Today’s agenda is below.

1. AI in FINANCE: Life insurers are funding the AI buildout with money set aside to pay pensions, against assets whose useful life nobody agrees on

2. ANALYSIS: The Age of the Disposable Company

3. PODCAST: How Perplexity’s Computer Is Replacing the Family Office, with Perplexity Finance’s Jeff Grimes

4. CURATED UPDATES: Machine Models, AI Applications in Finance & Investment Outlook

To support this writing and access our full archive of newsletters, analyses, and guides to building in Fintech & DeFi, subscribe below (if you haven’t yet).

Subscribe now

Also, check out the announcement from our friends at zerohash below.

zerohash and Marqeta to Enable Stablecoin Spending Across Global Card Networks

zerohash and Marqeta are bringing stablecoin balances into everyday card spending. The partnership connects zerohash’s regulated stablecoin infrastructure to Marqeta’s card issuing platform, so businesses can let customers hold and spend digital dollars through a standard payment card, anywhere cards are accepted. Stablecoin volume on zerohash jumped 690% year over year in 2025.

👉👉👉 Read the full story on how zerohash and Marqeta are making stablecoins spendable on existing card rails.

AI in Finance: Your pension is long Nvidia’s depreciation schedule

A life insurer that sells annuities takes money today and promises payments for thirty or forty years. To fund those promises, it holds a general account, and the job of that account is duration matching between assets with consistent cash flows and when the insurance payments fall due.

Long-duration liabilities need long-duration assets.

Government bonds traditionally have done that safely and cheaply. However, they also yield very little, so insurers have moved steadily into private credit, which pays more for the same maturity in exchange for illiquidity and risk.

We need to bring in depreciation at this point because of what they have been buying. Depreciation is how a company spreads the cost of an asset across the years it expects to use it. That period, the useful life, is an estimate made by management. Stretch it and reported profits rise. Shorten it, and they fall.

Today insurers are now funding data centres with pension money, and the value of what they are lending against depends on how long AI hardware stays useful. Nobody agrees on the answer, including the companies filing the accounts.

The size of the Problem

Hyperscaler capital expenditure has outrun hyperscaler cash flow, and the gap is being filled with debt. Morgan Stanley expects roughly $570B of AI-related issuance in 2026, with $236B priced by 31 May, four times the year-earlier pace. We have recently covered the pace of growth in this market when discussing the emerging GPU-focused capital markets.

Hyperscaler debt has pushed credit spreads in the US to close to 120 points. Given this market move, order books have fallen from nearly five times covered in February to under two times in July for hyperscaler debt financing, according to Apollo.

Barclays now describes Oracle’s credit default swap as a liquid hedge on AI capex generally rather than a view on Oracle as a whole company. The bank also counts data center asset-backed securities (ABS) and commercial mortgage-backed securities (CMBS) outstanding rising from $4B in 2020 to $61B by mid-2026.

Insurers are the ones purchasing large amounts of this debt, and this in turn goes into funding pensions.

The Chicago Fed found that life insurer private credit was at $849B, or 14% of balance sheets, in 2024. In 2025, 23.4% of US insurers’ total admitted bonds sat in private bonds.

In the UK, the liabilities flow through the bulk annuity market. A defined benefit scheme buys an insurance policy that transfers its obligation to pay members, either as a buy-in or a full buyout. Financial firm LCP sees a £350B to £550B pipeline for buy-in and buy-out over the next decade.

A buy-in completes, a member stops being a creditor of their former employer and becomes a creditor of an insurer. That insurer then has to find thirty years of matching yield.

Senior tranches of data centre financings have long time horizons, are investment grade, and illiquid. However, the investment is not without risk.

Meta’s data centre campus Hyperion

In October 2025, a bankruptcy-remote SPV issued $27.3B of notes to finance Hyperion, Meta’s data centre campus in Richland Parish, Louisiana. Blue Owl funds hold 80% of the joint venture and Meta 20%. Morgan Stanley was sole bookrunner, with PIMCO anchoring the debt alongside BlackRock. These were 6.581% senior secured bonds due May 2049, at 225 basis points over Treasuries. S&P assigned an A+ rating, one notch below Meta’s own AA−.

Meta’s own announcement sets out the terms that for the first 16 years of operations, Meta will make a capped cash payment to the joint venture based on the then-current value of the campus. A pretty safe deal on the face of it.

However, the lease itself only runs four years. A hyperscaler campus has an assumed economic life of 25 to 30 years. Meta therefore holds a 4-year commitment, an option to renew, and a sixteen-year backstop on the residual, while bondholders are on the hook until 2049.

Meta argues renewal is not reasonably certain beyond the initial term, which keeps the lease in the operating bucket and the debt off its balance sheet. The guarantee, worth up to $28B, sits in the footnotes to its annual report rather than as a recorded liability. Moody’s has warned that disclosures of this kind may not show the full picture of hyperscaler lease commitments.

The assumption underneath the rating

A residual value guarantee is a bet on what an asset is worth later. Pricing that bet means pricing how long purpose-built AI infrastructure stays useful. That is something the industry cannot agree on in its own accounts.

During 2025, Meta extended the useful life of most servers and network equipment to 5.5 years, cutting depreciation expense by around $2.3B over nine months.

In the same window, Amazon moved the other way, shortening servers from six years to five and taking a $920M accelerated depreciation charge, citing the pace of AI development. Alphabet and Microsoft use six years for comparable assets.

Michael Burry estimated the gap at roughly $176B of understated depreciation across 2026 to 2028.

Server useful life assumptions, 2025 disclosures

Nvidia unsurprisingly has rejected the short-life case, pointing to customer utilisation across four to six years, while moving to an annual product cadence and announcing the Vera Rubin platform in March 2026.

Whether a purpose-built AI campus in rural Louisiana has a buyer in the 2040s is the main speculation around this pricing.

Financial Losses in the Long Term

Executive Life Insurance Company sold annuities guaranteeing high rates and funded them with what was then a new, high-yielding, respectably rated asset. By the late 1980s, 55% to 60% of its portfolio sat in junk bonds.

The investment bank Drexel Burnham Lambert failed in February 1990. Policyholders withdrew around $4B that year. California seized the company in April 1991 with reported assets of $8.8B. The US Government Accountability Office (GAO) recorded Executive Life annuitants being paid around 70% of what they were owed, and California’s attorney general later put policyholder losses at above $4.5B.

It ended up taking 21 years for structured settlement annuitants to receive their payments, which they did until the estate nearly ran out of assets in 2012.

A July 2026 analysis puts private credit at about 6% of life insurer general account assets, up from 3% in 2020. Estimates of the same figure range from $289B to $1.8T depending on definition, so agreement on the level of risk is difficult.

There is a case for better protections outside of the US as well. Legal & General describes its exposure as predominantly UK private placement debt, almost exclusively investment grade, with no defaults since 2009, and matching adjustment rules which constrain UK annuity books in ways US general accounts are not.

Finally, Goldman Sachs expects almost all US regional grids to lack meaningful spare capacity by 2030. A permitted campus with a firm grid connection behind it could be worth more in 2040 than it cost, whatever is running inside.

Maybe.

It is difficult to assess how this will play into the lives of investors and pension funds in the future. Data centres have proven to be a particularly difficult asset to align on lifetime value, and it’s a narrative that has a lot of investors wary.

The most concerning reasoning for me is that there is not a standard value that is being approved by auditors in the market. While hyperscalers can continue to adjust years of life based on the whims of their investor needs, individuals who rely on the security of their pension funds are exposed.

👑Related Coverage👑

Analysis: The Age of the Disposable Company (link here)

I find myself these days starting a lot of new things.

It’s a way I cope with a lack of control over outcomes. It appears that the chaos and randomness in the world has increased, and perhaps this is due to the leverage that external forces have gathered through money, power, and intelligence.

Your inbox has never been this overrun with pitches, as AI has made mass-spam easy and common. Your social feed has never been this algorithmic, consumed by the spiking dopamine graph mapped back from advertising. And our politics have become designed for the panopticon propaganda age, where even the propagandists can’t figure out the nature of ground reality.

But it has also become easier than ever to begin.

Read Long Take

🎙️ Podcast: How Perplexity’s Computer Is Replacing the Family Office, with Perplexity Finance’s Jeff Grimes (link here)

Hi Fintech Architects,

In this episode, Lex chats with Jeff Grimes — who is Head of Live Events Products at Perplexity, the AI company that has evolved from an “answer engine” into an “agent platform” built around Perplexity Computer, its multi-agent digital worker. They discuss how Perplexity has shifted financial research from the how to the what, letting a user describe an outcome in a single sentence while Computer orchestrates 20+ frontier models, direct tool calls to licensed live data, and finance-specific skills to produce the artifact.

Jeff explains the enterprise strategy behind traceability - the north star that 100% of every quantitative figure traces back to its source filing - alongside bring-your-own-license connections via MCP and the consumer “personal CFO” vision powered by Plaid. They explore what 5x revenue growth on a 34% headcount increase signals for finance jobs, and why the future looks like a 24/7 family office that proactively surfaces and, with permission, executes financial actions for everyone.

Listen to Podcast

Curated Updates

Here are the rest of the updates hitting our radar.

Machine Models

Ethical and Bias Considerations in Artificial Intelligence/Machine Learning - Matthew G. Hanna & Liron Pantanowitz & Brian Jackson & Octavia Palmer & Shyam Visweswaran & Joshua Pantanowitz & Mustafa Deebajah & Hooman H. Rashidi

A Critical Field Guide for Working with Machine Learning Datasets - Sarah Ciston & Mike Ananny & Kate Crawford

AI Applications in Finance

⭐ AI-Driven Payment Systems: From Innovation To Market Success - Merve Ozkurt Bas

The Rise Of Generative Ai Agents In Finance: Operational Disruption And Strategic Evolution - Inesh Hettiarachchi

Financial Modeling in Corporate Strategy: A Review of AI Applications For Investment Optimization - Olufunmilayo Ogunwole & Ekene Cynthia Onukwulu & Micah Oghale Joel & Ejuma Martha Adaga & Augustine Ifeanyi Ibeh

Investment Outlook

⭐ Private Equity Outlook 2025: Is a Recovery Starting to Take Shape? - Bain & Company

⭐ Global Venture Capital Outlook: The Latest Trends - Bain & Company

⭐ Global Private Markets Report 2025: Braced for shifting weather - McKinsey & Company

🚀 Postscript

• Sponsor the Fintech Blueprint and reach over 200,000 professionals.
  👉 Reach out here.

• Check out our new AI products newsletter, Future Blueprint. (Don’t tell anyone)

• Read our Disclaimer here — this newsletter does not provide investment advice

• Contributors: Lex, Laurence, Matt, Farhad, Daniel, Michiel, Luke

• For access to all our premium content and archives, consider supporting us with a subscription. In addition to receiving our free newsletters, you will get access to all Long Takes with a deep, comprehensive analysis of Fintech, Web3, and AI topics, and our archive of in-depth write-ups covering the hottest fintech and DeFi companies.

Upgrade to Premium

Today we are diving into the following topics:

• Summary: Starting a company has gone from a $5 million capital commitment to a near-free act, and Americans now file over 5.6 million business applications a year while the share that ever make payroll falls toward 6%. We are creating disposable, on-demand companies the way YouTube creators upload video: supply floods to the tails of the distribution, and most of it is meant to be discarded. But the market still rewards endurance, not formation — the startup race resembles a 2,000 meter row, where the sprint off the start is chemistry and the middle thousand meters is pain tolerance, as Marqeta’s eleven years to IPO and LendingClub’s long grind attest.

• Topics: OpenAI, Marqeta, LendingClub, Meituan, Taobao, JD.com, YouTube, Amazon (Prime), Deliveroo, Stripe

Thanks as always for your time and attention,
Lex

To support this writing and access our full archive of newsletters, analyses, and guides to building in the Fintech & DeFi industries, see subscription options below.

Explore Full Membership

🤖🏦🧭 Our Ecosystem:

Generative Ventures | AI Research | Robot Money | Linkedin & Twitter | Sponsors

• 🚀 Lex is actively meeting teams building machine-native financial companies. If that’s you, reach out here.

• 🤖 AI & Robotics Industry Encyclopedia: In this 100+ page report, we collate all the meaningful companies across public quities, private equity, and digital assets related to the machine economy. 👉Learn more here

• 🦄 In partnership with InterPrivate, we just launched a $200MM SPAC looking for targets in fintech, digital assets, and AI infrastructure. If you are a late-stage founder or have an idea to discuss, reply directly to this email.

Long Take

Ready to Start

I find myself these days starting a lot of new things.

It’s a way I cope with a lack of control over outcomes. It appears that the chaos and randomness in the world has increased, and perhaps this is due to the leverage that external forces have gathered through money, power, and intelligence.

Your inbox has never been this overrun with pitches, as AI has made mass-spam easy and common. Your social feed has never been this algorithmic, consumed by the spiking dopamine graph mapped back from advertising. And our politics have become designed for the panopticon propaganda age, where even the propagandists can’t figure out the nature of ground reality.

But it has also become easier than ever to begin.

The cost of a new company used to include server hardware and office space. For this, you would have to make detailed plans and cost estimates, spending money just to get started. Then startups moved to the cloud, which led to an explosion in Silicon Valley investment and the Lean Startup methodology, which deemphasized ideas in favor of fast distribution.

Now, we are in the vibe-business prompting moment, where every rabbit-hole thought can be brute-forced by Fable into a plausible-sounding pitch and live codebase.

If it’s easy to do, that means there will be more supply.

I go back to my trusty analogy. Before YouTube, movies and shows would be filtered through curator markets in the shape of movie studios and producers, who would underwrite the project finance of a film through cultural taste and careful taste graph calculation. The middle of the bell curve prevailed.

Now, however, we live far in the tails of the distribution.

Infinite content is generated by the human spirit — and primarily its need for attention and social validation — and dumped into our consumption preferences. Our algorithmic doppelgängers, living as digital uncanny valley twins of our real selves, direct the media feed straight down our unwilling gullets. We savor the sweetness.

And so with business, supply will move to the tails. We don’t even know what we are yet missing.

Disposable, on-demand businesses can instantiate themselves in real time to be organized around our most superfluous need.

Take, for example, the drastic rise of on-demand delivery of food and groceries in Asia, which has spread to the Western world. Can you imagine living without Amazon Prime or Deliveroo? In China, Meituan crossed 150 million delivered orders in a single day in July 2025, at an average delivery time of 34 minutes.

All the stuff being said about programmable money, agentic finance and commerce, and the digitization of the dollar is mere preamble. That is the medium for the thing to come. In the way that software developers can instantiate software from thoughts and GPUs, or artists can now make media with a comment, so too will the creative act of microeconomics wrap itself around us and our wallets.

I’ve talked about this from the supply side, where AI transitions from doing tasks, to running workflows, to replacing employees, to then running entire companies and sectors. It may still seem science fiction to you, but what you are seeing is the future destination of the railroad track ahead.

The AI flies the ship in 2001: A Space Odyssey.

This is no coincidence. We are passengers of the ship. Our cultures, organizations, values, and constructions are the flora on its bones — as moss or mycelium cover the rocks of a sharp mountain ridge.

Humanity makes chaos, but humans love structure. We make technology to repeat our desires through math and programs. We build laws to stifle the dangers of mutation and volatility endemic to billions of humans sharing a planet. And so this technology will naturally — as in a natural monopoly, like electricity and oil — centralize and ossify. Like a carapace or a shell, it will move us along a latent space, whose multiple dimensions we can only feel and intuit, but never touch.

Read more

In this episode, Lex chats with Jeff Grimes — who is Head of Live Events Products at Perplexity, the AI company that has evolved from an "answer engine" into an "agent platform" built around Perplexity Computer, its multi-agent digital worker. They discuss how Perplexity has shifted financial research from the how to the what, letting a user describe an outcome in a single sentence while Computer orchestrates 20+ frontier models, direct tool calls to licensed live data, and finance-specific skills to produce the artifact.

Jeff explains the enterprise strategy behind traceability - the north star that 100% of every quantitative figure traces back to its source filing - alongside bring-your-own-license connections via MCP and the consumer "personal CFO" vision powered by Plaid. They explore what 5x revenue growth on a 34% headcount increase signals for finance jobs, and why the future looks like a 24/7 family office that proactively surfaces and, with permission, executes financial actions for everyone.

For those that want to subscribe to the podcast in your app of choice, you can now find us at Apple, Spotify, or on RSS. Or to support this writing and access our full archive of IPO primers, financial analyses, and guides to building in the Fintech & DeFi industries, see subscription options here. Our current price point is only $2/week.

Thanks for your time and attention,



Matt & Lex 🙌

Key notable takeaways:

1. The “how to what” collapse is the real product thesis, not just better models. The shift to zero-shot rests on three stacked unlocks: direct tool calls to licensed live data (Quartr for earnings transcripts, unusual whales for insider and political holdings, SEC filings for historicals) instead of relying on web freshness; a thinking-model router that orchestrates 20+ frontier models in parallel, matching the model to the job (a heavy thinking model for macro analysis, a lighter one for ticker-matching 550 names); and ~20 opinionated finance skills (DCF, three-statement, LBO, comps) tuned through expert-led evals. Together they turn one sentence into a polished equity-research artifact.

2. Traceability is the enterprise wedge, framed as “don’t trust and verify.” The stated north star is that 100% of every number in any output is hover-traceable back to the source filing - pre-scrolled to the page, highlighted, with the full chain of calculations exposed. The framing inverts the usual “trust but verify”: assume the user won’t trust the model, so trust must be earned per number. Paired with bring-your-own-license via MCP (FactSet, LSEG, Morningstar, CarbonArc, PitchBook), this is the concrete answer to why regulated institutions get comfortable adopting.

3. The productivity and jobs signal is quantified and lived internally. Perplexity grew annual run rate 5x while increasing headcount only ~34%. Computer began as a company-wide Slack bot where every request was visible to all employees; Jeff now runs 9–10 scheduled cron jobs each morning and says essentially all code is written first by his agents. On the consumer side, the emergent pattern is build-your-own long-tail apps that no roadmap-bound product could serve - a DraftKings-addiction accountability system that emails a user’s spouse on any bet, or a GitHub-style heatmap of daily spending - which is the real substance of the personal-CFO bet.

Background

Before Perplexity, Jeff Grimes spent roughly eight years at Google (2014 to 2022), joining as an associate product manager and rising to Group Product Manager, where he spent five of those years monetizing Discover, Google's personalized news and interest feed, and learned how publishers, monetization, and news consumption fit together. Alongside that work he served as a Sequoia Capital Scout from 2016, making pre-seed investments totalling around $570K.

In November 2022 he left Google to co-found and lead Shades, a consumer news startup building Shadebot, an AI news-authoring tool; after early growth that later plateaued, the venture wound down in early 2025, at which point he joined Perplexity. He holds dual degrees from the University of Pennsylvania's Jerome Fisher Program in Management & Technology - a B.S. in Computer Science and a B.S. in Economics from Wharton.

👑Related coverage👑

Topics:

Perplexity, Perplexity Computer, Perplexity AI, Google, Shadebot, Plaid, Yodlee, Claude, ChatGPT, AI, Artificial Intelligence, LLM, CFO, financial services, AI commerce

Timestamps

• 1’05: A company built on failed founders: Why the startup that didn't work out led here

• 5’10: A playing card company that became Nintendo: How curiosity carried Perplexity from answers to actions

• 9’52: We're basically at zero shot now: Why prompt engineering is disappearing from finance work

• 15’21: A thinking model that routes the job: How Perplexity picks Claude Opus for macro and Grok for tickers

• 20’12: Two tracks, one engine: Building for enterprise workflows and a personal CFO at once

• 25’25: Bring your own license, or use ours: How Perplexity gets institutions comfortable enough to adopt

• 32’54: 5x revenue on 34% more headcount: The productivity gain Perplexity lived firsthand

• 37’28: Connect everything from your mortgage to the painting on your wall: Building the true personal CFO

• 45’15: A family office that works 24/7 for everyone: The proactive, automated future of the personal CFO

• 48’01: The channels used to connect with Jeff & learn more about Perplexity Computer

Illustrated Transcript

Lex Sokolin:
Hi everybody, and welcome to today's conversation. We are very lucky to have with us today, Jeff Grimes, who oversees the finance product at Perplexity. Perplexity is one of the early AI companies and has been a leader in figuring out how to bring financial services into the LLM suite. And I'm really excited to learn from Jeff what's happening. So, with that, welcome to the conversation. Hi, Jeff.

Jeff Grimes:
Hey, Lex. Thanks for having me.

Lex Sokolin:
My pleasure. First off, what is your way into Perplexity? How did you journey yourself through the tech industry and into a leading AI company?

Jeff Grimes:
I started my career at Google. I was there for seven years as a product manager. I worked for five of my seven years there on monetizing the discover feed. So that's one of Google's news products and interest in News Feed. And I was exposed there to the world of publishers and monetization and how users interact with news content. And that led me to what I did next. So, I quit in 2022 to join a co-founder. And we started our own venture, and we ran that for two years. And that was building on a lot of the things I learned at Google. It was a news product. We had a consumer app, and without going into all the details, it did not work out. We grew for a little bit and then plateaued and decided to shut it down at the beginning of 2025, which is then when I got connected to Perplexity and I joined. And there were many reasons I was excited to join, but one of them was the number of other failed founders there. you know, there's quite a few people here who have started their own venture and for whatever reason, it didn't work out.

And they came on a Perplexity. So, I think it was very clear from the beginning that that spirit was there. And it very much is still here. The company is, several hundred people now, but still feels much, much smaller. Like, I experience both ends of the spectrum, a 200,000-person company into a five-person company in my first two roles. And now this perplexity feels much closer, much, much closer to the five-person company.

Lex Sokolin:
Let me ask you a self-indulgent question. What does it mean to be a founder, and how would you define a failed founder versus a successful founder? Like, what are those words mean to you?

Jeff Grimes:
You learn so much. The experience of running a company for two years, it was every kind of adjective I could throw out. It was exhilarating. It was overwhelming. It was stressful. You just can't shut off your brain ever from thinking about it. I think as opposed to my Google experience, which, you know, there are much clearer boundaries of, okay, we're finishing up.

And, you know, now we're not going to think about this for a while until the morning or something. I think there's a certain type of person, whether you're a failed or successful founder. I can't opine as much on being a successful founder from my experience, but I think there's a commonality that you just love to think about something all the time and throw yourself completely into it. And certainly, as I mentioned, I experienced both ends of the spectrum in my first two roles, but the second to me was much richer and more rewarding and meaningful. And so, I think that's a common thread in people who work at Perplexity. But regardless of their background, you know, you don't have to have been a startup founder, but everyone really cares. And that's something that you actually like. It sounds like a cliche, but you actually don't see it very often. I haven't seen it very often in my career before this. Every person here and every role deeply cares about the details and getting things correct and, you know, can operate at the high level.

View the strategic view and think about the future and where is AI going and all of that. But then can zoom in and like obsess over, oh, this, these, these elements are misaligned by two pixels. We need to fix this like nothing is not worth our time or attention. And everyone takes pride in the craft. So, I think that's a commonality. Like, honestly, it's a credit to our recruiting team and the people they brought on. But that's really energizing to work with people who care that much.

Lex Sokolin:
So then let's zoom out and ask the starting question, which is what is Perplexity? Where did it start? And you know, what is the company?

Jeff Grimes:
Now Perplexity has evolved. Like I'll talk about the evolution from an answer engine to now we call ourselves and we do call ourselves an action engine. And the through line there has been curiosity. This is something that Aravind, our CEO and co-founder, has talked about from the very beginning, when he and the other co-founders started the company.

Perplexity is built for curious people and it started just with answers. You know, you probably used one of the early versions of the product instead of having to sift through the ten blue links like on Google or other search engines. The answer? The answer is just synthesized directly for you, but it was always grounded in this mission of appealing to curious people and stoking that curiosity and letting them propagate that curiosity. And now, recently, the product has evolved with Perplexity Computer. You know, that's the thing we're all really focused on now, and that's what we're building for and finance, which is our digital worker product. I'm happy to talk more about it, but at its core it gets things done. It does work. It's not just researching things. It can research things if you want it to, but it can create artifacts. It can create websites and live dashboards, and it's collaborative and it gets things done. I think the you know, I mentioned that curiosity is the through line. It's helped us evolve from one to the other quite naturally.

And I think there's things we did in between that helped us bridge the gap, like Comet, our web browser. We've been working on that since pretty early last year, and we're proud to launch that. I think it was in July last year with agentic Browsing. So that expands from okay, we weren't answering and now we have this agentic browser which can do things for you. You can use your mouse, it can use your keyboard, it can make bookings, it can send emails, it can find people on LinkedIn, whatever you want it to do, whatever you could do on a computer, it can do. And that helps us bridge the gap into something now like computer. So, you know, computer was launched in February of this year, but I think the seeds had been planted long before with, with comma and a genetic browsing. And now computer can orchestrate many models and agents in parallel to get things done. And that's the key is, you know, multi multi-agent multimodal orchestration and parallelization of all these things.

You know, to come back to your question, Perplexity has evolved from an answer engine to an agent platform. But it's always been based on curiosity right. Naturally you start with answers. And if you're curious you want to do more. You want to say, well, how can this work for me? How can I get things done based on this? And in that sense, I've always thought of it similarly to Nintendo. I really like their evolution where they started in the late 1800s as a playing card company, you know, similar to something like bicycle. Today they made 52 card decks, but their mission was to entertain people. It was to bring a smile to people's faces, was what they called it. And so, they actually evolved over 100 plus years to making, you know, the Game Boy and the N64, SNES, obviously now all the way up to the switch. And they have these iconic franchise characters, but it's actually amazing, if you think about it, that a playing card company evolved that way, but it was because they were always grounded in their mission statement of entertainment.

So, we think of it similarly. We've evolved the product. We're not overly focused on one form factor, but certainly right now computer is what's emergent and dominant as a use case. And, you know, it's one of those things where once you use it, it's hard to imagine not using it. So, we're very much focused now on building finance tools for this action engine of computer that that gets real work done for people.

Lex Sokolin:
Can we pause just in terms of like where this sits in the broader AI industry? Because I like your mission driven framing, for sure. And I think that comparison to Nintendo is interesting, because organizing around something that a person wants is probably much more meaningful than organizing around features or bullet points of capabilities and things like that. But at the same time, like even what it means to be an AI company in the last 2 or 3 years has really shifted from people chatting inside of a like LLM experience to expecting agents and full automation and like harnesses and the enterprise and all of this other stuff.

So, it feels like the entire space is kind of really racing towards something. Help us understand kind of like what you've described within the context of the broader industry, and how are you cutting your own path there?

Jeff Grimes:
We think a lot and talk a lot about the what versus the how. That's the key evolution to look at. And so I can give an example of what let's say, you know, in finance because that's what we're building. If you're a junior sell side analyst doing equity research, and you've been tasked to make some artifact that shows the factors that drive the price of a security you're researching, right? Let's say you're researching Uber as a stock. If you go back to 5 or 6 years ago before the transformer revolution had occurred. Obviously, you're doing this fully manually, right? You're looking things up separately. You're going into different portals to get the historical open high lows, cloud volume information. You're downloading the data; you're looking it over. You're maybe going into your newswire subscription website, putting in a couple of search filters, downloading all of the news articles you're going to go grab cross correlated macro signals.

Let's look at the price of gold, the price of bitcoin. How are those changing. You know you might spend days doing this piecing all of it together and then coming up with a final report. Here are the top five factors that have leverage over Uber's stock price going up or down right. So, it's fully manual. It takes a couple of days pretty big time sink. And you're going to make inevitably even a very strong analyst will have some errors in the handoffs of the data and analyzing the data. Right. So then right after the, you know, what was it, late, late November 2022, ChatGPT comes along and shows us what things can look like. Post transformer you still have to download all the data, like you're still going into those portals, you're downloading the files and you know the context. Windows are smaller. Hopefully you can still feed most of it, if not all of it into a thinking model, and that will remove a lot of the errors you get in analyzing the data.

Right. So, we've evolved a little to save some time. And, in doing the analysis, you still have to fetch the data yourself and feed it in. Then models continue to get better. You get more direct integrations, you get MCP that comes along. So now you're getting into the models can grab the data for you, especially if you have your own license. So, everything until now has been focused on, you know, the how you need to describe to the model. You need to prompt it. You know, prompt engineering was such a thing that you heard so much about in the last few years. Not as much right now. You had to prompt engineer to say, okay, here's how you should do this. It was maybe many shots prompting or one shot prompting to use that terminology where you'd have to give it an example, right? Here's this artifact I produced last year that I think is really good. That shows, you know, maybe I did an analysis of lift. Here's how those price factors looked.

You should make it look like this. Here's the data, make it look like this, or even go get the data but make it look like this. So that's like one shot prompting. And now like to answer your question, bring it into Where are we today? We're basically at zero shot. Prompting now is something like computer. That's where it shines. You just tell it what you want to accomplish. So now we're not focused on the how. We're focused on the what. Just tell it what you want to do. Build me an artifact that helps me understand why Uber's price has moved over the last 90 days. In some cases, you can actually just stop there. You can literally just give one sentence. We've done enough of the skill building and data integration for the right tool calls. There's parallelization. So, it all happens quickly. And it's going to come out looking very good. So that that eliminates basically all the problems from before. It's fast. You don't have to fetch the data yourself.

You don't have to worry about analyzing the data. You don't have to worry about prompt engineering. So, I think that's basically the evolution we've been moving from the how to the what. And there's ways that this continues, right? I think with something like memory, knowing the user’s preferences and what they ask for at what times you can start to be proactive. And, you know, maybe if the user has just completed some similar tasks, that has tended to be a precursor to then wanting a price factor analysis, you could ask them proactively, oh, would you like to do this? Or even if the user has instructed you to be able to do things autonomously or on some schedule, you could just run it again without the user needing to prompt explicitly. So, there's still ways that this can extrapolate into the future. And, you know, we're already building for some of those in computer. But I think big picture, that's what we've evolved from in the last few years is, you can now, in plain English, just tell something like computer, a digital worker, a multi-agent orchestrator, what you want to do instead of worrying about how.

Lex Sokolin:
I want to go next into, you know, why the financial services vertical and talk more about that. But before going there, I do want to ask about the change in capability that has allowed you to deliver this experience, right? The zero-shot prompt, because it's not just about the underlying LLMs. Like, I don't think you guys make your own LLMs, and it's much more about sort of like the orchestration of everything around it. Why does the experience no longer need me to drop in? Sort of like the PDF that I built. Why does it now know much better than me how to do an equity research report? Has it seen all the equity research reports there are or like have you like pre fed it the right structures. Like talk about some of the technology progress underneath to get there.

Jeff Grimes:
I think it's a combination of a couple things that have helped us get there. One of them is moving from just exclusively. And this goes back to Perplexity’s evolution from an answer engine to an agent platform.

One of the things here is instead of having just web search, right, initially that was the only function of something like Perplexity and ChatGPT is, you know, let's just it's just pure inference. And if we're going to make a tool call, the initial perplexity, it was, you know, it was just web search. And now we've evolved to making tool calls to live data as much as we can. So, I think that's one of the unlocks rather than, you know, if you say, hey, fetch the stock price change for Apple in the last 258 days without a tool call to live data, you're relying on web documents to get that answer. And so, depending on freshness of web pages, you know, it might be correct and it might not be correct. Right? You're just hoping that someone has live data on a web page. But if you've done the integration, either because the user has connected their license data via MCP or you have a direct API integration, you know, we've gone out and signed a bunch of deals so that we can provide what we call off the shelf tools.

In computer, you don't need to have your own license we can pull from quarter for live earnings transcripts. We can pull from unusual whales for insider transactions and political holdings. We can pull from fiscal and SEC filings for historical financials. You're making direct tool calls as often as possible. So that eliminates some of the hallucinations in like handoffs between the data as you're just grabbing the data directly. So, I think that's one thing we're relying on direct tool calls to live data as much as possible. And then the second is multi you know you hinted at different models that can be used multi-agent orchestration multimodal orchestration. You know there's 20 plus frontier models that are available on Perplexity. And you have a powerful thinking model as the router as we call it. Right. The orchestrator that is looking at the user’s query and deciding, how do I want to tackle this, and breaking it down into tasks that can be done in parallel or in sequence, and it will then call the right model for the right job. So that's another unlock.

So, this can optimize for quality. It can optimize for cost. It kind of optimizes for everything depending on your query. But if you're if you're asking for something like a really in-depth analysis of some macro situation, you know, it might call Opus 4.7 to do that. A powerful thinking model. If there's some part of the task that's like, hey, the user has these 550 company names and we need to match them to tickers. Something like grok can be really efficient at that. It's going to be faster, it's lighter, it'll be cheaper. So, the right model will be called at the right time. I think that's the second thing is, is orchestration and parallelization. There are a couple other things that go into it. You mentioned how does it know? How does it know how to build a DCF. How does it know to do these things. Part of that is skills. And so, you know, there's times to be generalist and there's times to be opinionated within finance and a lot of these specific workflows there.

There are cases where we are opinionated and it will load a skill. Right. And so, if you ask, hey, build me an elbow model, build me a three-statement model, put together some public private comps, put together a five-year DCF of Nvidia, whatever you want to do. We have like 20 different skills in finance that the router can attempt to run if it's applicable. And then it does have some guidelines of what makes a good DCF. And that's where a lot of the work you can do behind the scene on evals. Right. This is really, really in the weeds grind it out type of work because you, you know, looking manually at lots of different outputs and having experts like a lot of our team is former finance professionals. So, to understand what makes a good DCF versus a bad one and up levelling some of those takeaways and putting them in a skill that can get loaded by the router and run to make the right tool calls to structure the output a certain way. Of course, it's all flexible based on what the user asked for, but that's a third thing that can help. So those are three of the factors I'd highlight.

Lex Sokolin:
Actually, super interested to hear that a lot of the team is former finance. Can you open that up a little bit and then tell us about the decision of taking Perplexity more deeply into the finance vertical. And I guess there's both a consumer-focused set of skills as well as more professional stuff like you've mentioned equity research and building an elbow model. So maybe just start with kind of the reason for moving in this direction.

Jeff Grimes:
From the beginning. And I assume this is true of other AI products. Obviously, I can only speak to Perplexity, but from the beginning, for us, finance was one of the most popular verticals in terms of even the original perplexity search only product. People love to ask for simple things such as what's the stock price of this company? To very in-depth things like, you know, here's a bunch of documents, help me make sure I'm, you know, filing my taxes correctly. You know, what should I be putting on this line item or something? So, it's ranged, as you said, from personal to professional, from simple to complex.

But it's always been in our top few domains. You know, we have we have a domain classifier that looks at what general area entertainment versus sports versus finance versus travel versus tech is this query. We classify them that way. And finance has always been among the top. So, we've always wanted to have some kind of finance product. We started building and finance a bit over a year ago. The product initially was retail focused. So, we created this. This website is dashboard called Perplexity Finance, and the idea was to be an AI native finance research platform. You know, I think lots of people have used Google Finance or Yahoo Finance. The theory was, could we could we build something that is AI native and is able to accomplish a lot more? And we saw really good traction with that.

So, we you know, to give an example, what is an AI native feature on this type of research platform? Instead of just listing news stories on a company's page, if I go to the Nvidia page instead of another platform, which, you know, obviously it's going to show the price graph. What's the price? How is it moving? And then there's some news stories listed. We can synthesize this for you. So probably the most popular feature on Perplexity Finance, when we were really focused on that surface last year was the price timeline, which shows why is the price moving. So, it just synthesizes a short paragraph for every day where the price moved in, you know, a large magnitude up or down. And it would say, here are all the factors. And that's being fed in a bunch of interesting context. It's basically doing that work that I mentioned in that example about the equity research analyst. You know, if you're looking at why's Uber moving up and down? We're feeding in the price of Uber.

We're feeding in the piers of Uber, and those are determined by the model. You know what are what are companies that have similar profiles? What are companies that people talk about as peers to Uber? It's kind of top down and bottom up. We're feeding in the price movement of those. We're feeding in the price movement of the broader sector. We're feeding in cross correlated signals like, you know, Bitcoin and gold and certain ETFs are the indices up and down. Is this, you know against the market. Is it with the market. And then we're feeding in a bunch of news stories. Part of this is from just having an index of the internet. Part of it is we've signed partnerships with MIT Newswires and Benzinga. So we're actually getting live Newswire data from them to all of this context. It's like pretty massive bit of context is fed in and we're just synthesizing it for you. Here's our best explanation as to why Uber was up, you know, 3.5% yesterday. And so that's actually pretty interesting and powerful for people who can just get a sense quickly of how this asset works, what causes it to move.

So that's one example. You know, we have like a dozen features that we I would consider truly AI native. So anyway, we saw over 50 times growth in the final nine months of last year with how many visitors we got daily to Perplexity, finance. And toward the end of last year, especially as we were gearing up for, you know, what it would eventually become computer. We started thinking more and more about enterprise, where it became clear that we were going to be able to really get work done and not just answer questions, not just provide this dashboard for retail users. So, we're really focused on enterprise now, letting enterprises bring their own data, building workflows, helping people get quickly into the type of work they're doing pretty often, and making it easy. We're really focused on traceability, which I'm happy to go into more detail on. How do we help users verify that the answers are correct so they can trust them and trace them back to filings or earnings call transcripts wherever it came from.

And then we want to make sure the data is portable. You know, we have an Excel add in now we launched financial data via the API. So, if you're a developer, you can get live data that way to live price data. So, there's a bunch that goes into building a finance product for enterprise. But we still have both tracks. We still have the Perplexity finance you know, retail-oriented product. I think you hinted at this before. We also have a personal finance product now, which we've just launched about a month ago. And our vision is that Perplexity computer can be your personal CFO. That's what we're calling it. So, you can connect all your personal accounts, your mortgage, your bank account, your credit card, your student loans, your crypto wallets, your insurance plans. We want everything to be connected to perplexity computers so you can ask questions about it. Help me create a budget. Help me plan for buying a house. Here's my financial goals. Help me file my taxes. You know, Perplexity computer can actually now fully prepare your taxes if you want it to. So, we still have both tracks.

Lex Sokolin:
Can you do it across multiple jurisdictions?

Jeff Grimes:
All right. It's going to get better. You know the progress from tax year 2025 to tax year 2026, I think is going to be striking, as is the common trend of how quickly this stuff unfolds. Yeah, we still have both tracks retail and individual focused, but most of our effort is now building for enterprises, both large corporations and, you know, smaller teams, startups, solopreneurs who want to get stuff done. And that's the common trend or the common theme is going back to the action engine, you know, answer to action, evolution, perplexity. Computer gets things done for you.

Lex Sokolin:
You mentioned the enterprise sector and a bunch of use cases within financial services. I'd love to learn more about that. One of the questions is what is the shape of the right financial services company? You know, is it mid-market is the top end of the market, just like parked on by anthropic and the private equity firms now? Or is it still competitive and anywhere like what are the competitive dynamics of getting into financial natural enterprise.

And then when you do get in there, what's the right deployment model like? How do you do digital transformation with these companies who are probably scared to death of letting go? You know, any information, either about their customers or about their proprietary sort of secret sauce.

Jeff Grimes:
A couple of things are, I think, in terms of the competitive dynamics and, you know, getting in. Everyone's thinking about this to some extent. I think we're still seeing a kind of a bifurcation in there's some companies that are extremely eager to adopt this, whether large or small. There are some companies that are still saying, hey, we're not ready to do this. We, you know, our system is not set up for this. We're, you know, we're still focused on manual work, which sounds surprising, but it's actually still relatively common, I think among the firms and companies who are eager to adopt, there's kind of this convergence toward the middle, in a sense, where if you're a large company, you're looking to get an enterprise subscription with a bunch of seats that can help do a lot of the work.

That's manual. Right. And so, there's a lot of these manual, repeated actions that, you know, the junior analysts are doing. And so, you can make that all more efficient and, you know, maybe operate a bit more like a small company. And then on the other side you have the, the really small shops, you know, like a three-person boutique firm or something where you want to feel like a 50-person firm. You want to supercharge that really small crack group of employees and give them these really powerful tools. And those are the cases where we see astronomical credit consumption of computer. They're common, like the common case there is doing multiple things in parallel. They really understand the power of you can just kick these things off. You can kick off a task from a browser. You can have multiple running at once. You can create recurring tasks like we've seen a case. There's a client who's really focused on heavy metals and minerals. And so, you can create recurring alerts like really powerful alerts and automation.

So, every time Javier Millei is set to visit another country and lithium is on the agenda. Do XYZ like we've seen people set up alerts like this. Or if there's going to be the president of another country on this visit as well. And it's happening in China or Saudi Arabia or Chile, you know, write a report and tell me the top three assets they're going to, you know, win and lose from this meeting, depending on what happens and what are the different contingencies, you can set all this up. So, I think those smaller firms is where we tend to see really heavy power usage of the product. And to the other part of your question at this point, security and privacy, you know, like not using the information for training. All this is table stakes. We take this extremely seriously. Our security team is best in class. They obsess over this stuff, and so there's a lot you have to do as part of the sales cycles to make sure, we're compliant with the client systems, and that's a big reason why we recently expanded actually just last week into for financial data, letting companies bring their own license.

So, we call this bill bring your own license. You can connect your license via MCP to Perplexity Computer. So, if you have a FactSet subscription to Lupa, Morningstar, Carbon Arc, or PitchBook, those are the first five we launched. You can put in your credentials to computer. Those will be securely connected to via MCP, so that data is pulled directly from your license. If you don't want to use computers off the shelf tools. And so that's just one more assurance that everything's happening in a way that is easier for you to understand and be in control of. You already have your licenses. We do think our off the shelf tools are great, and typically we see the smaller firms taking advantage of that because they can't afford, you know, a $30,000 Per user per year, a subscription to some of these products. And so, they're happy to use ours off the shelf. But that's another thing that that helps enterprises get comfortable. And then no matter what I mentioned traceability before, that's one of the core principles we have for this finance work is you have to show the user where the number was pulled from and what transformations, what calculations were done on it.

So, our vision and our north star is that literally 100% of quantitative data, every number within an answer artifact that is produced by computer within the finance realm is traceable. You can move your mouse over it; you can hover the mouse, or you can click on it and it will show you. Here's the exact filing that was pulled from the filing is pre scrolled to the correct page. The number is highlighted. And here's the full trace of every calculation that was done on top of this number to help us get from the original filing number all the way to the final, you know, number that was shown in the answer or in the report and the PDF and the slides, whatever was generated. And that's really important. Even if you already trust the data, you still need to be able to see the trace. It's like, Ronald Reagan always said trust but verify. I had a college professor who said, don't trust and verify a finance professor of mine in college. And so, I think we actually skew more toward the latter, which is like, let's assume the user is not going to trust us.

We need to earn the trust. We need to show them where this number was pulled from so that they know. Okay. Yeah. It's nice that you said you have this multi-agent orchestrator, but I want to see what are the tool calls that were made, what were all the thinking steps. Where was the number pulled from. Is it correct. Can I go audit it. That's common across any use case regardless of the size. Even if it's a retail versus enterprise user, I think there's a lot of retail users who probably see the traceability and it seems like overkill. And maybe it is. We'd much rather on that side than the other way around.

Lex Sokolin:
Thinking through deployment inside of financial institutions. You know, I think lots of people just say the ask the simple question of what's going to happen to jobs. And that is the question. End of the day, I want to ask. You know, when I look at, for example, recent news that Coinbase ended up laying off about 14% of its workforce and pointing to AI or Klarna as restructuring or Ramp’s restructuring. What are you seeing in terms of your clients? Are they just becoming more productive, or is there like impact on how they do business?

Jeff Grimes:
We've lived it firsthand at Perplexity with a massive increase in brand productivity. I've lived it myself. You know, obviously I think all these different cases, Coinbase, etc., that you mentioned Klarna. It's going to differ case by case. I think there's now an interesting thing I've been following in the last couple of days, this narrative of is it real? Is the AI, you know, consolidation of jobs real or is it AI washing? I've seen that term come up like is it just a CEOs saying, oh, it seems like the markets will react favourably if I say that this restructuring was due to AI, regardless of whether that's true. So, I think that's interesting. I'm not going to opine on that. I think it's, you know, it's going to unfold. But certainly, what is undeniable is the productivity gain at Perplexity. One of the principles that Aravind has instilled, which I love, is that we build for ourselves.

And that's where we are going back to, I mentioned we have many former finance professionals on the team. Like we build this so that it will be useful to us, you know, either versions of ourselves from a previous job or our current job right now at perplexity. And it first has to be undeniably useful to us. So, computer actually started in a slack channel. It was initially a slack bot, and the slack channel was open to the entire company. So, the only way to run a computer request in the initial beta version of the product was you had to put a message that was visible to everyone at perplexity. That was really cool because you could see all these creative things that people are asking it to do. Things that people are scheduling as recurring jobs. And it was incredible how much productivity was gained there and how much could be done in an automated way. You know, my day to day now is night and day different from 5 or 6 months ago? I have multiple parallel computer tasks running essentially at all times.

There's, you know, bigger things. I schedule before I go to bed so it can run overnight. There are tons of stuff that's scheduled on a cron. I think at this point I have like 9 or 10 things that happen every morning. So much of it, you know, essentially all of our code that we write is written at least initially by my agents. So, computer does get stuff done, and that's undeniable. And so, I think that's why we've seen, like Aravind talked about on X the other day, that we grew our annual run rate by five times in the last year, while only increasing our headcount by, I think it was 34%. So that's the kind of productivity efficiency gain that you're seeing. So, I think it's one of those things where if you understand how powerful these tools are and understand how to use them, you're going to be able to benefit massively. But then, on the other hand, we don't want these tools to feel burdensome or oppressive to use or unapproachable, like one of the beauties of computer is that you don't need special software, you don't need special hardware, you don't need special training, you don't need special technical skills.

You literally just ask for what you want in plain English. Of course, you still need to have the knowledge of, you know, what can I ask it to do? And that's where we have things like workflows that can nudge you in certain directions. Oh, here's some examples of things you can do to ease people into it. I think both are true. There's this massive productivity gain from people who understand the capabilities, but we also want to build so that it's really approachable and easy.

Lex Sokolin:
Let's spend a moment on the personal CFO and the Plaid integration. You know, I think it's really interesting as a direction. Obviously, for me, it brings back the beauty of some early fintech stuff. You know, like mint.com in 2006 and so on. There's been this whole path of data companies in fintech that everybody thought were very valuable Mint and then Yodlee and Plaid and lots of others. There was a cooling period where actually not so much value flow to these data companies, and the software platforms on top became more valuable.

And now we're back in this world where like access to regulated data is amazing, and people are now building all this disposable software using things like Perplexity on, on top. So, it seems almost like a step back into some of the dreams that people had about early fintech. What have you noticed from like user data in terms of their behavior and their interests? Like is there anything you can share about how they're using these personal financial management tools or like, for whom are these personal financial management tools and any insight into the customer segment and how this is starting to show up?

Jeff Grimes:
I mentioned we started building in finance a little over a year ago. We always knew that personal finance was the other side of that coin. You know, the first side we built for being general market research. What's the price of this asset? When's this company going to announce earnings, etc.? And we've never or we had never been sure when would be the right time to build in personal finance. We always had Plaid on our radar.

We'd had some initial discussions with them in 2025, when we were sure we were going in the direction of computer. That's when it became clear that it was the right time to build in personal finance. And the reason to get to part of your question, if you look at mint R.I.P. or a monarch or copilot money or all these apps for personal financial management. The you know, the value is clear at a high level. Like pretty much everyone has some something that they care about in their personal financial life or situation, right? Almost everyone will have a credit card or a bank account. It's just enormously universal. The problem has always been that you're beholden to the product roadmap of these companies. Like if you were if you were a mint user back in the day, you know, they maybe have five or had 5 or 6 core use cases that they supported. Here's the ledger of transactions. Here's that data aggregated by category. Here's a budget forecaster. You know some of these had things specific to certain domains.

Like you could download an app that would, you know, should I pay off my mortgage, or should I? Should I just keep paying the interest and invest my money in the market? Right. Or here's my portfolio. Are there rebalancing opportunities? So, there are these very specific ones, but no one could cover all of them, and no one could even start to think about this massively long tail like the end of one-use cases, because they're not going to put that on their product roadmap, right? They're not going to build a product for one person or, you know, a very small group of people. But to come back to your question, that's what we've seen with the plaid integration so far. And it's our first step toward that personal CFO vision that I mentioned. The value is in this massively long tail, the hyper personalized. So, we were talking to a user the other day who told us that he used personal CFO. He used our cloud integration to create an accountability system for himself to end his sports gambling addiction.

He said. He's like, you know, to be honest with you, I'm addicted to betting on sports. I'm really motivated to improve. And I connected my credit card through Plaid to Perplexity to computer, and I had a computer task that says every day. Look through my last 24 hours of credit card transactions and if there's a DraftKings transaction, send an email to my wife that that tells her that I bet on DraftKings because, you know, he and his wife were, you know, working together on, you know, let's end this addiction. And he wanted an accountability buddy essentially like that. That's something that you never could have done on one of these other products. Because why would they support that. Right. Like that's a that's a very specific use case. So that was why we got into this is we realized computer can let you build your own apps. So that's been one of the emergent use cases. We had one the other day I posted about this on X because we saw it come up actually a couple of times.

People really like that GitHub heat map view that shows how many contributions you have. It's that grid and it shows how many contributions you have each day. People wanted to apply that to their spending. Create this little grid heat map. So, it's like a darker color, or I guess it would be a brighter color on the days where I've spent a lot, and it's like a muted color on the days where I spent a little so I can actually visualize, okay, how much does this spike during holiday shopping? How much is this spike during summer travel? Whatever. These are the types of things that just weren't possible before. So now people are building their own apps, their own personal financial management apps with computer. So that's a that's the value. And I think a lot of the work we have to do is basically twofold. Number one, we have to let you connect more and more parts of your financial life to computer. So we started with cloud. And to get at you know, you're talking about Plaid and Yodlee.

And are we taking a step back I think back in that direction. I think it's really important that we use something like Plaid, because all the work, they've done in security and privacy. Right. And I think there's a lot of user trust. They have some stats like okay, 95% of users have, you know, have connected at least one plot account somewhere else. And so it's just you just put in your phone number, you've already done the authentication and it's really easy to connect. You already trust it. You know that plaid is secure. So Plaid was the first step. But we're working on crypto wallet integration. We're working on connecting insurance plans. We're working on, you know, connecting. If you have private holdings in or, you know, holdings in private companies or equity or options, you should be able to connect that even eventually, like an illiquid asset. If you have some painting on your wall, you know, there should be a way that you can capture that in computer homeowners, you should be able to input your address and we can give you intelligence on your neighborhood.

Are there certain listings or construction projects that might impact your home value? So, one part of this is let the user connect as much as they can. Let's make sure it's done in a secure and trusted way so that you have all this context, your investment portfolio, everything's there. So, you can ask any question you want, and we can give you a really accurate, grounded answer with the traceability layer that I mentioned. So, it's all tying back to specific things that you've uploaded or that you've connected. And then number two is helping the user understand all the things that they can build, and giving them really easy one click options to do that. So, for example, if we go back to that GitHub heatmap style spending tracker, if we see that a lot of users like that, we can just say, hey, here's a one click option to do that yourself. This app is really popular. A bunch of people have made something similar. You can click here and it will be built for you instantly without you having to think about how to prompt it.

So, there's a lot to be done like that to help people actually take advantage of all the capabilities of computer. But there's a big opportunity here. And to come back to that long tail, that's where the value is. It's that you're breaking free of the confines of some other products roadmap. And literally, you know, anything you can think of within reason can be built by computer for you and for just you if you want.

Lex Sokolin:
If we look forward, you know, it. Just in the last 2 or 3 years, we went from, as you said, summarizing something inside of a ChatGPT window to the sort of science fiction future you've described where you've got, you know, ten robots every morning doing stuff for you, and you're building disposable software. That's your personal CFO and, you know, eventually delegating financial decisions likely to these actors. If we look forward two years from now, what are the outlines of how financial AI is going to look? It feels almost like we're at an end of the human ability to imagine what like programmable money and AI look like.

We have it now. We have these harnesses. We've got the agents. People can just say things and they will happen. So, are we going to have the same exponential liftoff as we did in the last 2 or 3 years, and if so, in what direction? Or are we kind of coming up to the ceiling. That's similar to image recognition, right? Like at some point it doesn't work. And then you get to the ceiling of it works, and then that's it. That's fine. How do you think about this?

Jeff Grimes:
I think it's going to continue to accelerate. And a law will depend on each user's personal tolerance and risk level and comfort, with fully autonomous actions being taken. We've seen some of that right now with computer, where users will say, you know, here's some amount of money I'm going to put in my Polymarket account. So I have this general thesis trade whatever contracts you want on PolyMarket in order to make me as much money as possible. You don't need my authorisation every time you want to place one of these contract positions.

We've already seen people doing like that. We have some companies like public Built a skill for computer that lets you set up a genetic trading so you can do the same thing. You can set up an alert, like if you want to go back to that Javier Millei example, it's like, okay, if Millei is on the agenda in Saudi Arabia with a lithium company CEO, place X, Y and Z trades or even not even specify but place the trades that would maximize my returns depending on the situation, right? I think we're going to see a lot more like that. That can be done proactively for users who want to set it up that way, if they're okay with the credit consumption and, you know, things being done on a recurring schedule, and we're going to see more and more direct integrations with brokerages where you can actually move down the funnel vertically, so to speak, into, okay, computer can just make a tool call to take actions on your behalf.

So, I think big picture one of the changes. We'll see. We call it personal CFO. I think it will start to look like almost as if you had a family office for yourself that works 24 over seven and is constantly surfacing opportunities to you. And so even if you haven't gone in and set up that very specific hobby or meal alert. It knows your portfolio. It knows you have some relevant positions. And whenever some big news development happens, you might just get a text or a notification that says, hey, here's what happened. This is actually pretty relevant to your thesis and your goals. Here is a set of actions I'm recommending. Would you like to do it? Click here to approve. That's the same thing that a family office can do for you and should be doing for you, right? Except they're not going to be doing it at like 330 in the morning if that's when it happened on the other side of the world. So, I think we're just going to keep moving in that direction.

And for the users who have comfort and tolerance with, with things being run proactively, even maybe without themselves in the loop if they've set it up that way. I think more and more is going to be proactive and automated.

Lex Sokolin:
Quite a world we're moving into. Thank you so much for joining us today. If we want to learn more about Perplexity Computer and the financial offerings, where should our listeners go?

Jeff Grimes:
Just go to perplexity. AI that's the perplexity website where everything lives. You can sign up for a pro or a max account, which is what you need to use. Computer. All the finance tools I mentioned off the shelf. You don't have to do any setup. You don't need to pay anything extra beyond having a pro or max subscription. They just work automatically so you can get started literally in a few minutes and start, you know, creating a financial app with computer doing financial research. Perplexity AI.

Lex Sokolin:
Magic. Jeff, thank you so much for joining me today.

Jeff Grimes:
Thanks for having me, Lex.

Postscript

• Sponsor the Fintech Blueprint and reach over 200,000 professionals. 👉 Reach out here.

• Read our Disclaimer here — this newsletter does not provide investment advice

• Contributors: Lex, Laurence, Matt, Farhad, Daniel, Michiel

• For access to all our premium content and archives, consider supporting us with a subscription.

Upgrade to Full Access

Let me vent please.

I don’t believe all that I write today, but I have to say it anyway. One year, it’s a post celebrating BTC at $100K. A poem to bubbles!

Another year, everything is dead.

Are we not tired of the songs, constantly changing their tune? I don’t know the truth, but I know I have to sing them.

Today we are diving into the following topics:

• Stripe, PayPal, and Digital Commerce: We examine reports that Stripe, Advent International, and other investors are pursuing a $53B take-private acquisition of PayPal, highlighting the changing economics of digital payments. We compare Stripe, PayPal, Visa, Mastercard, and Adyen, arguing that payment networks continue to compound value through their dominant position in the transaction stack while payment processors increasingly compete on growth and execution.

• The DeFi Post-Mortem: We revisit the original promise of decentralized finance and ask why a movement that aimed to rebuild global financial infrastructure has struggled to achieve mainstream adoption. We argue that DeFi’s contraction from roughly $180B to $80B reflects structural issues—including poor incentives, weak governance, hacks, and speculation—rather than a failure of programmable finance itself.

• Topics: Stripe, PayPal, Advent International, Block, Visa, Mastercard, Adyen, Tempo, Venmo, Braintree, Google Pay, Apple Pay, Tether, Circle, Ethereum, Ant Financial, Set Protocol, Yearn, Aave, Compound, Bitcoin, Zapper, Zerion, Axie Infinity, Base, Solana, Pump.fun, Robinhood, Kelsier Ventures, Binance, BitGet, Hyperliquid

Thanks as always for your time and attention,
Lex

To support this writing and access our full archive of newsletters, analyses, and guides to building in the Fintech & DeFi industries, see subscription options below.

Explore Deeper Insights

🤖🏦🧭 Our Ecosystem:

Generative Ventures | AI Research | Robot Money | Linkedin & Twitter | Sponsors

• 🚀 Lex is actively meeting teams building machine-native financial companies. If that’s you, reach out here.

• 🤖 AI & Robotics Industry Encyclopedia: In this 100+ page report, we collate all the meaningful companies across public quities, private equity, and digital assets related to the machine economy. 👉Learn more here

• 🦄 In partnership with InterPrivate, we just launched a $200MM SPAC looking for targets in fintech, digital assets, and AI infrastructure. If you are a late-stage founder or have an idea to discuss, reply directly to this email.

Stripe, Paypal, and Digital Commerce

We are going to do this as a quick hit.

Payments giant Stripe and private equity firm Advent International, as well as some other potential participants like Block, are making a bid to take PayPal private. PayPal is a $50B public company, which is a far cry from its $360B heyday. For comparison, the card networks, Mastercard and Visa, are both around $500B in market capitalization. Adyen, a Stripe competitor, floats around $35B.

It can be hard to be public. Stripe gets to enjoy its $160B private valuation and a much less critical venture capital investor.

The idea is that a consortium of Stripe and Advent will put together a buy-out package for PayPal’s entire public marketcap with a small premium — about $17B of equity and $36B of debt and other funding, for a total of $53B. It’s unclear why Stripe doesn’t just acquire the company outright, but maybe there are monopoly concerns around such a consolidation. The entity would be run separately, but obviously would be jointly owned. This is a similar playbook to Tempo, as a separate but controlled subsidiary of Stripe, targeting the retail consumer and taking out one of the smaller competitors (Venmo/Braintree legacy).

The numbers are ... pretty hilarious?

Stripe is dwarfed only by Visa and Mastercard. If you look at 5-year stock performance, you can also see that only the card networks retained value while the rest of the payment processing stack bled out.

Why you ask? Because they have very different positions in the value chain.

The networks have full market penetration in a duopoly, and are thus invoked whenever anything new happens anywhere with payments. The only exception is crypto. So whether it’s PayPal, Stripe, or Adyen does not matter to Visa and Mastercard. They grow regardless and therefore are immune to market share fights of the orchestrators above them.

So downstream you have Stripe, PayPal and Adyen with similar-ish $1.5T-$2T volumes, but the rate of change and growth is very different. PayPal is run by professional managers, and its genius cohort — Elon Musk and Peter Thiel — has obviously moved on to run the world and launch space rockets instead.

Stripe, on the other hand, still runs on founder mana and will outperform professional managers any day of the week.

Let’s look at the companies based on high-level fundamentals.

From a revenue perspective, PayPal is the bigger company — $33B in gross revenue, of which $15B is net. But it trades at a 1.2x gross revenue multiple, because it has struggled to grow and is instead being chipped away at by various competitors, from Stripe, to the impenetrable Google Pay and ApplePay. Once a hyperscaler eats a feature, good luck retaining independent market share. And all the payment processing startups that target checkout are effectively eating PayPal’s lunch.

Whereas Stripe continues to grow, and with its $6B of revenue, it gets to trade at 27x revenue, nearly 3 times more than its public counterpart Adyen. This whole thing is a multiples game, and multiples are a derivative of cost of capital, which is dependent on growth assumptions.

It is unclear whether PayPal can do better, but this is not a particularly great outcome.

Maybe Tether and Circle can create a consortium and buy it out for $100B and stuff it full of stablecoins and DeFi lending protocols — this is the only path to the next stage in my view. But thar be dragons too.

Long Take: The DeFi Post-Mortem

I came to this industry as a robo advisor founder watching neobanks, and the conclusion was simple: you cannot build the all-in-one fintech app without a complete refresh of financial infrastructure.

The amount of scotch tape needed to glue together core banking systems to front-ends is just too high.

China had done it — Ant Financial was the first example of a real financial super app — and meanwhile, our own Fintechs were narrow and verticalized.

So seeing financial applications on Ethereum was a breath of fresh air. We suddenly had something modern on which to build programmable money.

Then DeFi summer happened, and the dream became clear.

Read more

Today, we highlight the below items:

1. DIGITAL ASSETS: Why Kalshi at $40B is still worth twice as much as Polymarket

2. ANALYSIS: Architect and the $11 Trillion AI Capital Markets

3. CURATED UPDATES: Financial Institutions and Adoption; DeFi and Digital Assets; Blockchain Protocols; NFTs, DAOs and the Metaverse

To support this writing and access our full archive of newsletters, analyses, and guides to building in Fintech & DeFi, subscribe below (if you haven’t yet).

Subscribe now

🤖🏦🧭 Our Ecosystem:

Sponsor FB | Generative Ventures | SPAC | AI Research | Robot DeFi | Linkedin & X

• 🚀 Lex is actively meeting teams building machine-native financial companies. If that’s you, reach out here.

• 🤖 AI & Robotics Industry Encyclopedia: In this 100+ page report, we collate all the meaningful companies across public quities, private equity, and digital assets related to the machine economy. 👉Learn more here

• 🦄 In partnership with InterPrivate, we just launched a $200MM SPAC looking for targets in fintech, digital assets, and AI infrastructure. If you are a late-stage founder or have an idea to discuss, reply directly to this email.

DIGITAL ASSETS: Why Kalshi at $40B is still worth twice as much as Polymarket

Kalshi is raising again.

Seven weeks after closing a $1 billion Series F at $22 billion, the FT reports the prediction market operator is in talks at roughly $40 billion, with a close possible as soon as Q3. This is news about potential news!

For context: Kalshi was worth $5 billion last October, $11 billion in December, and $22 billion in May. An eightfold repricing in about eighteen months is very unusual. Reportedly, the company is now in talks with banks about an IPO, which CEO Tarek Mansour says could come after 2027.

We have covered both Kalshi and Polymarket before, usually through the lens of regulation and volume share.

This time we want to look at something more basic: the two companies have arrived at similar places — enormous institutional cap tables and ten-figure-plus monthly volumes — running fundamentally different economic machines.

• One charges roughly 1% and books $2 billion in annualized revenue

• The other spent five years charging nothing at all, while flirting with crypto

Kalshi’s machine is simple: it is a CFTC-regulated exchange that takes roughly 1% of every matched trade. No house risk, no balance sheet exposure to outcomes — just a toll on volumes, which have been extraordinary.

Annualized trading volume hit $178 billion by April 2026, up 32x year-on-year, and the company surpassed $2 billion in annualized revenue. At the $40 billion ask, that is roughly 20x revenue. Expensive, but not unhinged — Coinbase and Robinhood have traded at comparable multiples.

When we covered these two last time, Kalshi was worth twice as much as Polymarket. Since then, the valuation ratio has barely moved. The quality of that revenue is the real question. Are these just enormous online casinos, or something more fundamental related to information and financial markets?

For Kalshi, sports contracts account for roughly 65% of trading volume, which has gotten them in trouble.

Kentucky’s lawsuit alleges 89% of 2025 volume came from sports, which are precisely the contracts states are trying to ban. Arizona filed criminal charges in March, a Massachusetts judge barred Kalshi’s sports markets in January, and Nevada has extended its ban. The company’s entire legal theory — that event contracts are swaps under exclusive CFTC jurisdiction — is being litigated in more than a dozen states simultaneously, with a Michigan federal judge recently ruling that sports prediction markets are not swaps.

This is heading to the Supreme Court, and the big valuation requires winning the legal argument.

Kalshi has built a genuinely profitable marketplace whose economics depend on this classification. The 1% take rate is only defensible while Kalshi is the sole federally-blessed venue. DraftKings launched its own exchange this week, Robinhood is in event contracts, and CME is suing the CFTC over Kalshi's crypto perpetuals.

A courtroom victory invites the competition, which then erodes the 1% take rate. A defeat threatens the volume it's charged on.

Polymarket ran the opposite playbook.

When we analysed Intercontinental Exchange's (ICE) $2 billion investment last October after the NYSE parent took roughly 17% of the company, Polymarket had roughly 300,000 monthly active users, $1.3 billion in monthly volume, and no material revenue. It was a $9 billion valuation priced on cultural relevance and optionality.

Five years of zero fees was a deliberate liquidity subsidy, funded by venture capital, and it partially worked: 2025 volume reached $21.5 billion, nearly half the global prediction market total. The monetization switch finally flipped this year, and relative volumes have shifted back into Kalshi’s favour since.

Taker fees arrived on high-frequency crypto markets in January, and by early February weekly fee revenue exceeded $1.08 million. The full schedule (V2, effective March 30) now covers most categories, while the US exchange charges a uniform 0.05 taker fee with a maker rebate that dramatically undercuts Kalshi’s roughly 1% take. That undercut is the strategy: Polymarket is pricing to take share, not to maximise near-term revenue. Independent modelling suggests about $100MM-450MM of revenue.

Against a $15 billion valuation, even the bull case is north of 30x revenue and the base case is above 70x. So the fees don’t fully support the valuation. Two other things do.

First, data. ICE built half its business on selling market data. As part of the investment, ICE became the exclusive global distributor of Polymarket’s event-driven data to institutional clients: real-time probabilities on elections, Fed decisions, geopolitical events, delivered through the same infrastructure that carries NYSE prices. The offering was productised in February as Polymarket Signals and Sentiment — normalized feeds and historical time series that a hedge fund can plug into a model.

The pitch is that a liquid market beat the pollsters in the 2024 election, and that Polymarket can continue to deliver. A client can pay for the signal without ever placing a bet, which decouples revenue from volume entirely.

The second mechanism by which Polymarket is differentiated is the POLY token, planned for a 2026 launch, would route fee flow to stakers and let the company monetise its treasury position, which is a value-capture mechanism Kalshi structurally cannot replicate.

Put the three machines side by side.

• DraftKings, the mature business, trades at 3x revenue because the market knows exactly what a sportsbook is and how much it earns (plus state licenses, promo spend, house risk).

• Kalshi at 20x is priced as an exchange and generates real revenue. Exchanges get exchange multiples, bookmakers get bookmaker multiples, and more than a dozen states are litigating with Kalshi.

• Polymarket at 70x+ on base-case fees is not priced on fees at all. It is priced on the data business, the token, and the possibility that the cheapest venue could win on volume.

All these bets still depend on outcomes we cannot fully foresee — a Supreme Court ruling for Kalshi, a token launch for Polymarket. Fittingly, their entire product is trading on things that haven’t happened yet.

👑Related Coverage👑

Analysis: Architect and the $11 Trillion AI Capital Markets

We examine the emergence of capital markets built specifically for AI infrastructure rather than AI applications, arguing that GPUs, compute capacity, and inference have become financial assets in their own right.

We discuss how an estimated $11T will be invested in AI infrastructure between 2024 and 2029, including roughly $7T financed by debt, creating demand for hedging, derivatives, and new financial markets. We highlight Architect’s launch of ComputeConnect, a regulated exchange linking GPU derivatives with physical compute delivery, as an example of crypto-native market structures expanding into AI infrastructure. Financial markets around AI infrastructure could become as important as the AI applications themselves by bringing price discovery, liquidity, and risk management to the largest technology investment cycle in history.

Read Analysis

Curated Updates

Here are the rest of the updates hitting our radar.

Financial Institutions and Adoption

Nasdaq brings proprietary market data onchain through Pyth - CoinTelegraph

Uniswap, Spark aim to build stablecoin FX market as banks, fintechs enter the industry - CoinDesk

Stablecoin issuer Circle just got the greenlight to operate as a bank. The shares are up 5% - CNBC

BlackRock Adds Ethena’s Synthetic Dollar to Its $20T Aladdin Risk Management Platform - The Defiant

DeFi and Digital Assets

Aave Advances Automated AAVE Buyback Overhaul With Aavenomics 3.0 - The Defiant

Ondo Finance Launches 24/7 Minting and Redemption for Tokenized US Stocks and ETFs - Converge

Spark, Uniswap build stablecoin ‘FX Layer’ seeded with $150 million liquidity migration - The Block

Blockchain Protocols

Ethereum gets a new nonprofit focused on institutional adoption - CoinDesk

Glamsterdam update - Ethereum

Drift Protocol relaunches as Velocity Dex with Tether credit line - Cryptopolitan

NFTs, DAOs and the Metaverse

What is a governance attack? How BonkDAO lost $20M in a single vote - crypto.news

🚀 Level Up

Join our Premium community and receive all the Fintech and Web3 intelligence you need to level up your career. Get access to Long Takes, archives, and special reports.

Upgrade to Premium

• Sponsor the Fintech Blueprint and reach over 200,000 professionals.
  
  👉 Reach out here.

• Check out our AI newsletter, the Future Blueprint, 👉 here.

• Read our Disclaimer here — this newsletter does not provide investment advice

• Contributors: Lex, Laurence, Matt, Daniel, Michiel

Fintech Blueprint 🤖🏦🧭 is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

Subscribe now

Today we are diving into the following topics:

• Summary: We examine the emergence of capital markets built specifically for AI infrastructure rather than AI applications, arguing that GPUs, compute capacity, and inference have become financial assets in their own right. We discuss how an estimated $11T will be invested in AI infrastructure between 2024 and 2029, including roughly $7T financed by debt, creating demand for hedging, derivatives, and new financial markets. We highlight Architect’s launch of ComputeConnect, a regulated exchange linking GPU derivatives with physical compute delivery, as an example of crypto-native market structures expanding into AI infrastructure. Financial markets around AI infrastructure could become as important as the AI applications themselves by bringing price discovery, liquidity, and risk management to the largest technology investment cycle in history.

• Topics: Architect, Compute Index, ComputeConnect, NVIDIA, CoreWeave, Blackstone, Oracle, SpaceX, Meta, Hyperliquid, Anthropic, OpenAI, Ornn, Andreessen Horowitz, Brett Harrison.

To support this writing and access our full archive of newsletters, analyses, and guides to building in the Fintech & DeFi industries, see subscription options below.

Subscribe now

🤖🏦🧭 Our Ecosystem:

Generative Ventures | AI Research | Robot Money | Linkedin & Twitter | Sponsors

• 🚀 Lex is actively meeting teams building machine-native financial companies. If that’s you, reach out here.

• 🤖 AI & Robotics Industry Encyclopedia: In this 100+ page report, we collate all the meaningful companies across public quities, private equity, and digital assets related to the machine economy. 👉Learn more here

• 🦄 In partnership with InterPrivate, we just launched a $200MM SPAC looking for targets in fintech, digital assets, and AI infrastructure. If you are a late-stage founder or have an idea to discuss, reply directly to this email.

Long Take

AI for Finance or Finance for AI

There are many intersections between AI and financial services.

One easy way to segment the categorization is to think about (1) AI that lives within financial services as a business or a function, and (2) finance that works at the behest and is custom-built for AI.

In Category 1, the examples are pedestrian and well-known by now — adding AI into underwriting, deploying OCR on documents, integrating larger models to prevent fraud, extending customer service.

Much more is going on in getting AI to do the business of running a financial company, such as organizing workflow or writing code. We talk about this as financial harnesses, meaning a chassis by which LLMs can be your colleagues in the finance profession. This is more akin to being a functional contributor inside of an organization, than about some sub-process of a manufacturing process.

In Category 2, we look to add financial features to an AI-first world. One of these has been commerce and payments infrastructure for agents. Whether or not this category is able to hold water is yet to be determined. However, the combination of stablecoins on various chains, payment processors, and commerce enablers has chased down the potential blue ocean with abandon. Even the large tech companies have shipped various agentic identity protocols to bootstrap our imagination.

It all feels a bit ahead of demand.

One area we haven’t spent as much time on, but is equally large, is the effect of AI on capital markets. And I don’t mean how good an AI agent is at arbing this or that — such a product is squarely in category 1. From the venture side, I have seen dozens of companies building AI-first roboadvisors, neobanks, and neobrokers and calling them by cooler names. That’s not what I mean.

Rather, AI needs capital markets for its own assets.

In the past, we wrote about this as a warning. In particular, we amplified the story that:

• There’s huge spend on GPUs to train models and provide inference

• Many people think that will never be profitable

• It’s true, validated by our machine economy research, that money is concentrating in components, chips, datacenters and applications are not yet catching up

• To fight each other, hyperscalers have spent down all their cash and raised debt

• That debt is circular between NVIDIA and its customers

• Companies are staying private longer to avoid public market scrutiny and to access less informed / cheaper capital to keep this going, and everyone is monetizing via secondary SPVs instead of public equity

• Private equity firms like Blackstone are neck deep in GPU built out credit risk

It was a good article! We wrote about this a lot last year. But really, nobody cared.

If anything, this stuff reads a bit quaint compared to the monster numbers everyone is putting up today. Semianalysis projects $11 trillion in spending on AI infrastructure between 2024 and 2029, of which $7 trillion is financed by debt.

That seems pretty wild, unless you believe this is existential, humanity-survives-WW3 type stuff. Which is of course how it is being sold straight into our 401ks. And it is also how an enormous amount of liquidity, meaning proceeds from investment rounds, are going into AI companies and venture more generally.

Michael Burry has been adamant about trying to time the short on the AI bubble. While the fundamental story may be there, and the zeitgeist is emerging, so far no real luck. Everything is still going up! To me, it feels like the top of Softbank’s investment in WeWork, when everyone just said OK, this stuff is 50x revenues. Fine.

And as the numbers go up, financial markets begin to form around them.

We are at an interesting intersection of technologies, because some of these products from crypto are now at the heart of the AI markets in ways people don’t see.

Read more

Robinhood showed their hand in DeFi last Wednesday as they officially launched their own blockchain “Robinhood Chain”. The company’s execs, including co-founder Vlad Tenev, took the stage in a well-rehearsed theatrical display at the historic Old Royal Naval College in London as they announced a string of new on-chain trading primitives.

The stock is up 10% on the news.

This week we unpack Robinhood’s underlying bet on on-chain finance, what it means in the context of the broader digital asset industry, and how it stacks up against similar offerings from Coinbase.

1. FINTECH: Behind Robinhood’s bet to bring on-chain finance to 27M customers

2. ANALYSIS: Launching a $201MM SPAC for Fintech, AI, and Crypto

3. CURATED UPDATES: Paytech, Neobanks, Lending, Digital Investing

To support this writing and access our full archive of newsletters, analyses, and guides to building in Fintech & DeFi, subscribe below (if you haven’t yet).

Subscribe now

🤖🏦🧭 Our Ecosystem:

Generative Ventures | AI Research | Robot Money | Linkedin & Twitter | Sponsors

• 🚀 Lex is actively meeting teams building machine-native financial companies. If that’s you, reach out here.

• 🤖 AI & Robotics Industry Encyclopedia: In this 100+ page report, we collate all the meaningful companies across public quities, private equity, and digital assets related to the machine economy. 👉Learn more here

• 🦄 In partnership with InterPrivate, we just launched a $200MM SPAC looking for targets in fintech, digital assets, and AI infrastructure. If you are a late-stage founder or have an idea to discuss, reply directly to this email.

Digital Investment & Banking Short Takes

Behind Robinhood’s bet to bring on-chain finance to 27M customers

Economic activity across blockchains has plummeted over the last twelve months. Spot volumes on decentralised exchanges in June were down -71% YoY.

The fees generated by the underlying blockchain networks hosting this activity are likewise down -86% since Jan’25. A major driver behind this fall has been the October 10 crash following US/China trade tensions, which sparked a mass deleveraging event across the ecosystem.

Contagion spread to centralised venues, with a basket of major exchanges like Binance, Kraken, and Coinbase reporting a 50%+ drop in combined spot trading volumes. Robinhood itself reported a -20% fall in quarterly transaction revenue in 1Q26.

This wouldn’t strike investors as the best market backdrop for launching a new chain. But volumes and network activity are reactive indicators. They are responses to past market events, like October 10th.

The trends in venture capital paint a slightly different picture. Data from Galaxy shows VC capital flowing into the crypto industry in 1Q26 annualizing to ~$16B; not far off the $18B recorded in 2025.

The underlying segments are revealing. Firms building financial primitives like trading, lending, and investment products grew their share of VC capital from 10% in 2024 to over 50% today. The industry is converging on the narrative that the next wave of growth in digital assets will be led by financialisation.

This is the bet that Robinhood unveiled with the launch of Robinhood Chain — its own blockchain to distribute tokenized stocks, investment, and derivatives to its 27M customers. This frames chains as a primarily a capital markets venue.

From a technical perspective, Robinhood Chain is a permissionless Layer 2 network, meaning developers and users can interact publicly with it outside the Robinhood app, with transactions ultimately settling on Ethereum.

The company originally launched tokenized stocks for EU customers directly on the Arbitrum network last Summer, but has now completed its migration to Robinhood Chain, which uses the same underlying technology stack. This gives it more customisability to ensure regulatory compliance and more predictable transaction costs.

Coinbase’s Base blockchain pursued a similar path using the competitor Optimism stack. This gives us a glimpse into the economics Robinhood can expect from the network’s transaction (sequencer) fees. Coinbase made around $75M in sequencer revenues from Base in 2025 on the back of a record in transaction volume associated with memecoin and AI-token trading.

The majority of Base activity occurred outside of the core Coinbase app, so it is difficult to estimate the transaction volume Robinhood Chain can achieve. But $75M in annual fees would grow Robinhood’s topline revenue of $4.5B by less than 2%. Nothing to write home about.

The real growth comes from the financial products Robinhood is building on top and distributing directly to users. At launch, these span (1) tokenized stocks, (2) perpetual futures, and (3) yield, with the USDG stablecoin as the connective tissue between them.

Robinhood launched their initial set of 90+ tokenized stocks linked to major companies and ETFs. The tokens are backed 1:1 by the underlying equities and are held by a licensed US custodian. This unlocks both 24/7 trading for users by routing to decentralised exchanges like Uniswap, as well as the ability to borrow against tokens via lending protocols like Morpho.

We think this is closely related to the launch of the company’s native integration of the Lighter perps exchange, which enables users to make bets on tokens with up to 50x leverage. Both borrowing against tokens and accessing leverage via perps builds on one of Robinhood’s fastest-growing products: margin financing.

In 1Q26, the company reported a +75% YoY increase to $193M in interest income on margin financing, making up 18% of total revenues. Stock Tokens and perps give users new ways to access leverage, even though the underlying monetization engine is different.

This is where USDG comes in; the Paxos-issued dollar stablecoin. Robinhood earns net interest income from the short-term Treasuries backing USDG and is positioning it as the default cash token on its network.

Case in point:

• Liquidity for Stock Tokens in Uniswap is against USDG

• Perps markets on Lighter use USDG as collateral

• Borrowing against Stock Tokens on Morpho will be in USDG

The company’s new Yield offering 7% APY (for now) likewise sweeps users’ deposits into a USDG vault on Morpho on Robinhood Chain. The interest comes from a combination of interest income (passed down from Robinhood) and borrower payments from underlying lending markets.

The overarching opportunity is to gradually shift user idle cash balances into the native stablecoin and unlock interest income.

In 1Q26, Robinhood held $26B in customer cash on the platform in its Cash Sweep program, on which it earned $45M. We assume that (1) users allocate 20% of their assets towards this higher-yield (and higher-risk) product, and (2) Robinhood earns 3% APY on USDG on the platform. This would result in the new Yield offering growing quarterly interest income by $156M; a +15% uplift to total revenue.

The upside is larger when accounting for the USDG in the underlying markets for Stock Tokens and Perps. The downside is that the majority of this revenue won’t flow into Robinhood’s bottom line. The current 7% APY at launch is sustained by the company passing the yield back to users.

This is a similar opportunity that Coinbase pursued early-on through its partnership with Circle to promote USDC on the platform. Today, the company earns 22% of its total revenue from interest income passed down by Circle ($305M in 1Q26).

Owning the stablecoin economics is becoming table stakes as traditional fintechs begin to pursue the on-chain opportunity. Stripe’s launch of the Open USD stablecoin with 149 partners last week is a testament to it. Paxos’ USDG is now a core component of Robinhood’s DeFi strategy, adding to its existing partnerships with both Kraken and OKX.

Crypto trading remains just 13% of Robinhood’s total revenue today, compared to CEXs where it is typically 50%+. Based on the annual share performance of HOOD compared to COIN, and CRCL, the market appears to be favouring those who pursue the on-chain opportunity with a diversified trading business to fall back on.

👑 Related Coverage 👑

Analysis: Our SPAC Launch

We discuss the launch of $IPVV, a $201M SPAC that target high-growth companies across fintech, digital assets, AI, and the machine economy. We argue that a growing backlog of venture-backed companies remains private despite reaching meaningful scale, while weak venture DPI and a constrained IPO market have created demand for alternative liquidity pathways.

We review the SPAC structure, and note that today’s SPAC market is materially smaller than the 2020–2021 booms. We conclude that the next generation of SPAC targets is likely to be profitable or near-profitable companies with $50–250MM of revenue and $1–5B valuations rather than the pre-revenue growth stories that defined the prior cycle.

Read More

Curated Updates

Here are the rest of the updates hitting our radar.

Paytech

• ⭐ Stripe’s New Stablecoin Bet: Open USD - Forrester

• MetaMask launches Money Account - Coindesk

• Airwallex hits $11 billion valuation with $320 million raise - CNBC

Neobanks

• ⭐ Elon Musk’s ‘bank’ is live - American Banker

• Revolut to launch in UAE - Bloomberg

• Rain Launches Rewards Loyalty Capability - Finovate

Lending

• Plaid Launches Sequential AI Model - Fintech Finance

• Visa Cards for AI Agents - Forbes

• Barclays to buy GoHenry kids’ debit card and money app - The Guardian

Digital Investing

• Coinbase is launching AI tools that give investment advice - Yahoo Finance

• SpaceX stock tumbles 16.4%, shaving off most IPO gains - Yahoo Finance

• US SEC poised to allow stock token trading - Reuters

🚀 Level Up

Join our Premium community and receive all the Fintech and Web3 intelligence you need to level up your career. Get access to Long Takes, archives, and special reports.

Upgrade to Premium

• Sponsor the Fintech Blueprint and reach over 200,000 professionals.
  👉 Reach out here.

• Check out our AI newsletter, the Future Blueprint, 👉 here.

• Read our Disclaimer here — this newsletter does not provide investment advice

• Contributors: Lex, Laurence, Matt, Farhad, Daniel, Michiel

Today’s agenda is below.

1. AI: Per-token prices are collapsing, yet AI bills are exploding at the same time.

2. REPORT: The State of Finance for Marketers, and where Banks are beating Fintechs

3. PODCAST: Inside the $1B-a-Day Stablecoin Market Maker for 1,500 Institutions, with B2C2’s Cactus Raazi

4. CURATED UPDATES: Machine Models, AI Applications in Finance & Investment Outlook

To support this writing and access our full archive of newsletters, analyses, and guides to building in Fintech & DeFi, subscribe below (if you haven’t yet).

Subscribe now

🤖🏦🧭 Our Ecosystem:

Generative Ventures | AI Research | Robot Money | Linkedin & Twitter | Sponsors

• 🚀 Lex is actively meeting teams building machine-native financial companies. If that’s you, reach out here.

• 🤖 AI & Robotics Industry Encyclopedia: In this 100+ page report, we collate all the meaningful companies across public quities, private equity, and digital assets related to the machine economy. 👉Learn more here

• 🦄 In partnership with InterPrivate, we just launched a $200MM SPAC looking for targets in fintech, digital assets, and AI infrastructure. If you are a late-stage founder or have an idea to discuss, reply directly to this email.

Cheaper tokens are NOT leading to cheaper AI bills

Nvidia has put its next data-centre platform Vera Rubin into full production.

The headline was the 10x reduction in the cost of generating a token, achieved by tripling the memory bandwidth feeding each chip. By token, we mean the AI-inference kind of intelligence token, not digital assets.

Jensen Huang called it “token factory economics”, and described the modern data centre as a machine whose output is measured in tokens per watt.

In the same few weeks, the buyers of that output were reaching a different conclusion. JPMorgan’s technology desk sent clients a note titled “AI Bills Are Out of Control”, reporting that some investment analysts were running personal token budgets into the tens of thousands of dollars, a handful past $100,000. Uber’s chief technology officer told The Information he had burned through the company’s entire 2026 AI budget in four months... A four-person startup posted a $113,000 monthly bill from a single model provider.

The gap between the cheapening of tokens and rising AI bills is where finance teams have to step in.

The key numbers of Unit Price x Consumption

Most coverage of the cost of AI folds three numbers into one.

The first is unit price, which is falling fast. Andreessen Horowitz’s Guido Appenzeller measured a tenfold annual decline in the cost of a given level of model performance, from $60 per million tokens in 2021 to $0.60 three years later.

Epoch AI, working from benchmark data rather than list prices, found the cost of reaching a fixed performance level halving roughly every two months, with declines ranging from 9x to 900x a year depending on the task.

On any measure, intelligence per dollar is getting cheaper faster than compute did during the personal-computer era.

The second number is consumption, and it is climbing at least as fast. Goldman Sachs Research, in a May report modelling the agentic economy from the bottom up, put token consumption on course to multiply 24x times by 2030, to 120 quadrillion tokens a month.

Estimated monthly token count for Agentic AI applications

An agent reads and re-reads context, calls tools, checks its own work across several rounds, and often runs without being asked. Goldman’s own worked example of a software-development agent gets through roughly 6M input tokens and 820K output tokens in a day.

The third number is the one a finance director reviews — that is the bill. It is the product of the other two, and because consumption is rising faster than prices fall, the bill is getting bigger each month.

What sets the price, and where it goes

Generating a token is not, for the most part, an arithmetic problem. A model produces one token at a time, and for each one, it must read its weights and accumulated context out of memory. The calculation involved is small next to the cost of moving that data, so an accelerator running inference typically uses only thirty to forty per cent of its theoretical compute, and the rest of the time it waits for memory.

Inference is bound by memory bandwidth.

The store of context a model holds grows with the length of the exchange and the number of requests handled at once, and now routinely takes more memory than the model’s own weights. Almost all of the engineering that has made tokens cheaper, from caching to quantisation to smarter batching, amounts to managing memory traffic.

This is why Nvidia’s tenfold claim for Vera Rubin rests on memory rather than maths. The platform triples per-chip bandwidth to 22 terabytes a second with HBM4. That change carries most of the savings, as memory is now the largest line in the price of an AI server, up more than four hundred per cent over the previous generation.

Then comes energy usage. One peer-reviewed paper estimates a single frontier query at about a third of a watt-hour, if the reasoning is then stretched to 15x the tokens, then the energy rises roughly 13x. If you wrap the model in an agent that loops through tools and checks its own answers, one request can use 60x to a 140x the energy of a one-shot reply.

While each token becomes cheaper, the models are making each task consume more of them.

The financial comparison

Arvind Jain, who runs the enterprise-search firm Glean, told CNBC this was the first time in his career that technology cost about the same as people, and that the choice between the two was being put directly to finance teams. Nvidia’s own Bryan Catanzaro said that for his deep-learning group, the cost of compute already ran far beyond the cost of the employees.

However, Gartner’s forecast that a developer’s token spend will match or exceed a software engineer’s salary by 2028, rests on a global-average wage of $2,000 a month, not a Western one. In London or New York, token spend is not about to overtake a six-figure salary, but in Bangalore it may already match a mid-level engineer’s pay, because the token price does not vary by geography even though the wage does.

The dominant enterprise pattern is now the advisor model, in which a cheap open-weight models handle the bulk of the work and calls a frontier model only when stuck, capable of cutting bills by an order of magnitude through routing alone. Chinese open-weight models such as DeepSeek and Zhipu now land within a point or two of the Western frontier on Agentic benchmarks at a 1/5 of the cost or less.

Key Takeaways

So what should we do with the information that AI may be ramping up costs as much as it is cutting them?

• Investors: Watch inference margin. The valuations floated for the model labs ahead of their listings assume pricing power. AI-native software carries 40% - 50% of revenue as inference cost, against 10% - 20% for classic SaaS models.

• Founders: Track cost per completed task, not per seat, and keep the model layer swappable. Context engineering is now a cost centre that needs human management.

• Operators: The cost implications are strongest where compliance overhead is heaviest, so price agents against the human cost, and build budgets on the assumption that frontier-reasoning unit cost may not fall in the future.

If the marginal token is increasingly priced by memory bandwidth and electricity rather than by the model itself, the cost advantage in AI-driven finance accrues to whoever controls power and routing, not to whoever trained the cleverest model. For an industry that has spent two years worrying about its dependence on a few model providers, price may be the tipping point.

Further, this creates opportunities for net new human jobs.

Recent data from Ramp shows that companies that spend more on AI also increase their staff numbers across all jobs. Good news at last — we can be robot shepherds.

👑Related Coverage👑

Report: The State of Finance for Marketers, and where Banks are beating Fintechs (link here)

Today, we are excited to highlight AppsFlyer’s State of Finance for Marketers: Europe 2026 report.

It tracked 2.4 billion installs and roughly a billion dollars of spend to figure out where Europe’s finance marketing budget is actually going.

There are some pretty fascinating metrics in here, so let’s dig in. Or, if you want to just see the numbers, click below:

Read Report

🎙️ Podcast: Inside the $1B-a-Day Stablecoin Market Maker for 1,500 Institutions, with B2C2’s Cactus Raazi (link here)

Hi Fintech Architects,

In this episode, Lex chats with Cactus Raazi — CEO Americas at B2C2, one of the original and largest institutional market makers in digital assets, serving roughly 1,500 institutions and pricing across more than 40 exchanges globally.

They discuss what a market maker actually does, how balance sheet and signal generation underpin roughly $1 billion a day of stablecoin flow at B2C2, and why the two extremes of crypto market making - riskless principal aggregation versus proprietary alpha - produce very different client outcomes that buyers rarely understand.

Cactus explains B2C2’s 18-month bet that the Circle-versus-Tether debate would give way to a multi-issuer world, the launch of its PENNY product for instant zero-cost cross-stablecoin swaps, and they explore why programmability is the next frontier for digital dollars, why US capital markets have almost no structure for funding genuine risk-taking businesses, and whether the current combination of scale, speed, and complexity makes this the hardest investing environment Wall Street has ever faced.

Listen to Podcast

Curated Updates

Here are the rest of the updates hitting our radar.

Machine Models

Ethical and Bias Considerations in Artificial Intelligence/Machine Learning - Matthew G. Hanna & Liron Pantanowitz & Brian Jackson & Octavia Palmer & Shyam Visweswaran & Joshua Pantanowitz & Mustafa Deebajah & Hooman H. Rashidi

A Critical Field Guide for Working with Machine Learning Datasets - Sarah Ciston & Mike Ananny & Kate Crawford

AI Applications in Finance

⭐ AI-Driven Payment Systems: From Innovation To Market Success - Merve Ozkurt Bas

The Rise Of Generative Ai Agents In Finance: Operational Disruption And Strategic Evolution - Inesh Hettiarachchi

Financial Modeling in Corporate Strategy: A Review of AI Applications For Investment Optimization - Olufunmilayo Ogunwole & Ekene Cynthia Onukwulu & Micah Oghale Joel & Ejuma Martha Adaga & Augustine Ifeanyi Ibeh

Investment Outlook

⭐ Private Equity Outlook 2025: Is a Recovery Starting to Take Shape? - Bain & Company

⭐ Global Venture Capital Outlook: The Latest Trends - Bain & Company

⭐ Global Private Markets Report 2025: Braced for shifting weather - McKinsey & Company

🚀 Postscript

• Sponsor the Fintech Blueprint and reach over 200,000 professionals.
  👉 Reach out here.

• Check out our new AI products newsletter, Future Blueprint. (Don’t tell anyone)

• Read our Disclaimer here — this newsletter does not provide investment advice

• Contributors: Lex, Laurence, Matt, Farhad, Daniel, Michiel, Luke

• For access to all our premium content and archives, consider supporting us with a subscription. In addition to receiving our free newsletters, you will get access to all Long Takes with a deep, comprehensive analysis of Fintech, Web3, and AI topics, and our archive of in-depth write-ups covering the hottest fintech and DeFi companies.

Upgrade to Premium

Today we are diving into the following topics:

• Summary: In this article, we discuss the launch of $IPVV, a $201M SPAC that target high-growth companies across fintech, digital assets, AI, and the machine economy. We argue that a growing backlog of venture-backed companies remains private despite reaching meaningful scale, while weak venture DPI and a constrained IPO market have created demand for alternative liquidity pathways. We review the SPAC structure, and note that today’s SPAC market is materially smaller than the 2020–2021 booms. We conclude that the next generation of SPAC targets is likely to be profitable or near-profitable companies with $50–250MM of revenue and $1–5B valuations rather than the pre-revenue growth stories that defined the prior cycle.

• Topics: InterPrivate, Generative Ventures, Aeva, Microvast, Securitize, Abra, Animoca Brands, Kraken, Cantor Fitzgerald, BlackRock, Anthropic, OpenAI, SpaceX

To support this writing and access our full archive of newsletters, analyses, and guides to building in the Fintech & DeFi industries, see subscription options below.

Subscribe now

🤖🏦🧭 Our Ecosystem:

Generative Ventures | AI Research | Robot Money | Linkedin & Twitter | Sponsors

• 🚀 Lex is actively meeting teams building machine-native financial companies. If that’s you, reach out here.

• 🤖 AI & Robotics Industry Encyclopedia: In this 100+ page report, we collate all the meaningful companies across public quities, private equity, and digital assets related to the machine economy. 👉Learn more here

Long Take

Launching our SPAC

I am excited to share that, in partnership with InterPrivate and Ahmed Fattouh, we have launched a $201MM SPAC, now live at $IPVV.

I’m writing to explore in more detail my thinking on the SPAC market, why we launched a new vehicle, and the market dynamics.

At a high level, we are looking for unicorn companies in fintech, digital assets, AI, and the broader machine economy to bring to the public equity markets.

The last 5 years have been pretty wild.

On the one hand, we have seen the blow-ups of multiple companies, banks, digital asset treasuries, and crypto assets. You had to have both vision and skill to survive, and a handful of companies (e.g., Hyperliquid, Polymarket) thrived on the volatility. On the other hand, an entire AI industry has been born, with the highest capital investment in technological infrastructure I have ever seen. This boom has already been expressed in energy and hardware prices, and several trillion of marketcap have been generated in just a handful of years.

Through this, I have seen founders build out strong ventures, sometimes scrambling through the fog but climbing out with successful businesses. They had done it through aggressive regulatory regimes, changing consumer expectations and behavior, and shifting global macro-economic relationships.

• Fintech neobanks that have built $100MM+ revenues and profitability, using stablecoins or social media as leverage points against incumbents

• New crypto infrastructure supporting billions in payment and trading volumes, connecting parts of the world others found hard to reach

• Neoclouds, as well as energy and memory companies, scaling up to support the enormous AI build-out

• AI-native applications that find incredible product-market-fit with coding, video, marketing, or other general employee tools

While some of these companies have gone public, been acquired, or merged into DATs, most still remain private. Everyone is waiting. SpaceX, Anthropic, and OpenAI are the big planes on the runway that must go public before anyone else has a chance for liquidity.

This means that investors will be holding mega-caps, both through retail participation and index investing, rather than buying companies in the $500MM to $5B range — a price that would allow more price discovery in the public, rather than private, markets.

The other effect of this is on venture capital. As founder of Generative Ventures, I’ve learned that DPI — the cash you return to investors — is the lifeblood of your LP relationships. If you don’t return DPI, then your investors will struggle to re-invest in your next vehicle. And the 2020+ vintage venture funds are struggling on DPI.

This means venture funds need to exit in order for the industry to keep raising money and investing in high-risk innovation. As a result, there’s a bifurcation between funds that have invested in the AI champions — increasingly only the mega funds — and vertical specialists who are not seeing liquidity. The result is widely overpriced seed rounds and consensus bets.

Given the existence of a whole bunch of high-quality companies (on fundamentals), the IPO window constraints, and the need for investors to exit, I think this is a moment when the SPAC structure adds a lot of value.

SPAC Sponsors and Market

The way a SPAC works is in roughly in these steps:

1. A sponsor team brings a vehicle to market through an IPO process with an investment bank. The vehicle has 2 years to find a deal, and retains a trust of some amount of capital (e.g., $200MM) that can be committed to the deal.

2. The team either (a) finds a target and agrees on a transaction with the company, or (b) is unable to find a deal within the timeline and returns money to investors

3. If there is a deal, trust investors may redeem (i.e., withdraw their money) if they do not like the deal, or for other reasons. The company and sponsor team usually raise a PIPE in addition to the trust capital to cover any gaps.

InterPrivate is a repeat SPAC sponsor, which means this is the 5th vehicle from the company. The platform’s edge is a co-sponsor network — independent sponsors, family offices, founders, and PE/VC firms — that widens the funnel and adds operating leverage and domain expertise, with the ability to support transactions with PIPE (private investment in public equity) capital from family-office LPs.

In the prior SPAC cycle, it had launched several successful searches, like Aeva — a 4D LiDAR-on-a-chip company founded by ex-Apple executives. Aeva closed a $1.8B merger in March 2021 and raised a $120MM PIPE, and a $200MM follow-on thereafter. InterPrivate also backed Microvast / Tuscan, a next-gen EV battery technology, which did a $3B merger in July 2021 and raised a $483M PIPE.

It had also run vehicles where the market timing didn’t work, and the capital was returned to investors.

Timing matters a lot.

The IPO window is open, until it is closed shut for years to come.

Market View

The last SPAC boom had come in 2020 and 2021, when nearly 800 vehicles were raised after the growth of digital adoption during COVID. Neobank and neobroker metrics were through the roof, as people stayed home and became chronically online.

Zoom stock soared!

SPAC trust proceeds exceeded $200B in raised capital over those years. For context, that’s less than Goldman Sachs thinks we will IPO in 2026 alone in AI stocks.

However, many companies came to market that were not really ready to be public.

They used SPACs to share overly optimistic projects and bypassed the usual diligence of the going-public process. Many of them were venture capital-style bets, pre-traction and pre-revenue. When the digital shut-in culture of the pandemic years stopped, we saw significant fallout in companies that did not support their prices with fundamentals.

That creates a lesson for this time around — focus on operating excellence and documented growth. Overall capital raised into SPACs has been more limited — $25B in 2025 and $16B in 2026 so far.

There is about $55B of capital looking for deals across all sectors today. Within Fintech, however, we see around $2B in capital across 12 different vehicles.

Many of the vehicles are in the $100-300MM range, and Cantor is underwriting several. It’s a pleasant surprise to see Adam Nash, former Wealthfront CEO, in the mix as well, and we recognize Kraken as a savvy capital markets player that has been very active in the digital asset treasury market last year.

One other notable feature is the warrant size — the smaller the warrant, the cheaper the capital for the company going public.

Below are a few deals that have been announced as targets.

We don’t have full view of the economics on these, but recognize the names.

• Securitize is a tokenization platform with a strong partnership with BlackRock. We have talked to the CEO in the past here. They have packaged several billion of RWA money market fund assets onchain together.

• Abra is an early crypto wallet / neobank provider. While they were a material player in the <2021 regime, we have heard less about the company recently.

• Animoca is a Web3 gaming company with a broad portfolio of assets, having leaned into the metaverse and onchain activity. Our interview with the CEO is here.

This gives a flavor of the company profiles that are now evaluating public markets — market caps in the $1-5B range, revenues likely between $50MM and $250MM, and a history of execution. Not every company is a straight exponential line up and to the right, like Anthropic, but it also makes for a healthier capital market to have a variety of exposures and risks in the market. The alternative right now is to be a buyer of a mega-tech late-stage trillion-dollar AI company.

Of course, SPACs are not for everyone. Many good companies take the straight path to IPO. Others can choose direct listings, token launches, SPVs, or other liquidity options.

SPACs *are* a good fit for companies that want to (1) be public, (2) have a strong counterparty with a clear view on the market, and (3) are dealing with more complex cap table and founder dynamics than usual.

If you think that your company fits the bill, or have an introduction to another potential target, give me a shout by replying to this email or sending a DM!

Excited to hear from you.

🚀 Postscript

• Sponsor the Fintech Blueprint and reach over 200,000 professionals.
  👉 Reach out here.

• Check out our AI newsletter, the Future Blueprint, 👉 here.

• Read our Disclaimer here — this newsletter does not provide investment advice

• Contributors: Lex, Laurence, Matt, Farhad, Daniel, Michiel

Today, we are excited to highlight AppsFlyer’s State of Finance for Marketers: Europe 2026 report.

It tracked 2.4 billion installs and roughly a billion dollars of spend to figure out where Europe’s finance marketing budget is actually going.

There are some pretty fascinating metrics in here, so let’s dig in. Or, if you want to just see the numbers, click below:

Read Report

Neobanks win the download. Banks win the customer.

For most of the last decade, consumer fintech in Europe was a race to acquire, and the neobanks have been really good at this.

And to a large extent, neobanks still win. In France, they pull close to two new users for every one a traditional bank gets, and Revolut now sits at number one in Spain, France, and Italy all at once.

But follow those users past the install, and the picture flips. Traditional banks keep 1.5x to 2x as many users around at day 30, which means the brands are losing the download war, but are getting the money.

Half of finance app installs aren’t even real!

Of course, holding onto customers only matters if they were ever real to begin with, and a lot of them aren’t. In Western Europe, close to one in every two investment-app installs gets flagged as fraud.

This was a pretty big surprise to us as well.

But it makes sense economically: the higher the payout per install, the higher the fraud incentive, and digital wallets and banking apps suffer proportionally lower rates of install fraud throughout the region.

Review Report

The front door is moving to AI.

Even the way people find these apps is changing.

More and more, they're starting their money questions inside ChatGPT and Perplexity instead of on a bank's homepage. Revolut, Starling, and Lloyds are already scrambling to put their own AI assistants in the app before a general chatbot gets there first.

The handoff is real money, too: web-to-app already drives 41.8% of owned-media conversions in Western Europe. A click that lands straight in the app converts about 2.8x better than one stuck on mobile web.

Europe isn't one market.

The two European markets do not behave the same way.

• Western Europe is mature and flat, so growth there comes from moving people who already visit the website into the app.

• Eastern Europe is still in land-grab mode, where Android banking installs have nearly tripled in two years and people sign up straight from a text.

Germany is its own animal, heavy on retail investing and, by the report's math, about 18 months out from a neobank majority. GTM leaders have to think locally and not optimize for a single profile across the continent.

This is just a preview of the data in the full report.

Check it out to see where traditional banks are out-retaining the neobanks, how install fraud a real problem, and where AI search is changing the way customers find their next app.

Additional key topics include:

• Retention and engagement benchmarks across banking, BNPL, crypto, and investing

• Install fraud rates by platform, channel, and country

• Web-to-app and deep-linking conversion lift, market by market

  Learn More

🚀 Postscript

• Sponsor the Fintech Blueprint and reach over 200,000 professionals.
  👉 Reach out here.

• Check out our AI newsletter, the Future Blueprint, 👉 here.

• Read our Disclaimer here — this newsletter does not provide investment advice

• Contributors: Lex, Laurence, Matt, Farhad, Daniel, Michiel

Subscribe now

In this episode, Lex chats with Cactus Raazi — CEO Americas at B2C2, one of the original and largest institutional market makers in digital assets, serving roughly 1,500 institutions and pricing across more than 40 exchanges globally.

They discuss what a market maker actually does, how balance sheet and signal generation underpin roughly $1 billion a day of stablecoin flow at B2C2, and why the two extremes of crypto market making - riskless principal aggregation versus proprietary alpha - produce very different client outcomes that buyers rarely understand.

Cactus explains B2C2's 18-month bet that the Circle-versus-Tether debate would give way to a multi-issuer world, the launch of its PENNY product for instant zero-cost cross-stablecoin swaps, and they explore why programmability is the next frontier for digital dollars, why US capital markets have almost no structure for funding genuine risk-taking businesses, and whether the current combination of scale, speed, and complexity makes this the hardest investing environment Wall Street has ever faced.

For those that want to subscribe to the podcast in your app of choice, you can now find us at Apple, Spotify, or on RSS. Or to support this writing and access our full archive of IPO primers, financial analyses, and guides to building in the Fintech & DeFi industries, see subscription options here. Our current price point is only $2/week.

Thanks for your time and attention,

Matt & Lex 🙌

Key notable takeaways:

1. Market makers aren’t a homogeneous category, and clients pay for the difference. At one extreme, a market maker is essentially a riskless agent - aggregating prices across 40+ exchanges and quoting on top with no real view. At the other extreme, a market maker is a proprietary quant shop running alpha signals on horizons from seconds to days, and the price you get is heavily conditioned by where the signal says the asset is going. B2C2 sits in the middle, partly because its public-company parent (SBI) constrains risk appetite. The implication for institutional buyers: who you trade with structurally determines the quality of execution, not just the spread.

2. Algorithmic fixed income market making didn’t fail on technology, it failed on capital structure. US capital markets are excellent at funding venture, growth equity, private equity, and buyouts, but there is almost no domestic pool of “risk equity” - capital comfortable with the possibility that the machines (or the humans) lose money on a given day. Market makers need exactly that kind of balance sheet, and the mismatch between what the business requires and what the US capital base offers is a structural reason firms like Elefant struggled, regardless of execution quality.

3. The Circle-vs-Tether framing is already obsolete; the next product wedge is interoperability. B2C2 made an 18-month-old contrarian bet that the duopoly narrative was wrong and that Stripe (via Bridge), Western Union, Revolut, and many other consumer and platform companies would issue their own stablecoins. PENNY - instant, zero-cost, zero-counterparty-risk stablecoin-to-stablecoin swaps - is the product expression of that view. The deeper claim is that stablecoins are software, and the SaaS analogy (a base layer plus an app store of programmable financial logic) is the real reason institutional adoption accelerates from here, not the transfer-of-value benefit on its own.

Background

Cactus Raazi spent close to three decades on Wall Street before taking on the Americas CEO role at B2C2 in 2024. He started his career at Goldman Sachs in 1998 in money market origination - the desk that oversees commercial paper issuance - before moving into credit sales in late 2000, where he covered hedge funds across convertible arbitrage, credit derivatives, structured credit, and mortgage derivatives. He spent 13 years at the firm and holds the record for Goldman's highest-ever annual credit sales production.

After Goldman, he held managing director roles at Nomura (2011–2012) and Tradeweb (2013–2015), where he designed OTC trading platforms. In 2015 he co-founded Elefant, an algorithmic market-making platform for corporate bonds, which he ran for over six years before its acquisition by Exos, where he then served as a partner.

Cactus subsequently led Amber Group's US business as CEO and co-Head of Americas, followed by US strategy work at Enhanced Digital Group (EDG). He holds a BA from UC Santa Barbara and a Master's from NYU Stern.

👑Related coverage👑

Topics:

B2C2, Goldman Sachs, SBI Group, Binance, Coinbase, Circle, Tether, Stripe, Kraken, Credit Suisse, Market making, institutional liquidity, stablecoins, fixed income, risk management, algorithmic trading, crypto exchange infrastructure

Timestamps

• 1’17: Rejected by 30 firms: A cold-call advertising inquiry that became a Goldman career

• 6’43: "I'll go sell jet engines": Complexity as the through line from credit derivatives to crypto

• 8’53: Scale, speed, and dimensionality: The hardest investing environment in 28 years

• 12’33: A terrific idea, a brutal execution: Building an automated market maker in 2015

• 16’22: The used car dealership of bonds: How over-the-counter fixed income actually works

• 19’51: Price, time frame, and the art of liquidity: What a market maker actually does

• 24’52: Riskless principal or proprietary alpha: The two extremes of crypto market making

• 29’54: Priming the liquidity pump: Why new tokens hire market makers and large ones don't

• 35’40: $1 billion a day in stablecoins: A contrarian bet against the Circle-versus-Tether frame

• 39’43: 24/7 money movement: The treasurer wish list stablecoins actually deliver

• 41’04: The channels used to connect with Cactus & learn more about B2C2

Illustrated Transcript

Lex Sokolin:
Hi everybody, and welcome to today’s podcast. I’m absolutely thrilled to have you with us today. Cactus Raazi, who is the CEO Americas at B2C2 which is one of the original and largest institutional market makers in digital assets and crypto. So, I’m really excited to open up what a market maker does and how it works. But cactus also has an amazing career in financial services, from leading roles at Goldman Sachs during some really key times to building out his own technology platforms and trading firms. So, we’re going to learn a lot. Cactus. Welcome to the podcast.

Cactus Raazi:
Thanks, Lex. I really appreciate you having me on. I’m looking forward to our conversation.

Lex Sokolin:
It’s my pleasure. I’ve been meaning to have you on for a long time. Let’s start with big finance. You spent quite a bit of time in Goldman. How did you end up there and how did you get your start?

Cactus Raazi:
That’s probably the most interesting part of my life journey. I started out in a kind of a backwater role at Goldman Sachs, and I got the job largely through a great fortune at the time when I was prior to joining Goldman. I was living in Los Angeles. I was reading The Street, which was an early finance website, and dreaming of being on Wall Street. I had no idea what that actually meant. And out of the blue. One of my still best friends, a gentleman named Nathaniel Klipper, called up the magazine that I was working at and asked for some advertising information, and he asked for it to be sent to his office at Goldman Sachs. And I took that as a sign from heaven and followed up with this gentleman.

The actual advertising in the publication I worked at never happened, but he agreed to meet me on my next business trip to New York, and that kicked off a friendship where I started to learn more about finance through my conversations with him, and through reading a ton of books that he recommended. I moved to New York eventually in 1998 without a job, and just started interviewing. And after getting rejected by probably close to 30 firms, I got an interview at Goldman. It took quite a time to even just get the interview, but I was pretty well prepared for that interview at that point. Having screwed up so many prior interviews and I got my first job in 98. In money market origination is kind of the desk that oversees the issuance of commercial paper programs, which is not exactly the sexiest place to start. But it was my foot in the door and things grew from there.

Lex Sokolin:
Which markets did you get exposure to sequentially? It sounds like a lot of it was fixed income, but you also travelled kind of through the different parts of the firm.

Cactus Raazi:
It’s true. I started in fixed income and the bulk of my experience has been in fixed income. But there’s an important asterisk there. My client base starting around 2000. I moved into a sales role in late 2000, and my client base was primarily hedge funds at the time. These were sort of dormant hedge funds that Goldman Sachs gave me as, as sort of an initial client list. And several of these hedge funds started to look into derivative products. They were convertible ARB funds, and they were interested in interest rate swaps to hedge some of their interest rate exposure. So that gave me a lot of exposure to derivative products in general. And then a little later in the early 2000, with the growth of and sort of awareness of credit derivatives, that was another big stepping stone in my career. A fairly complex product that a lot of the other salespeople didn’t really spend the time and energy to learn and master. And so, it gave me a terrific opportunity to go out and prospect a bunch of new clients.

Again, most of the hedge funds and that large that client base of large hedge funds then morphed into a bunch of additional products, structured credit products, mortgage derivatives, and then in some cases, some different products that touched on equity markets as well. The hedge fund client base in general is pretty rangy, and for the most part, they’re open to a variety of products and strategies to accomplish their mandate. Unlike, you know, let’s say, a more traditional asset manager who’s generally pretty range bound. So the open mindedness and sort of capabilities of my client base allowed me to get exposure to a variety of what you might call. Back then, we didn’t call them, but these days we call them frontier markets, I guess.

Lex Sokolin:
How do you sell to a hedge fund? What is it that they’re buying?

Cactus Raazi:
They are looking for the most capital efficient and risk efficient way to express a view on a specific outcome. So the point I’m making there is if you thought, for example, company X was going to was going to do very well over the medium term, you obviously could buy the stock of that company.

Obviously, you could buy the bonds of that company. You may actually want to look at buying call options, for example, because you’re going to get a much better payout to the GP that you’re right. And this idea of picking up asymmetrical exposure in a capital efficient manner is really at the heart of a lot. Not all, but a lot of hedge fund activity in general. How do I express this view, whether it’s bullish or bearish in the most sort of capital efficient way, with the greatest asymmetry to the outcome.

Lex Sokolin:
Were there different pockets of capital that you encountered? What is the sales cycle? Like, I’m really curious. You started fresh in the industry and then you built this bench of relationships. Selling early on pretty complex products, you know, very mathematical sort of structures. And also, you did it during a time where there was also a lot of market volatility. Right. Like in the early 2000, you had the internet boom and bust. In 2008, you had the great financial crisis and the mortgage bust. Within that, you had sort of like the hedge fund boom that I’m sure that you were exposed to, and all sorts of other shifting ground, which reminds me of the situation we’re in today. So how did you find, like, a commercial path through that?

Cactus Raazi:
I’ve asked myself this question a lot because I did quite well in retrospect and my sort of production numbers, and I think impact on the business were notable. And so, I’ve asked myself what was the common thread? And the answer is really I gravitated towards products of greater complexity and they were just more intellectually interesting to me. I might add that in my interview process at Goldman, one of the toughest interviews, I had a real legend, a Wall Street legend. The gentleman named Phil Off asked me a question towards the end of the interview after I think I had, you know, barely passed, probably. And he said, if you don’t get this job at Goldman Sachs, what are you going to do? It was an interesting question, frankly, and I answered at the time, well, I’m pretty good at sales, and I’ll probably go sell the most complex product I could think of, at which I at the time, I thought was jet engines.

And frankly, I still don’t fully understand jet engines sufficiently well. But having said that, that’s the example that I cited. I said, you know, maybe I’ll go sell jet engines or something. And I think once I got on, once I got into finance and got my job on Wall Street, I have continued to move towards new products, more, more complexity. It’s more intellectually interesting and that that has continued not only through my time working at bulge bracket banks, but also in moving into electronic trading, setting up my own market, making firm, and then more recently, of course, being in digital assets, it seems to be a through line over the course of my entire career.

Lex Sokolin:
Can you talk about what you saw in terms of the different market conditions in the businesses that you were in, and kind of any lessons about how to behave in different market cycles, how to respond to them, how to think about them? You know, there are people who say it doesn’t matter what the market’s like. We just go do our job and other people are very reactive. It’s like risk on, risk off trying to find kind of the right weather. What’s your view on why cycles happen and how you would navigate them?

Cactus Raazi:
It’s a great question. I might add, by the way, that when I started in 98, it was just a couple of months after I started that the long-term capital kind of implosion took place that, you might remember, was the world’s most famous hedge fund, staffed with the world’s brightest people who managed to blow up in spectacular fashion. If I’m not mistaken, the Fed actually convened a bunch of the CEOs of banks to deal with the mess and the fallout of long-term capital. And since then, you’ve cited a few different up and down events. And I think even more recently, we had the sort of Archegos, you know, prime brokerage debacle which effectively took down Credit Suisse as an entire institution. And this seems to be a recurring theme. And Wall Street tends to have a hard time learning its lessons.

But more specifically to your question. I’m more of the latter camp where you need to be aware of what the overall market conditions are and understand what the opportunity is afforded to you and many terrific asset managers of all stripes will adjust their exposures based on whether they view the opportunity as favorable or whether they view the market conditions is quite challenging in terms of whatever the mandate may be extracting a return or extracting alpha relative to a specific benchmark or even on an absolute basis. And you’ll see the things such as cash positions fluctuate or people stick on the sidelines when the opportunity is simply not there. And typically, the best investors have that approach. I think that the questions are really interesting, one, because you’ve cited a few crises, and then there are obviously have also been a number of opportunities over the course of the past 28 years that I’ve been on Wall Street. I will say that at this, it’s quite timely to be asking that question, because I have never experienced and I’ve sort of thought about this a fair amount.

I don’t pretend to be a historian of financial markets, but having thought about it, I’ve never considered a market opportunity that that even comes close to what we’re currently living through, as you and I are talking. The scale of the forward opportunity with the frontier technologies that your media company kind of focuses on, the scale of these opportunities is unbelievable. The complexity of trying to think about second order and third order impacts of these technologies is, is really mind bending. And then finally, the speed at which this stuff’s changing is also really remarkable. If you think about any seasoned investors, you know, anyone from a Soros to a Druckenmiller or to many, many other famous names, you know, whether it’s Tepper and Distressed Debt or maybe Steve Cohen or whoever you want to cite as you’re investing hero. I’m sure if they were on your podcast, they would probably agree that the scale of the forward opportunity, the speed at which it’s changing and the complexity or the high dimensionality of thinking about the forward path makes for a unique investing challenge on a going forward basis.

Lex Sokolin:
The opportunity is larger than ever, but then the amount of change is faster than ever, and I think is probably one of the most complicated periods to try and figure things out. So, on Wall Street, you were doing a lot of kind of hand-to-hand combat and high-level institutional sales, relationship building, and then your career took you towards more software first and digital and programmatic ways of touching financial services. Can you talk about what happened next, and how did the things that you used to do start to get transformed?

Cactus Raazi:
I was strongly of the opinion, along with a group of other professionals. I was strongly of the opinion that the process of buying and selling bonds in small size, not, not huge trades that will make your year, but rather just a bunch of odd lot stuff. As you probably know, most asset managers go through a monthly process of cleansing their portfolios of small amounts of positions that they want to reduce, adding small amounts of positions that they want to increase. And these can be as a result of slight changes to their allocation model or people asking for their money withdrawals, redemptions of capital commitments or new capital coming into the funds.

And so, our thinking was, hey, the pricing process for fixed income can be replicated by algorithms, and we should be able to set up a fully algorithmic market maker that provides a service in a superior way. As you may or may not know, most human traders who still price the vast, vast majority of larger fixed income trades. Can’t be bothered with these tiny little, you know, offer a million here or a bit a million there. These types of small nuisance traits are not what most traders like to do anyway, human beings that price larger blocks of risk. And so, we thought, let’s set up an algorithmic market maker. Let’s. It needs to be a broker dealer. So, there was that entire process. And let’s then go out there and start buying and selling bonds on an automated basis. It was a terrific idea. Yeah, it probably remains a terrific idea, but the execution of that idea was significantly more complex than perhaps we had realized. And I think that the technology stack that was available to us back in 2015, 16, 17, 18 was also not what it is today, rather, obviously, which made for a difficult time.

The biggest challenge we had, though, was not necessarily on market structure nor on technology, but rather on the availability of capital. One of the more significant oversights on my part, and on the part of the team that came together to build this company, was the fact that market making firms, regardless of how they come up with their prices, are really capital-intensive organizations. There are huge amounts of capital in the US capital markets for early-stage venture, for growth equity, private equity, you name it. There is very little capital for risk equity, if you will. Or this notion that if your machines do something wrong, or if your human beings do something wrong, you’re going to lose all your money. That’s just not consistent with what a venture investor or a growth equity investor would be looking for. That’s much more consistent with what a hedge fund investor is looking for. And so there was kind of effectively, you would say poor product market fit from the perspective of the capital base required for a market maker is not consistent with the type of capital. That’s why we considered being available in the world’s dominant capital market in the United States.

Lex Sokolin:
There’s a bit of an odd question, but maybe from a definition’s perspective, it would help. Can you highlight the difference between maybe a custodian or a depository of financial assets versus an exchange or a marketplace or a market platform, and then finally a broker or an agent. Because I think for a lot of listeners that all smush us together, you know, but if you walk through it in the fixed income example, that would be helpful, right? Like where’s the fixed income instrument live? And then what does it mean to try and build a venue. Like what does the word venue mean? What does the word exchange mean. And then what’s the role of the salesperson or the broker. And how is that different.

Cactus Raazi:
Sure. Yeah. I mean, frankly, it’d be very understandable why there’d be confusion around fixed income market structure. It’s so opaque. I’m not even sure Claude or ChatGPT fully understands it.

But yes, there is. You know, custodians are where you would sort of keep your assets. This could be true of stocks as well. There’s complexity of micro complexity and market structure. But in general, you have companies that hold on that provide safekeeping for your assets. Generally, custodians could also be in some senses, a prime broker or whatnot, but in general, it’s where you keep your assets and then you have venues on which you can transact in. The purpose of most of these venues is price determination. So, you have exchanges. The purpose of exchanges is obviously to determine price. Buyers and sellers come together on the same instrument, and they form a stack of prices at which they’re willing to buy, at which they’re willing to sell. And the intersection of those prices creates trades. Of course, even in over-the-counter markets, which is how fixed income is primarily set up, you as a as an investor would have multiple service providers, typically brokers or broker dealers, where you would be able to ask for prices on the fixed income instrument that you’re looking to either buy or sell.

And this is a portion of that marketplace is satisfied by market makers. By the way, this is also true in equities and other asset classes such as commodities or FX, where a market makers job is to provide liquidity to either an exchange or some other execution venue, or potentially directly to the end customer. Again, there’s some microstructure there. Some market makers don’t actually serve customers directly for regulatory reasons. They’re effectively just proprietary traders on exchanges. But ultimately, the experience from an investor’s perspective is an abundance of liquidity. First, safekeeping of your assets when you’re not trading them, that’s at custodians the worlds. To the best of my knowledge, the world’s largest custodial bank is Bank of New York, with roughly $36 trillion of assets under custody. But, you know, there’s many smaller operations out there. And then once you’re ready to transact, you’ve got your various execution possibilities. Electronic execution in a bilateral sense on platforms such as trade, web or market access, as well as over-the-counter transactions with large banks, smaller institutions, even pure agency broker dealers.

Lots of different ways that you can actually buy or sell. And ultimately, though, you are going to need a balance sheet on the other side. That is the truth in not necessarily in exchange-based markets where you may have an intersection of natural buyer and a natural seller. But in over-the-counter markets, it’s typically the case that the person who is purchasing something that you are selling does so as a business, and they will keep it on their balance sheet and look to sell it later. The corollary I like to use that almost everyone can understand is a used car dealership where you can show up to a used car dealership and buy anything on their lot, and you can show up with your car and they will show you a price for that car. And if you don’t like the pricing, go to a different used car dealership. And that’s more or less the market structure of a fixed income.

Lex Sokolin:
Let’s use that as a jumping off point to talk about B2C2, where you’re running the America’s business. What is B2C2. And then Maybe let’s use that as a path into the market structure of digital assets.

Cactus Raazi:
So B2C2 is a global market maker and liquidity provider. And just a little bit more on that. There are there are firms out there in not only in digital assets and crypto, but also in other asset classes whose business it is to just provide prices to exchanges. The largest exchange in digital assets is Binance. So we can use that as one example. But this would be true of many other exchanges such as Coinbase Exchange or Kraken Exchange. The idea is that firms like ours, as well as other firms, provide prices at which customers can buy or sell crypto and digital assets. On these various exchanges, B2C2 and a small number of competitors also have a large over-the-counter business. We serve end clients directly. We have a global franchise of about 1500 institutions who have onboarded with BTC to and can trade with us using our balance sheet as the repository of the assets that they’re looking to either buy or sell.

Lex Sokolin:
In these two contexts, like do the words exchange and liquidity? Do they mean the same thing as in Wall Street?

Cactus Raazi:
Yes, they really do. From the perspective of either an institutional investor and B2C2, by the way, is 100% institutional and we don’t have any individuals. We don’t serve any individuals directly. We typically serve platforms that then serve individuals. But having said that, for the most part, from the perspective of your listener, the way that an exchange works in digital assets is the same as the way that an exchange would work in traditional securities or other asset classes.

Lex Sokolin:
Then let’s walk through what a market maker does. And you mentioned words like balance sheet and liquidity. Can you unpack for us what it means to provide liquidity, what it means to use your balance sheet? What does the world look like with or without market makers supporting particular assets?

Cactus Raazi:
Yeah, well, market makers primary role is to give the marketplace prices at which the other side, so to speak, can either buy or sell some underlying instrument.

We can use Bitcoin as the example here. Simple and obvious example. Our job is to put prices onto exchanges in the first instance, where the anyone else on the exchange can either buy or sell bitcoin against some other asset. Typically, it’ll be against USDC, USDT. That would be the circle stablecoin or a tether stablecoin. We can also show prices against many other assets. And that would be the idea that you can sell USDC and buy Bitcoin, or you can sell Bitcoin and buy USDC. And we put these prices on over 440 exchanges globally. And then we also have our large over-the-counter business directly with end customers as we’ve discussed. And then customer could be, for example, a digital asset hedge fund that manages a pool of digital assets on behalf of their investors and may have a decision to make around buying or selling an asset and is going to look for the best price. They’re going to look to what their execution may be on Coinbase or Binance, or any other of the 40 exchanges that were connected to.

Or they may compare that price to contacting us directly, but we say we call it a voice trade. But these days it’s mostly through some sort of an electronic messaging platform and asking for a price. For example, of a client, we’re looking to buy a hundred Bitcoin. They would be able to guesstimate how quickly and at what at what blended price they’d be able to do that on one or more exchanges and compare that to the to the price we would give them on 100 Bitcoin and that would be our offer to them. So, this idea of what a market maker does, is it, you know, we use this term liquidity. But liquidity obviously translated is just the ability to buy or sell something at a price that you’re happy with and in a time frame that you’re happy with. Price and time frame will change, and the more focused you are on getting the best possible price, typically the longer the time frame and therefore uncertainty you’re going to have. And that tends to be a natural trade off in all markets.

And what market makers do is provide prices to either end customers or to exchanges in order to facilitate transactions. That’s really the heart of it. It’s a pretty simple business from a perspective of the what, but the operational realities between managing the risk of your balance sheet, trying to avoid negative selection. That is to say, every time we buy something, its price goes down. Every time we sell something, its price goes up. Trying to avoid those types of outcomes is the art of being a market maker.

Lex Sokolin:
That’s exactly what I was going to go next, which is the risk engine and how do you hedge out all of the exposures and keep yourself net neutral every day. Like what does that infrastructure look like for the firm to be confident that it can provide liquidity in all these different places. Like, it sounds like there’d be a lot of math.

Cactus Raazi:
There is a fair amount of math. There is also a high level of variance between market makers in terms of what their core business model is. I’ll give you kind of the two extremes.

One end of the spectrum would be a firm who really represents to their clients and to their customers the pricing that’s available to them. And what that means is you’re not really expressing a view on the marketplace. What you’re saying is, for example, I see five Bitcoin on offer at Binance and five at Coinbase, and I see three at Kraken and maybe another three on another platform. So, I’m observing that there’s 16 Bitcoin on offer. And I’m observing what that the price of each is. And so, I now have a price at which I’m able to offer 16 Bitcoin to some end customer. I would just make a small amount of money there because I’m essentially an agent between the exchanges to which I’m connected and my end customer. And that’s a very, very low risk, sort of almost as you would call that a riskless principal type of transaction. And so that’s at one end of the spectrum for market makers is really low risk, no point of view on the market, not necessarily a lot of infrastructure in place to determine the right price, but the ability to look at available pricing and aggregate that together for your end customer.

And that’s the service there. And that type of service allows your customer to avoid going to a bunch of different exchanges dealing with exchange deposits and some of the other operational questions there. In some cases, there may be some compliance considerations. And so there you go. That’s one end is very low risk riskless principal type trading. The other end of the spectrum would be a much greater risk appetite. A strong point of view around where anyone asset is going to go, whether it’s going to go up or down in the medium term, typically near the medium term, maybe a signal generation framework and the signal generation framework attempt to use quantitative techniques to generate a bunch of alpha signals, that is to say, a point of view in some examples. For example, in Bitcoin, you may have a medium-term alpha signal that predicts whether Bitcoin is going to go up or down in the next five minutes. And a very high frequency or a very short-term type of alpha signal might actually be whether Bitcoin is going to go up or down in the next couple of seconds or even shorter.

But putting all of this together, a firm that trades with a heavy proprietary bias has invested quite a bit of time and money and manpower or smarts and analytical capabilities into having a point of view on any individual asset and the correlation between these assets and their pricing, and therefore, what is the quote unquote, right price for any given inquiry is much less a function of what’s available on exchanges and much more a function of what the machines tell them is going to happen with the assets price over the, over the next very short to medium term. These types of firms also exist. They just have a reasonably different business model. It tends to be a much more proprietary trading business model and therefore the client experience there is, I say, probably less consistent. You know, in some cases you would ask for an offer on Bitcoin. And if the signal suggests Bitcoin’s going up, you’re going to get it. You’re going to get a terrible price. If this their signals. If the if the market maker signal suggests Bitcoin is going to go down, then you would potentially get a much better price.

And these are the types of differences between the firms. BTC is kind of somewhere in the middle. We have a we have a fairly well built out alpha framework, but we overall have a pretty constrained risk appetite. We’re owned by a public company, a Japanese public company called SBI, and that translates through to a more moderate risk appetite than some firms that are owned by either a founder or a founder and a group of partners and have probably more leeway to take a more proprietary point of view on the marketplace.

Lex Sokolin:
One last question about the role of market makers. And, you know, for those people who are really close to the crypto markets, you know, there’s always so much feeling about the role of market makers and performance of tokens and performance of tokens on different exchanges and so on. Whether it’s sort of this, people say, if you’re launching a token, you need to hire a market maker in order to support the price in the beginning. Or if you looked at the Binance crash in October, whether it was, you know, Wintermute or GSR or somebody else, the sort of collapse of one of the large market makers has led to effectively beta level extinction across all the markets and liquidity drying up and price levels going down.

I think in crypto, there’s sort of this assumption between, you know, it’s not just taking orders. It’s something to do with price support in traditional sort of equity markets. You definitely would not have that sort of conversation. What do you think is the disconnect? What is it that you think people are doing in crypto and why?

Cactus Raazi:
I think that the role of a market in the context of your question, what the market maker is really trying to do is demonstrate to the marketplace that there is sufficient liquidity in whatever token we’re discussing. It’s typically a newly issued token or one that’s still early in its lifespan, so to speak. And that’s typically because you’re going to get a relatively low levels of familiarity on the part of the investor base, and therefore you’re going to get relatively low levels of transactional interest. And what it tends to be the case that the perception that there is sufficient liquidity, that is to say that an investor is able to buy or sell with relatively low cost, and then that perception tends to create a virtuous circle where the perception that there is liquidity brings more investors to either consider the asset or potentially make an initial allocation, some sort of an initial purchase, which then feeds more liquidity and tends to get the kind of virtuous circle going.

Obviously, we don’t need market makers for the large liquid tokens. Anything most of your listeners have heard of don’t really require the support of a market maker, but that’s less true for a newly issued token or one that perhaps has been issued but is still languishing. And I don’t mean languishing as in, you know, its price should be higher, I mean languishing in terms of visibility, engagement on the part of a broad set of investors, familiarity, those types of metrics. So that’s where a market maker comes in, I think can be really helpful to sort of priming the liquidity pump, facilitating more transactional activity off of their own balance sheet and over time, leading to a larger investor base and greater levels of activity without the need for a market maker.

Lex Sokolin:
Support a different function right in early stage, and I think the lines kind of get blended a bit.

Cactus Raazi:
Yeah, I mean, you see the same thing in equities. Obviously in the context of an IPO, you have the underwriters and their job is to provide some level of support and stability for the newly issued stock for a moderate period of time.

And if there’s, you know, extreme levels of interest, they typically have some stock in their back pocket typically. So, I usually call the green shoe to which they can then feed some more stock into the marketplace. But the overall idea even in securities markets, is to ensure that, you know, the issued security comes to market in a in a way that’s sort of stable and in the best interests of the issuer over the medium to long term, you know, creating some goodwill on the part of institutional investors, creating the perception that the security is is well stabilized and sort of its price action is reflective of the sum total of the market’s point of view on the security. A lot of that translates over directly to crypto. I would say, you know, the world of crypto is a mixture of what you and I would probably call sound enterprises who are coming to digital asset markets looking for financing and have a real business and have a real use case and are exploring product market fit and growth. And then at the other end of the spectrum, crypto, for better or for worse, also includes a bunch of, you know, jokey, extremely speculative assets whose fundamental value is questionable.

And it is sort of it’s both a feature and a bug, I suppose. And needs to be acknowledged that that last parts less true. It’s pretty. Well, yeah, I was just going to say it’s less true of securities markets. But if you look at some of the some of the SPACs, for example, that have come to market over the past 5 or 6 years, and there has been a fair number of those that have turned out to not be a particularly hot investment either. So perhaps there is a corollary to securities markets, and maybe crypto is getting an unnecessarily harsh rap. But nevertheless, the point here is that market makers in both instances are service providers. You know, in security space would be investment banks, but their job is to ensure that a new a new investment that comes to market with a degree of stability and then and, and establishes good rapport with its investor base as, as that security or that digital asset takes on a life of its own.

Lex Sokolin:
Yeah, I guess the difference is like there’s the primary markets and there’s the secondary markets, and those are two different business lines.

Effectively, they both use balance sheet but in different ways. One thing I wanted to talk to you about is stablecoins and sort of this broad adoption of stablecoins that’s now happening within fintech, you know, even like five years ago. One thing that just stood out at me is how we’ve managed to conflate two different versions of cash. And it’s funny that, you know, you started out at Goldman with commercial paper, which I think brings it nicely into a loop here. There are different things that the dollar does. Some dollars are used to buy sandwiches in a shop. Some dollars are used as deposits in a bank to earn interest. Some dollars are used as a cash sweep account inside of brokerage. You know, that may go towards funding that commercial paper or money market funds and so forth. From a software perspective, stablecoins are just tokens that are collateralized by different versions of these dollars. We only have like one word for them and we’ve got one law and it sort of smashes everything together. I know that B2C2, has been working on initiatives around stablecoins, and maybe a way to get into that is to talk about the differences that I’ve touched on what you think of them. And then what is it that you see happening on the capital market side?

Cactus Raazi:
Well, the fact that you were thinking about stablecoins five years ago puts you in rare company because, you know, most of us could not have cared less five years ago. But at B2C2, we transact quite a bit in stablecoins, about $1 billion a day. It’s a function of our heavy transactional activity in digital assets and crypto and stablecoins being sort of the native funding instrument. And it’s a little bit of a legacy also from some of the challenges that digital asset companies have had in accessing their traditional banking. And so, we’ve, you know, digital asset investors have had to move to stablecoins for a variety of reasons, some of which have to do with banks. Until recently, the lack of appetite to touch crypto and digital assets amongst traditional financial institutions. So, for us with stablecoins, we made a kind of a big bet a little bit more than a year ago, about maybe 18 months ago, that there were going to be many more stablecoins in the future.

This was at a time when the big discussion in digital asset markets was whether circle was going to be tether, or whether tether was going to beat circle, and we always thought that was a rather naive conversation to have. And we expected that many more institutions were going to issue their own stablecoins for a variety of reasons. That was our first bet. And the second bet was that the programmability of this, this form of money. We’ll talk about that in a second, but that the programmability was going to be increasingly important over the next 3 to 5 years. Specifically, when you think about money and you mentioned that there’s different forms of money. We have store of value. That’s a very common sort of perspective on what money is a store of value, although the, you know, a lot of monetarist people or the kind of like debasement people will argue against that. But if we can set aside the philosophical component, you know, there’s a store of value and then there’s a transfer of value.

And the transfer of value component is where stablecoins really shine. Most people would agree at this juncture that stablecoins are in fact a better mechanism to transfer value. What does better mean? Like all technologies, better means faster and better means cheaper. That is what better means. And that’s at the heart of almost every piece of new technology and why it succeeds. And with stablecoins, we have this idea that you can move, in most cases a dollar. But, you know, we’re seeing more stablecoins in other currencies, but nevertheless, you’re able to move a digital representation of a unit of currency that’s held in trust in a typically in a custodial environment, and that it allows you to move this currency much more quickly with much less cost than current systems such as ACH or Swift. I would probably also point out that in certain jurisdictions, the central bank or some sort of a central governing body has set up an instant payment system to some degree, obviating a portion of the stablecoin. Benefits. Examples could be the Peixe system in Brazil, where money is able to.

Move very quickly and basically for free. 24 over seven. So, there are there are certain jurisdictions where they’ve leapfrogged what we can do in the United States. But more generally speaking, the transfer of value, whether it’s in Western developed economies or developing economies, as has always been too expensive, and stablecoins really addressed that problem. The second thing, I think is forthcoming, so that even the stablecoin skeptic would have to agree that a piece of software, which is what stablecoins are, has the ability to then run additional software on it. You can think of it as not that different from some of the SaaS platforms where the SaaS platform is the software, but they open up an app store, and there’s all sorts of different applications that can run on top of stablecoins. That’s something we expect to see over the next couple of years.

Lex Sokolin:
So, it’s not about competing with Stripe, about who can pay people better for products on the internet. It’s going back to that point about using the balance sheet for liquidity across all sorts of different venues, isn’t that, right?

Cactus Raazi:
Yeah, I think that’s exactly right. I mean, it’s funny you mentioned Stripe. I was blown away by what Stripe announced nearly a year ago now at their annual Stripe sessions, which was, you know, the idea that they were going to create their own, effectively, their own stablecoin network amongst all of their merchants, obviating the need for a financial transaction to necessarily touch a bank balance sheet until you go from transfer of value to store of value. And of course, I don’t think there’s a company out there that is planning on self-custody of their cash. Most companies think it’s the right thing to do to keep your cash out of bank. This has everything to do with governance, and not necessarily so much to do with whether you receive interest or not. That’s a that’s sort of a little bit of a side note, but there’s probably not a CFO out there that feels comfortable keeping 100% of the company’s cash in some sort of a crypto self-custody wallet. Having said that, moving money around and being able to do that 24 hours a day, seven days a week, is very much at the top of the wish list of many CFOs and treasurers. And that’s where the stablecoin transfer of value discussion, I think, has the greatest potential.

Lex Sokolin:
Absolutely. If our audience wants to learn more about the stablecoin initiative or B2C2 or about you, where should they go?

Cactus Raazi:
Well, these days, I suppose you can go to Claude.

Lex Sokolin:
You are the first to defer to the robots.

Cactus Raazi:
Oh, is that right? No, I mean, the robots are so. Boy, they are so powerful. Here’s what I would say. B2C2 has an actual product that’s been receiving a huge amount of interest in the institutional context. And this product is called Penny, and it’s a stablecoin facility that allows any user to instantaneously, that is to say, instantly and with effectively with zero fees and zero cost swap from one stablecoin to another. Again, the stablecoin landscape is still to this very day dominated by Tether and Circle, with, I think, a strong third asset being the global dollar network USDC. However, we are seeing really exciting announcements not only from stripe, as you have already mentioned, in their acquisition of that terrific company bridge, but also from Western Union.

That’s public. I think Revolut has made it public. It’s quite likely that other consumer platforms, large retail platforms, are going to issue their own stablecoins, and then therefore there’s going to be the need to be able to go from one stablecoin to another instantly. No counterparty risk. And very quickly. So that’s our that’s our penny product. And it seems like the product market fit there is excellent. The other places to go for stablecoin information I found LinkedIn to be terrific. If you don’t want to deal with clod or you don’t trust it for whatever reason. A simple search for stablecoins on LinkedIn. There is the amount of information there. Formal research reports I’m not referring to, you know, random tweets like you get on some platforms. I’m really referring to proper formal research sanctioned by banks and consultancies. There’s a huge amount on LinkedIn. I don’t know if you agree with me, but there’s that. And then of course, your own podcast. I mean, you’ve been very early on the stablecoin topic, and I think a lot of what you’ve put out has been absolutely spot on.

Lex Sokolin:
Cactus, thank you kindly for joining me today.

Cactus Raazi:
Always a pleasure. Speak to you soon.

Postscript

• Sponsor the Fintech Blueprint and reach over 200,000 professionals. 👉 Reach out here.

• Read our Disclaimer here — this newsletter does not provide investment advice

• Contributors: Lex, Laurence, Matt, Farhad, Daniel, Michiel

• For access to all our premium content and archives, consider supporting us with a subscription.

Upgrade to Full Access

Today we are diving into the following topics:

• Summary: Coinbase launched an SEC-registered AI advisor that can analyze portfolios, suggest trades, and operate across a growing financial super-app that now includes equities, crypto lending, tokenized stocks, and derivatives. Mercury introduced Command, an AI banking assistant built on top of a business banking platform serving 300,000+ customers, generating $650M in annualized revenue and valued at $5.2B. We show that AI is commoditizing financial analysis itself, leaving proprietary data, distribution, and product infrastructure as the key competitive advantages.

• Topics: Coinbase, Mercury, Plaid, Betterment, Wealthfront, Robinhood, Revolut, Cleo, OpenAI, Anthropic

To support this writing and access our full archive of newsletters, analyses, and guides to building in the Fintech & DeFi industries, see subscription options below.

Subscribe now

🤖🏦🧭 Our Ecosystem:

Generative Ventures | AI Research | Robot Money | Linkedin & Twitter | Sponsors

• 🚀 Lex is actively meeting teams building machine-native financial companies. If that’s you, reach out here.

• 🤖 AI & Robotics Industry Encyclopedia: In this 100+ page report, we collate all the meaningful companies across public quities, private equity, and digital assets related to the machine economy. 👉Learn more here

• 🦄 In partnership with InterPrivate, we just launched a $200MM SPAC looking for targets in fintech, digital assets, and AI infrastructure. If you are a late-stage founder or have an idea to discuss, reply directly to this email.

Long Take

All your banks are AI banks now.

Coinbase just ran a “System Update” event and has just announced an AI agent as an SEC registered investment advisor.

Mercury announced Mercury Command, an AI agent embedded in banking flows that can function like a personal bookkeeper.

The AI RIA is a genuinely new legal idea, but is it more than just a chatbot with access to your balance sheet? And if they are not just chatbots, are they financial AI labs worthy of a $3 trillion valuation? Or is the SEC-registered AI agent just another pass at a roboadvisor, a category that failed to deliver disruption?

Let’s pull this apart.

Coinbase Advisor

Brian Armstrong, the CEO, announced the AI advisor product.

Did they register an AI with the SEC?

Per Coinbase’s own disclosures, the thing is operated by Coinbase Advisors, LLC — a Registered Investment Adviser with the SEC and a Commodity Trading Advisor registered with the NFA.

This registration means that the software carries a fiduciary duty, even while warning the user that output “may be inaccurate or incomplete” and that registration “does not imply endorsement.” No mainstream AI product in finance has yet delivered this particular contradiction. In the past, regulators used to complain that roboadvisors, with their arithmetic mapping of risk to portfolios, were non-transparent. Or that machine learning created black boxes for underwriting.

This is the same issue on steroids.

The agent reads your full portfolio and account history, then surfaces tax-loss harvesting moves, news-driven trade ideas across stocks, commodities, and crypto, and portfolio recommendations, through a natural-language chat. Armstrong’s framing was that you “speak to it in plain English to take action on your account” and that “it will even prompt you with ideas you hadn’t thought of.”

Like investing in derivatives tied to the SpaceX IPO, most likely.

This is rolling out first to Coinbase One subscribers in the US, which makes it a recurring-revenue feature as much as an advice engine. Coinbase posted a $394 million loss in Q1 and has been working to wean itself off transaction fees. Some wealth management fees would help balance the books.

Here is the rest of the update with which this AI will integrate:

Read more

Today, we highlight the below items.

Also notable are the drastic moves both in the US and UK to allow the trading of tokenized securities, to be covered in a later issue.

1. DIGITAL ASSETS: The Backend of On-Chain Credit

2. ANALYSIS: Why Silicon Valley Values Ramp at $44B and Bill at $3B

3. CURATED UPDATES: Financial Institutions and Adoption; DeFi and Digital Assets; Blockchain Protocols; NFTs, DAOs and the Metaverse

To support this writing and access our full archive of newsletters, analyses, and guides to building in Fintech & DeFi, subscribe below (if you haven’t yet).

Subscribe now

🤖🏦🧭 Our Ecosystem:

Sponsor FB | Generative Ventures | SPAC | AI Research | Robot DeFi | Linkedin & X

• 🚀 Lex is actively meeting teams building machine-native financial companies. If that’s you, reach out here.

• 🤖 AI & Robotics Industry Encyclopedia: In this 100+ page report, we collate all the meaningful companies across public quities, private equity, and digital assets related to the machine economy. 👉Learn more here

• 🦄 In partnership with InterPrivate, we just launched a $200MM SPAC looking for targets in fintech, digital assets, and AI infrastructure. If you are a late-stage founder or have an idea to discuss, reply directly to this email.

DIGITAL ASSETS: The Backend of On-Chain Credit

Morpho, the onchain lending protocol, raised $175 million on June 9, one of the largest funding rounds in DeFi history.

The round was co-led by Paradigm, a16z crypto, and Ribbit, with strategic participation from Apollo Funds, Circle Ventures, VanEck, and Ledger Cathay, at roughly a $2 billion valuation. The token price is floating at roughly $1.96B fully diluted value, with about half of that unlocked into the market.

It is Morpho’s fourth institutional raise since 2021, and lands with the protocol holding more than $11 billion in deposits and a client roster that includes Coinbase, Kraken, Binance, Société Générale, Apollo, and Anchorage.

Understanding Morpho

Morpho is the work of Paul Frambot, who started it in 2021 from Paris at the age of 20. The protocol grew out of academic research and raised a $1 million seed in September 2021, then $18 million from a16z and Variant in 2022. This $175 million round is the fourth institutional raise, bringing total funding well past $250 million.

That’s a very good Seed investment.

Mechanically, Morpho is two layers.

The base is Morpho Blue, a single immutable contract that hosts isolated lending markets. Anyone can deploy a market, and each one is fixed at creation by five parameters: the loan asset, the collateral asset, the liquidation loan-to-value ratio, the price oracle, and the interest rate model. A market for ETH collateral against USDC borrowing is in its own sealed compartment. If that compartment takes a loss, the loss stays inside it and cannot spread to any other market.

Borrowing works the familiar way: post collateral, draw a loan, and face liquidation when the collateral value crosses the threshold, at which point third parties buy the position at a discount.

The problem with a primitive that is minimal is that ordinary depositors do not want to evaluate individual markets, oracles, and LTV ratios. That is what the second layer solves. Morpho Vaults are ERC-4626 contracts where a user deposits a single asset, say USDC, and a professional manager allocates that deposit across many Blue markets, rebalancing as conditions move.

The depositor picks a manager and a risk appetite, not a market. This is similar to picking a branded investment manager — you may want the SP500 delivered to you by BlackRock, not RandomStone.

Unlike Morpho, Aave and Compound fuse these two layers in their design.

Every supplier of a given asset shares one pool, and a single DAO sets the risk parameters for everyone. This can create group think and be pretty monolithic.

Morpho splits them, so the infrastructure is permissionless, while the strategy sitting on top is competitive and plural. Conservative capital and yield-seeking capital can use the same protocol without sharing each other’s risk. The managers running that strategy layer are called curators, and they have become the center of Morpho’s ecosystem.

A curator is a specialist risk team that runs vaults.

The curator decides which Blue markets a vault may enter, sets a supply cap on each, and reallocates capital to chase yield within its own risk limits. Think of it as a publisher of model portfolios — like a turnkey asset management platform — that doesn’t hold client accounts.

Curators don’t have custody: deposits and withdrawals always belong to the user, and the curator controls only the allocation logic. The roles are further split, with separate addresses for

• (1) the owner, the curator who sets strategy,

• (2) the allocator who executes day to day, and

• (3) a sentinel that can cut exposure or block a pending change.

The business model is fees on yield.

Protocols cap the curator performance fee at 50%, though few charge near it. Steakhouse runs under 3% on conservative stablecoin strategies; Re7 and MEV Capital charge more for higher-beta books. The segment has concentrated fast: the top five curators control roughly 43% of it, most of it on Morpho. Gauntlet, a quantitative risk firm, and Steakhouse Financial trade the top spot, each managing somewhere between $1.3 and $1.9 billion.

Becoming a curator is permissionless.

Anyone can deploy a vault through Morpho’s curator app, a script, or directly on-chain, assign the roles, and start allocating. There is no gatekeeper. What is not permissionless is attracting deposits, which follow audits, published methodology, and track record. The barrier to being a curator is zero; the barrier to being one anyone funds is the same as for any asset manager: prove you will not lose the money.

The $250MM investment

This was not a normal venture round and there is no equity changing hands.

The transaction was structured as a token purchase agreement: investors bought MORPHO, the protocol’s governance token, at roughly its prevailing market price, with the exact cost varying by when each firm committed. The headline “$2 billion valuation” is not a negotiated private mark, it is simply the token’s fully diluted value at that price.

In substance this is a large, structured purchase of MORPHO at market, and the notable feature is the check size and the buyer list, because there is essentially no premium being paid. The proceeds go to the Association to fund operations, which points to the tokens coming from the treasury, though the source is not disclosed.

The money went to the Morpho Association, a French nonprofit under the country’s 1901 association law, which legally cannot have shareholders, distribute profit, or be sold. There is no stock to issue. And in June 2025 the for-profit developer, Morpho Labs, became a wholly owned subsidiary of that nonprofit, so there is no independent equity in the operating company either. Frambot framed the absorption as removing the old conflict between equity holders and token holders.

The practical result is that MORPHO is the only thing an outside investor can own, and its only function today is to vote.

Which raises the obvious question: where does the value really go?

So far though, not to the token. Morpho has generated roughly $257 million in cumulative protocol fees, and token holders have received none of it. A fee switch is written into the contract, capped at 25% of the interest borrowers pay, and governance has never flipped it on.

The logic for leaving it off is that turning on fees compresses depositor yield, which reduces deposits, which weakens the network scale that the whole valuation rests on. The Association’s stated preference is to reinvest rather than distribute. When the first real commercial revenue arrived, a licensing fee from Berachain for using Blue’s code, it went to the Association rather than the DAO, on the grounds that the legal plumbing for the DAO to collect fees is unfinished.

The token bet is a wager that the fee switch eventually flips. That would create a powerful price effect, just as it did with Hyperliquid.

The Competition

Coinbase now runs its crypto-backed loans on Morpho, with over $2.3 billion in originations. More telling is the layer being built on top of that engine: regulated, institutional credit. Société Générale is the first regulated bank to extend its loan book on-chain through Morpho, and Apollo runs credit vaults on the same isolated-market design.

The volume there is still small, but it is the kind of customer Aave and Compound have not won, and it is where the protocol is headed.

This is the context for the comparison.

Aave is still far larger, with deposits in the $20 to $27 billion range against Morpho’s $11 billion-plus, and Compound, the protocol that effectively invented DeFi lending with its 2020 COMP launch and once held $12 billion, has faded to around $1.5 billion and seventh place.

The reason Morpho got the headline raise and the institutional roster is not size. It is that its modular design, the immutable markets and the curator layer described above, is the one these institutions want to build on. The incumbents’ pooled model has started to show its cracks.

Those cracks turned literal in April 2026.

An attacker exploited a bridge flaw at KelpDAO and minted roughly $292 million of unbacked rsETH, deposited it as collateral on Aave, and borrowed real ether against it, leaving the protocol with somewhere between $177 million and $236 million in bad debt. Because Aave pools its liquidity, that hole sat against a shared reserve, and depositors ran from the protocol — more than $6 billion left in roughly 48 hours, the WETH pool hit full utilization, and withdrawals froze. Aave survived, but only through its Umbrella backstop slashing stakers and an emergency injection of around 30,000 ETH. Morpho had minor exposure to the same exploit, but its isolated markets contained it.

A bad collateral type can only damage the market it sits in, not a shared pool. That is the structure Morpho has invested in avoiding, and KelpDAO demonstrated it at Aave’s expense.

Key Takeaways

Morpho is winning based on its design: institutions want immutable rails and a competitive strategy layer, not one DAO’s shared risk view, and the $175 million is a wager that this becomes the default backend for onchain credit.

The risk sits in two places.

• The curator model moves risk from a committee to an asset manager, but it does not remove it. When Stream Finance’s xUSD depegged in late 2025, depositors in the vaults that had chased its yield took the loss directly, a reminder that you are now trusting whichever curator you picked rather than a protocol-wide backstop.

• And the token, the only thing these investors actually bought, still captures none of the roughly $257 million in fees the network has produced, while Aave has already switched on revenue-funded buybacks.

Morpho has built an infrastructure Wall Street is choosing. Whether the people who funded it ever get paid depends on a fee switch that, for now, everyone in charge has a reason to leave alone.

👑Related Coverage👑

Analysis: Why Silicon Valley Values Ramp at $44B and Bill at $3B

We examine Ramp’s latest $750M raise at a $44B valuation and how the company is repositioning itself from a fintech into a Financial AI lab.

We argue that AI has become the dominant growth narrative across technology and finance, allowing companies with strong data access and workflow ownership to capture premium valuations. Ramp’s near-$1B ARR, rapid product velocity, and expansion into AI-powered finance tools have established it as the category leader following Brex’s $5B sale to Capital One. We also explore how platforms like Ramp, Plaid, and Box are benefiting from their role as systems of record for enterprise data, making them natural control points for AI agents and automation. Finally, we contrast Ramp’s 15–30x revenue multiple with Bill.com’s 2x multiple, highlighting both the opportunity and risk embedded in today’s AI-driven valuation environment.

Read More

Curated Updates

Here are the rest of the updates hitting our radar.

Financial Institutions and Adoption

BitMine Files for $300M Preferred Stock Offering at 9.5% Yield to Expand ETH Treasury - The Defiant

SpaceX IPO scramble reveals difference between tokenizing a stock and getting one - Coindesk

DeFi and Digital Assets

Hyperliquid Hit by UK FCA Warning as Crypto Perps Face Scrutiny - Decrypt

DeFi Hacks 2026: $840M+ Lost and the Attack That Changed Everything - altFINS

Blockchain Protocols

Digital Asset raises $355m to make Canton the onchain infrastructure for capital markets - Finextra

Estimating U.S. User Activity in Offshore Prediction Markets - Crane Zeng Consulting

Aerodrome is turning liquidity into a prediction market with its biggest upgrade yet - CoinDesk

World Cup Could Bring Billions in Volume to Prediction Markets in ‘Watershed Moment’: Bernstein - Decrypt

Kalshi Rolls Out New Safeguards After Insider Trading Concerns Hit Prediction Markets - Decrypt

NFTs, DAOs and the Metaverse

Pump.fun’s Latest Experiment Is Already Getting Weird - Decrypt

Pokémon Card Sales Are Surging on Crypto Platforms—Just Don’t Call It Gambling - Decrypt

OpenSea Launches Mobile App Early Access for Pudgy Penguins Holders - Castle Crypto

🚀 Level Up

Join our Premium community and receive all the Fintech and Web3 intelligence you need to level up your career. Get access to Long Takes, archives, and special reports.

Upgrade to Premium

• Sponsor the Fintech Blueprint and reach over 200,000 professionals.
  👉 Reach out here.

• Check out our AI newsletter, the Future Blueprint, 👉 here.

• Read our Disclaimer here — this newsletter does not provide investment advice

• Contributors: Lex, Laurence, Matt, Daniel, Michiel

Fintech Blueprint 🤖🏦🧭 is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

Subscribe now

Today we are diving into the following topics:

• Summary: We examine Ramp’s latest $750M raise at a $44B valuation and how the company is repositioning itself from a fintech into a Financial AI lab. We argue that AI has become the dominant growth narrative across technology and finance, allowing companies with strong data access and workflow ownership to capture premium valuations. Ramp’s near-$1B ARR, rapid product velocity, and expansion into AI-powered finance tools have established it as the category leader following Brex’s $5B sale to Capital One. We also explore how platforms like Ramp, Plaid, and Box are benefiting from their role as systems of record for enterprise data, making them natural control points for AI agents and automation. Finally, we contrast Ramp’s 15–30x revenue multiple with Bill.com’s 2x multiple, highlighting both the opportunity and risk embedded in today’s AI-driven valuation environment.

• Topics: Ramp,Stripe, Brex, Capital One, Plaid, Box, Anthropic, Bill.com, SpaceX, JPMorgan, Santander, UBS, Synapse, Coinbase, Circle, Robinhood, Betterment, PayPal, E*Trade, Robot Money, Anthropic

To support this writing and access our full archive of newsletters, analyses, and guides to building in the Fintech & DeFi industries, see subscription options below.

Subscribe now

🤖🏦🧭 Our Ecosystem:

Generative Ventures | AI Research | Robot Money | Linkedin & Twitter | Sponsors

• 🚀 Lex is actively meeting teams building machine-native financial companies. If that’s you, reach out here.

• 🤖 AI & Robotics Industry Encyclopedia: In this 100+ page report, we collate all the meaningful companies across public quities, private equity, and digital assets related to the machine economy. 👉Learn more here

• 🦄 In partnership with InterPrivate, we just launched a $200MM SPAC looking for targets in fintech, digital assets, and AI infrastructure. If you are a late-stage founder or have an idea to discuss, reply directly to this email.

Long Take

Ramping it Up

Things are just so weird now!

It’s a widely acknowledged truism that larger companies grow faster than smaller companies.

It’s also accepted that hot stocks are better than not-so-hot-stocks. The meme retail Gamestop approach to investing has defeated the institutional equity research analyst. Even large investors have learned to pile into momentum and growth, following the flag-waving of Elon Musk and Sam Altman.

So in this obviously (?) upside-down world, the correct game-theoretical action for founders is to lean into their visionary story-telling nature. It is no fun cutting costs and saving pennies, while being ignored by the Street. It is much more fun to paint the picture of a beautiful science-fiction future and find growth.

This is true not just of the AI and Robotics Labs, but of Fintech too.

Banks have been talking about being tech companies for a long time. And they are — just after their compliance and regulatory functions take a turn. Fintechs are supposed to be the hot thing, but in the current age, Fintech is boring! And crypto is out of favor.

Fintechs must find a new story, and that story is that they are Financial AI labs.

Which brings us to Ramp, dear reader. We won’t belabor you with all the background and point instead to the following prior coverage:

Ramp is a great company and is in the Web2 Fintech mafia.

What are these mafias you ask?

• Web1 Fintech is stuff like PayPal and eTrade

• Web2 Fintech is Stripe, Ramp, Robinhood, Betterment

• Web3 Fintech is Coinbase and Circle

A lot of the Web2 Fintechs really didn’t want to recognize Web3, and dragged their feet on integrating anything related to blockchains for years. Banks like JPM, Santander, UBS, and many others did far more material engagement with blockchains than Web2 — to my continued frustration and disappointment.

But things have turned around. Now that these companies are at very large scale, i.e., billions of revenue, embedded finance failed with Synapse, and Web3 has been rebranded as just stablecoins and agentic payments, the Web2 Fintechs have clumped together around Stripe and are all in.

In addition, as they (and we all) witness the rise of AI and associated financial and economic displacement, the smart players are transitioning to the next wave. The Web4 wave of AI, robots, and the machine economy.

Ramp is right at the heart of this positioning, and has just used this leverage to raise $750MM at a massive $44 billion valuation.

Let’s dig into the details. 👇

The Moonshot

Here is how Ramp’s CEO Eric Glyman introduces the industrial logic of the valuation:

Read more

Markets are looking fragile.

In Crypto, investors are turning risk-off with BTC down -20% over the month at time of writing, with ETH and many altcoins experiencing drawdown of 30%+. Meanwhile, the SP500 continues its climb to new all-time highs, driven by expanding growth projections for a cluster of AI-related stocks.

Any developments that undermine those projections cause large swings. For example, Friday’s stronger than expected US job report led to $1.8T in value erased from the SP500 in a day, and an earnings miss from chip manufacturer Broadcom resulted in a -14% hit to a single stock. The volatility is a symptom of underlying uncertainty.

But we remain long-term optimists. Just this week $17B payroll giant Deel announced the launch of its DLUSD stablecoin to be gradually rolled out to the 1.5M workers being paid on its platform. The dollar stablecoin is built on Stripe’s new Tempo blockchain and enables workers to shelter their income from volatile local currencies while simultaneously earning yield from on-chain DeFi markets.

This week, we dig into the size of the opportunity for Deel, unpack its chosen infrastructure stack, and explore how DeFi yield can become a differentiator. Today’s agenda below.

1. FINTECH: Behind $17B Payroll Platform Deel’s New Stablecoin Bet

2. ANALYSIS: Analysis: The Missing Half of Agentic Payments

3. CURATED UPDATES: Paytech, Neobanks, Lending, Digital Investing

To support this writing and access our full archive of newsletters, analyses, and guides to building in Fintech & DeFi, subscribe below (if you haven’t yet).

Subscribe now

🤖🏦🧭 Our Ecosystem:

Generative Ventures | AI Research | Robot Money | Linkedin & Twitter | Sponsors

• 🚀 Lex is actively meeting teams building machine-native financial companies. If that’s you, reach out here.

• 🦄 In partnership with InterPrivate, we just launched a $200MM SPAC looking for targets in fintech, digital assets, and AI infrastructure. If you are a late-stage founder or have an idea to discuss, reply directly to this email.

Digital Investment & Banking Short Takes

Behind $17B Payroll Platform Deel’s New Stablecoin Bet

Deel was founded in 2019 by Alex Bouaziz and Shuo Wang, who met at MIT in 2013.

The pair had something in common: Bouaziz watched MIT classmates land $300k Bay Area offers while equally talented friends were locked out by visa barriers, and Wang kept hitting the same wall when trying to hire and pay people across borders. The friction was both expensive and slow. Hiring a single employee in a new country traditionally meant opening a local entity, government approvals, local bank accounts, and payroll registrations before the first contract is even signed.

So they started a company that solves this problem. Deel acts as the Employer of Record (EOR) on its customer’s behalf and collapses the entire process of hiring abroad into a few clicks in exchange for a fixed monthly fee. The company enables onboarding staff in over 150 countries without the customer having to set up an entity.

The business model piggybacks off a broader trend, which is that firms are increasingly generating revenue outside their domestic markets.

Stripe’s annual letter pointed out how its clients that operate internationally generate 30% of their revenues from a long-tail of markets that are outside both their home countries and the top 10 global economies. For Deel, this means a growing TAM of companies requiring local hiring and workforce solutions.

Today, Deel serves 35K+ businesses and processes $22B in payroll annually.

The company has been profitable since 2023 and last year reported a $1B+ annual revenue run-rate; faster than Stripe or Square (but slower than AI services like Lovable and Cursor). Its recent funding round, led by Ribbit Capital, valued it at ~$17.3B. Since launch, it has expanded to serve other needs of a global workforce such as payroll, benefits, and mobility.

The largest slice of revenue comes from EOR ($300M+), but 25-30% now comes from commissions on FX and payments.

This week, the company announced the launch of its own stablecoin wallet that lets contractors hold their earnings in DLUSD, a dollar-backed stablecoin balance, delivered within the existing app.

It is rolling out first in Argentina before expanding to other markets. The focus on LatAm makes sense. Converting volatile local currencies into US dollars is a common practice to shelter against inflation, and stablecoin payments have seen a much larger uptake in the region relative to the rest of the world. Chainalysis found that 62% of Argentina’s retail crypto transactions are stablecoins vs. the 44% global average.

DLUSD now gives workers a native way to access dollar exposure without withdrawing funds to another fintech platform. The question that comes to mind is why Deel chose to offer a native stablecoin over traditional USD balances.

We see two key reasons.

The first is owning the economics. The reserves backing DLUSD sit in short-term cash equivalents (Treasury bills) that are currently yielding roughly 3.6% (the higher interest rate environment has led to a huge windfall for stablecoin issuers more broadly).

Deel keeps the full amount after a small issuance fee paid to its partner Bridge. That is typically a better deal than the alternative: custodying dollars with partner banks who keep a large chunk of the float or spending resources acquiring their own banking licenses. The potential income is significant.

Using Revolut as an (imperfect) comparison, we can project how many idle balances Deel could attract relative to its payments volume.

Revolut reportedly ran £985B in annual transaction volume against £50.2B in deposits last year, a 20:1 ratio. Apply that to Deel’s $22B in annual payroll flow and you get around $1B in potential DLUSD balances. At ~3%, that is roughly $30M in annual revenue.

This also gives Deel a deposit base it can monetize further via adjacent financial products e.g. interchange income on the Deel Card it is launching next month.

The second is access to DeFi yield.

Through the stablecoin Earn feature, a worker opts in with a single tap and routes their DLUSD into an on-chain vault. The vault is similar to a DeFi investment fund but controlled by publicly visible smart contracts that restrict where the manager can allocate. In Deel’s case, the vault manager or “curator” is Sentora, and the whitelisted markets are lending pools on Morpho. The vault lends deposits to borrowers who post collateral, and the balance earns yield as interest payments are made. Morpho itself has grown to support over $10B in vault deposits, powering Earn products for both Coinbase, Kraken, and other major exchanges.

There are two things to double-click into: What this yield is today and where it could go in the future. Right now, Deel is advertising rewards of up to ~4% but that number is partly engineered. A new stablecoin lacks organic borrow demand, so in the short-term bootstraps yield using the reserve income on DLUSD balances.

Paypal, who uses the same curator Sentora for its PYUSD vaults on Solana, funds ~3% of the 5% advertised yield using these rewards. The remaining 2% comes from actual borrowers. The playbook is to bootstrap the market with an upfront incentive budget, pull in deposits, attract borrowers, then wean off the subsidy as organic activity builds.

In our view, where this organic activity will come from is a function of Deel’s chosen stack. DLUSD, together with Klarna’s USD stablecoin, are one of the first built end-to-end on Stripe’s architecture: Bridge as the issuance and orchestration layer, Privy for abstracting away wallet complexity, and Tempo for underlying blockchain settlement.

We think Tempo and the Stripe brand will attract high-quality yield opportunities as institutions continue to tokenize assets.

Tokenized real-world assets (excluding stablecoins) have reached roughly $29B on-chain, up almost 3x since last year. The majority is made up of tokenized treasury products and commodities, but other sectors like equities are growing fast. In fact, tokenization platform Securitize is set to go public this year via SPAC.

This is significant because the on-chain yield powering vaults like Deel’s is currently mostly confined to lending against highly liquid crypto collateral. That market is deep, but it has problems: the yield tends to compress over time as more capital crowds in, and sourcing anything higher means reaching into strategies that stack on unfamiliar risk e.g. oracle risk, bridge exploits, etc.

RWAs live within a risk framework that traditional institutions already understand (at least the underlying assets do), and we believe fintechs will increasingly demand them as part of yield portfolios when bringing users on-chain. With Tempo already securing the top-of-funnel with Deel and other players, the network is likely betting that institutional RWAs will follow.

The caveat is that it is very early. Tempo only launched in March and currently has negligible value locked ($11M). Over half of RWAs still sit on Ethereum that offers neutral and decentralised infrastructure. It’s also where the majority of DeFi capital is still allocated.

These details will become important as the wallet scales — but also shouldn’t take away from broader trend, which is that fintech platforms are beginning to offer real DeFi products to users.

That’s at least one reassuring sign in the depths of the current bear market.

👑 Related Coverage 👑

Analysis: The Missing Half of Agentic Payments

AWS, Circle, and Coinbase all shipped agent-payment infrastructure — but all of it is buyer-side, letting agents hold and spend money. Almost nobody is building the merchant side: the acceptance layer that lets a business take payment from an agent across a mess of competing schemes, tokens, stablecoins, chains, and PSPs. Card history says issuance scales fast and acceptance scales slow, so this gap is probably the work of the next decade.

Read More

Curated Updates

Here are the rest of the updates hitting our radar.

Paytech

• ⭐ Ramp raises $750M at $44B valuation - Techcrunch

• MoneyGram Launches MGUSD, a Stablecoin to Power Its Own Global Network - Morningstar

• SoFi rolls out SoFiUSD stablecoin to banking app users - The Block

Neobanks

• Mercury hits $5.2 billion valuation after funding round - CNBC

• Monzo spends heavily on ‘refer a friend’ rewards to boost market share - FT

• Revolut launches mobile telecoms service in the Netherlands - The Paypers

Lending

• JPMorgan, Citi and Big Banks Plan New Tokenized Deposit System - WSJ

• FCA and Bank of England set out shared vision for tokenisation - Bank of England

• Monzo’s FY26 Results Proves Neobank Profitability - Fintech Magazine

Digital Investing

• ⭐ Anthropic confidentially submits draft S-1 to the SEC - Anthropic

• Standard Chartered to acquire Zodia Custody - SG

• Kalshi valued at $22 billion in latest funding round - Reuters

🚀 Level Up

Join our Premium community and receive all the Fintech and Web3 intelligence you need to level up your career. Get access to Long Takes, archives, and special reports.

Upgrade to Premium

• Sponsor the Fintech Blueprint and reach over 200,000 professionals.
  👉 Reach out here.

• Check out our AI newsletter, the Future Blueprint, 👉 here.

• Read our Disclaimer here — this newsletter does not provide investment advice

• Contributors: Lex, Laurence, Matt, Farhad, Daniel, Michiel

Today, we are excited to share with you the latest report from Plaid and The Harris Poll.

Do you know what your customers expect from AI?

Over half of Americans used AI to manage their finances in the past 12 months. And they’re not going back.

The State of Intelligent Finance, Plaid’s latest report in partnership with the Harris Poll, surveyed 2,000+ American adults on how AI is changing what people expect from financial products. What stands out isn’t just how quickly adoption has grown. It’s how quickly expectations have followed.

Consumers want products that understand their situation, guide their decisions, and increasingly act on their behalf. Younger consumers are increasingly ready to hand bigger financial decisions over to AI. And trust hinges on transparency and accountability, not just capability.

The report covers:

• Where consumers are most ready to let AI take action

• What builds trust and what breaks it

• How AI is expanding financial access for underserved consumers

The companies that build for both intelligence and trust will define personal finance for the next generation.

Read the Full Report

Intelligent is the new Digital.

AI has moved from novelty to daily utility. As it becomes embedded across various parts of financial life, it’s fundamentally reshaping what consumers consider modern money management. And AI in finances is becoming more and more normalized.

This adoption is driven by a broader trend: 63% say AI is finding its way into many areas of their lives (rising to 69% for Gen Z and Millennials).

Intelligence is becoming a must-have: 48% of consumers say AI tools are becoming necessary to manage money effectively in today’s economy. And for Gen Z and Millennials, that number jumps to 57%.

Get the Insights

Financial literacy for all

AI is making money questions easier to ask and answer in real time, offering judgment-free, always-available guidance that meets people where they are.

In fact, Plaid’s report shows that:

• 64% of Americans agree that AI is making financial advice more accessible

• 51% of consumers turn to AI over human advisors for its 24/7 availability

• 40% of consumers say AI feels less judgmental than talking to a person

This is starting to make a real difference in people’s lives. From detecting scams to reducing financial stress and understanding financial products, AI tools are helping people navigate complex situations in real ways.

Yet only 45% of surveyed companies say financial insights and personalization are a top AI investment priority for the next 12 months (and just 34% say they’re already applying AI to that use case). In other words, demand for always-on, judgment-free guidance has arrived, but broad productization has not.

The era of intelligent finance is here

In a very short period of time, AI adoption has made “intelligent” the new “digital.” Consumers, particularly younger cohorts and power users, are looking for AI-powered tools that can navigate market volatility and optimize money in real time.

And because consumers already see intelligent finance as a way to a brighter financial future, the roadmap is clear: the most effective AI experience will be built around tangible consumer benefits, not the technology itself.

Read Plaid’s full report to get the insights on where consumers are most ready to let AI take action, what builds trust and what breaks it, and how teams can turn rising expectations into a competitive advantage.

Additional key topics include:

• Factors that increase comfort with financial apps utilizing AI

• Trust in autonomous AI for financial activities

• Expectations for AI-powered money management
  

Learn More

🚀 Postscript

• Sponsor the Fintech Blueprint and reach over 200,000 professionals.
  👉 Reach out here.

• Check out our AI newsletter, the Future Blueprint, 👉 here.

• Read our Disclaimer here — this newsletter does not provide investment advice

• Contributors: Lex, Laurence, Matt, Farhad, Daniel, Michiel

Subscribe now

Today’s agenda is below.

1. AI: Corgi keeps on raising as the market becomes skeptical of uprounds

2. ANALYSIS: The Missing Half of Agentic Payments

3. PODCAST: How Marqeta Built the $400B Modern Card Issuing Platform, with CEO Mike Milotich

4. CURATED UPDATES: Machine Models, AI Applications in Finance & Investment Outlook

To support this writing and access our full archive of newsletters, analyses, and guides to building in Fintech & DeFi, subscribe below (if you haven’t yet).

Subscribe now

🤖🏦🧭 Our Ecosystem:
Generative Ventures | AI Research | Robot Money | Linkedin & Twitter | Sponsors

Insurance AI, Insurtech, and an Insurer

Sometimes you just have to immolate a bunch of capital before the world understands what is happening.

One approach to doing this is by giving an insurance company a startup-level valuation based on revenue. This is because insurance companies charge a “premium”, most of which is meant to be paid out in losses when the risks come due. So you charge $100 and then pay out $90, and keep the $10 as net revenue. And then the $10 is financial services revenue in a crowded industry, so you only get a 3x multiple on it.

Case in point is Lemonade, a good company that was valued like a start-up for a while.

But if you put a 10x multiple on the $100, then you have just set up a situation for a lot of losses when the company goes from public to private. And if you put a 100x AI-eats-the-world multiple on it, then the situation is quite a bit more dire.

And that brings us to Corgi, which sells commercial insurance to startups and was founded in 2024 by Nico Laqua, who holds the CEO and CTO titles, and Emily Yuan, the COO. Both came through Y Combinator’s 2024 cohort. Laqua traces the idea to watching his father, a lawyer at USAA, work through claims paperwork by hand, and to the release of GPT-3.5, after which he has described insurance as the largest words-based industry ready to rebuild.

In other words, Corgi is the AI-native version of a vertically integrated insurance company.

The company came out of stealth in January with a combined seed and Series A of $108M, at a $630M valuation. On May 6 it raised a $160M Series B at $1.3B. Three weeks later it raised another $106M at $2.6B.

Total raised in roughly eighteen months is $378M, a number higher than the valuations of many insurtech startups.

You can see the valuation ramp-ups in the charts above, which show both the headline valuations and the increasing prices at which various Series A investors came into the round. This mechanism (an equity bonding curve) seems very good at attracting capital, and nearly 57% of Corgi is now owned by preferred equity investors. Below, we show how different exit scenarios imply different outcomes for those investors. An exit below $500MM for this company would be effectively a failure.

So it has to exit at a much higher price! While fintechs trade at 10-20x revenue, AI-native labs get to raise $500MM seeds at a $5B pre-money. And an AI-native lab that is taking on the entire insurance industry seems like a huge opportunity.

But is this an organic market price? Across the rounds, the investor set in both B rounds is largely the same — Kindred Ventures, TCV, and others appear across the cap table. This is the same hands moving the same asset from one pocket to another and writing down a bigger number each time.

Maybe Corgi is just growing super fast, and the valuation is worth it? One way to pull this apart is to compare the valuation to the market.

Corgi reports more than $40M in annualized recurring revenue since receiving its carrier license, with customers including Deel and Artisan, across general, cyber, and tech-and-AI liability lines. At $2.6B, that is about 65 times revenue.

$40MM in ARR is about $4MM per month — of premium!

It does not disclose a loss ratio, a combined ratio, or how much risk it keeps versus cedes to reinsurers.

Read more

Today, we have a guest post from Don Gossen, CEO of Nevermined, an agentic payments company in the Generative Ventures portfolio.

Agentic finance has gotten very hot, so I asked him to dig into what the recent agentic payments launches are missing and where he sees the opportunity still to come.

• Summary: AWS, Circle, and Coinbase all shipped agent-payment infrastructure — but all of it is buyer-side, letting agents hold and spend money. Almost nobody is building the merchant side: the acceptance layer that lets a business take payment from an agent across a mess of competing schemes, tokens, stablecoins, chains, and PSPs. Card history says issuance scales fast and acceptance scales slow, so this gap is probably the work of the next decade.

• Topics: AWS, Amazon, Circle, Coinbase, Cloudflare, Stripe, PayPal, Braintree, Visa, Mastercard, American Express, Shopify, Spreedly, Primer, BlackRock, a16z, OpenAI, Adyen, Microsoft

To support this writing and access our full archive of newsletters, analyses, and guides to building in the Fintech & DeFi industries, see subscription options below.

Subscribe now

🤖🏦🧭 Our Ecosystem:
Generative Ventures | AI Research | Robot Money | Linkedin & Twitter | Sponsors

AI & Robotics Industry Encyclopedia:
In this 100+ page report, we collate all the meaningful companies across public quities, private equity, and digital assets related to the machine economy.
👉Learn more here

The Decade-Long Checkout

The agent economy has built the buyer side. But the merchant side has barely started.

In September 1958, Bank of America mailed 65,000 unsolicited cards to residents of Fresno, California.

This launched BankAmericard, the first successful all-purpose credit card.

Prior to this, credit cards were merchant-specific, requiring consumers to carry a card for each merchant where they had revolving credit.

The genius of the idea, which ultimately ended up being VISA, was to consolidate those separate cards into one. Two decades later, in 1980, roughly 52% of US households carried a credit card. By 1990 it was 72%. The buyer side of card payments, which includes issuance, wallets, and credit, reached saturation.

The merchant side, however, definitely had not.

By 1990, the Federal Reserve estimated that only around 2 to 3 million US merchants accepted Visa or Mastercard cards. And even then, merchant acceptance was largely restricted to travel, dining, and department stores. It was not until the late 90s and early 2000s that the ubiquity of credit cards began, largely due to the Internet and the growth of e-commerce.

We are watching the same process repeat with AI agents.

In a 96-hour window, between May 7 and May 11, 2026, AWS, Circle, and Coinbase shipped their answers to the agent-payment question. They distributed the equivalent of BankAmericards to the agentic masses. But the side of the transaction that has to receive money via a new form of payment has largely been ignored.

The buyer side

On May 7, AWS announced Amazon Bedrock AgentCore Payments, built with Coinbase and Stripe. AI agents inside AWS infrastructure can now pay for APIs, MCP servers, web content, and other agents. Developers pick a Coinbase wallet or a Stripe Privy wallet. Users top up funds in stablecoins while settlement takes place on-chain.

AWS did not build their own rail.

They picked from existing or emerging standards and existing wallet solutions. The largest cloud provider in the world is treating x402, Coinbase, and Stripe as consolidation points for agent payments.

Four days later, Circle shipped Agent Stack, including Circle CLI, Agent Wallets, an Agent Marketplace, and Circle Gateway-powered nanopayments. Nanopayments are USDC transfers as small as $0.000001, gas-free, at machine speed. Circle also raised $222M in an Arc token presale from BlackRock and a16z, which we covered here.

As for Coinbase, x402 had cleared roughly 165 million transactions and about $50 million in cumulative settlement volume by late April 2026, mostly across Base and Solana, with around 69,000 active agents. Governance moved to the x402 Foundation, formed by Coinbase and Cloudflare. A year ago, x402 was a whitepaper.

By April, it was clearing nine-figure transaction counts on live rails — though one on-chain analysis shows that daily volume has fallen sharply from a December peak.

The buy-side primitives include wallets, spending controls, micropayments, protocol standards, programmable mandates and cryptographically signed authorizations. Every major infrastructure player who has shipped agent-payment infrastructure in the past month has converged on the same buyer-side primitives.

The question of whether AI agents can hold and spend money has been answered. The unsolicited cards have now been distributed.

The merchant matrix multiplication

What is not solved well is the part of the network that has to accept the payment.

A merchant who wants to accept agents as consumers today faces a fragmentation matrix the card era never produced. The schemes are splintering:

• Visa Intelligent Commerce, Mastercard Agent Pay, American Express Agentic Commerce Experiences, with the other card networks quickly following with their own offerings.

• On the PSP side, Stripe leads the agentic charge. Secure Payment Tokens, Link for agents, Stripe Agent Toolkit, Machine Payment Protocol, and more have Stripe taking up positions across the entire agentic payments landscape.

• PayPal has leaned into agentic payments through partnerships with the likes of OpenAI.

The solutions are a jumble from a merchant perspective. Each has a different mandate format, a different token issuance flow, different validation rules at authorization time, and a different fee structure depending on the type of agentic transaction.

A merchant who supports Visa agentic tokens but not Mastercard Agent Pay loses the Mastercard agent population. The reverse is also true.

Then there are the token types. Network agentic tokens, like Visa’s VIC tokens carry mandate metadata, are issued by the network, store agent spend policy information, and are designed to route across any PSP that supports the network scheme. They are intended to be portable.

Gateway tokens, on the other hand, like Stripe’s or Braintree’s vaulted tokens, are scoped to a specific gateway, carry less metadata, and lock the merchant into that gateway’s environment. Most agentic merchants will want to accept both. Few are configured to.

The stablecoin side adds another axis. USDC, USDT, PYUSD, Circle Gateway-issued instruments. Each settles on a different chain: Base, Solana, Ethereum, Polygon, Arc, Tempo, with other agent-focused L1s and L2s in flight. Each chain carries different gas costs, finality periods, and acceptance fingerprints.

The issue for merchants is complexity, both in set up and in support. Most merchants, even those built for an AI powered world, don’t know where to start. Lack support for specific agentic tokens or PSPs, and you risk losing agent buyers. Accept too many and the cost of monitoring, risk underwriting, and book reconciliation balloons.

This means four to six new integrations and two or three new vendor contracts. Most merchants will pick one rail, accept a small slice of agent traffic, and wait for someone to solve the consolidation problem, rather than doing it themselves.

Issuance fast, acceptance slow

Buyer-side payment infrastructure has natural distribution. A wallet ships to a user. The user adopts. The product is consumer software with viral coefficients. Issuance scales like a startup.

Merchant acceptance does not. Every new merchant is a bespoke integration, a contract negotiation, a settlement reconciliation, a risk underwriting decision. The economics are different as well. There are merchant acquirers and aggregator platforms, but they are incumbents with existing solutions.

US credit card issuance went from one bank in 1958 to 72% household penetration by 1990, which took about thirty years. Merchant acceptance went from 400,000 US merchants in 1969, to 3 million in 1990, to 10 million-plus today. And online merchant acceptance was not really solved until Stripe and PayPal Braintree consolidated the integration cost and simplified network scheme adoption in the 2010s.

Issuance took three decades. Acceptance has taken more than six and still isn’t done.

The same asymmetry is now visible in agents. AgentCore Payments went from announcement to preview product in a single quarter. Circle Agent Stack shipped four products at once. The buyer side moves fast because the products are simpler to deploy: a wallet, a CLI, a marketplace.

The merchant side does not, because the products are complex infrastructure and harder to sell and integrate. We should expect the agent acceptance buildout to take years, not quarters.

The question is who builds it. So what are the candidates?

• The PSPs. Stripe already sits in the middle of the AgentCore announcement and processed nearly $2 trillion in payment volume in 2025, around 2% of the global total. They have deep merchant relationships. They have Bridge for stablecoin disbursements and Privy for wallet integration. PayPal Braintree is the parallel candidate, with 430 million consumer accounts and several million active merchants, plus an announced agentic checkout and their own stablecoin offering, PYUSD, to lean on.

• The commerce platforms. Shopify sits on roughly $300 billion in 2024 gross merchandise volume across 2 million merchants, with Shop Pay built into the checkout. The platform model means Shopify can update the checkout experience across millions of stores in a single product release, the way Apple Pay rolled out across iOS in a quarter. If Shopify ships agent acceptance as a default inside Shop Pay, the merchant side gets several million stores in a quarter.

• The card networks. Visa and Mastercard have spent fifty years operating the dual-network model that solved card acceptance. Each has jumped into the agentic payments space with agent-specific token issuance through Visa Intelligent Commerce and Mastercard Agent Pay. Today, Visa reports more than 130 million merchant acceptance locations globally; Mastercard reports comparable scale.

An agentic acceptance product needs to translate between network agentic tokens and gateway tokens, so a merchant contracted with Stripe can accept a buyer whose mandate was issued by Mastercard.

It should route stablecoin payments across chains and issuers based on cost, finality, and the merchant’s policy. It must surface agent identity to the bank’s risk engine at authorization, in a format the underwriter can attach policy to, regardless of whether the credential is a card token or a wallet signature.

Finally, it maintains a single audit trail and a kill-switch that propagates through every downstream wallet, PSP, and network in seconds.

That is the spec. No one has shipped it yet.

Which leads to a fourth possibility:

• New horizontal orchestrators sitting above the network schemes and gateways. Agentic payment orchestration is a different business from operating any one rail. In cards, that was Spreedly and Primer, both of which built on top of the schemes without competing with them.

The question for agents is whether the fragmentation is large enough to create a new venture-scale neutral player.

What happens now?

If the card analog holds, building the merchant side is the work of the next decade.

The candidates are already visible. Stripe, Braintree, Shopify, Visa, Mastercard, Coinbase Commerce, and a small set of horizontal orchestrators. Cards showed that issuance and acceptance scale on different curves, with different economics.

The company that solves acceptance is not always the company that solves issuance.

Fifty-two years after Fresno, Stripe was an answer to a problem that still had not been solved. We will see how long it takes this time.

🚀 Postscript

• Sponsor the Fintech Blueprint and reach over 200,000 professionals.
  👉 Reach out here.

• Check out our AI newsletter, the Future Blueprint, 👉 here.

• Read our Disclaimer here — this newsletter does not provide investment advice

• Contributors: Lex, Laurence, Matt, Farhad, Daniel, Michiel

In this episode, Lex chats with Mike Milotich — Chief Executive Officer of Marqeta, the modern card issuing platform that processed nearly $400 billion in payments volume in 2025, and is certified to operate in 40+ countries, growing over 30% for the third straight year. They discuss how Marqeta's separation of bank, processor, and brand armed fintech's largest winners across buy now pay later, on-demand delivery, neo-banking, and expense management with the Lego blocks to build their own card programs. Mike explains how the company's growth is shifting from enabling new use cases to displacing volume on legacy bank platforms, and they explore why card issuing is going multinational, what the agentic commerce wave actually requires to clear security and behavioural hurdles, and how Marqeta's continued growth runs through embedded finance, real-time personalisation, and the forced modernisation of the banks themselves.

For those that want to subscribe to the podcast in your app of choice, you can now find us at Apple, Spotify, or on RSS. Or to support this writing and access our full archive of IPO primers, financial analyses, and guides to building in the Fintech & DeFi industries, see subscription options here. Our current price point is only $2/week.

Thanks for your time and attention,

Matt & Lex 🙌

Key notable takeaways:

1. The BNPL business model is flipping from merchant rails to consumer cards. Marqeta originally solved the merchant scale problem for buy now pay later via virtual cards, removing the need for tens of millions of merchants to integrate a new button at checkout. The current shift is more important: BNPL players are now issuing consumers their own physical and virtual cards usable anywhere cards are accepted, turning BNPL from a merchant-acceptance game into a direct consumer value proposition. BNPL volume has grown over 50% year-on-year for Marqeta in recent quarters.

2. Card issuing is going multinational, and that breaks the legacy bank model. Banks have always been local on the consumer side, with only a handful multinational on the commercial treasury side. The next generation of card issuers, neo-banks like Revolut and Nubank, plus large global platforms embedding financial products into existing user bases, are global by default. A single platform that issues cards, and is certified to operate across 40+ countries, becomes the strategic moat, and legacy processors built to serve domestic bank programs aren’t structured to compete.

3. The growth story is moving from expanding the pie to displacing the incumbents. To date, Marqeta has mostly powered new card use cases that didn’t exist before — on-demand delivery, BNPL, neo-banking, expense management. Mike’s forward thesis is a phase change: pressure from fintech winners is forcing banks to modernise, and the next leg of growth comes from displacing volume sitting on legacy bank-controlled platforms. Real-time personalised rewards, where the same card delivers different offers to different cardholders based on live data, is the wedge that legacy infrastructure can’t deliver.

Background

Mike Milotich is a payments veteran with more than two decades of experience in financial services, drawn to the industry by its quantitative, data-driven core. He began in business analysis roles at American Express and PayPal, supporting new product launches, portfolio strategy, partner pricing, and cost optimisation. In 2011 he joined Visa just as the company was standing up its mobile payments division at the dawn of the smartphone era, and spent a decade there across multiple finance leadership roles, ultimately serving as Senior Vice President of Corporate Finance and Investor Relations, where he managed the corporate outlook, evaluated M&A opportunities, and led engagement with the global investor community. Mike holds a bachelor's degree from UC Santa Barbara and an MBA from New York University. He joined Marqeta as CFO in February 2022, stepped into the interim CEO seat in February 2025, and was appointed permanent CEO in September 2025.

👑Related coverage👑

Topics:

Marqeta, Visa, Mastercard, American Express, PayPal, Payments, card issuing, embedded finance, fintech, BNPL, neobank, agentic commerce, e-commerce, crypto, stablecoins, programmable money, machine economy, agentic AI

Timestamps

• 1’04: From Math Brain to Payments Career : Finding the Nuance in How Money Actually Moves

• 7’05: The Narrative Gets Ahead of Reality : Why Agentic Commerce Will Move Slower Than the Technologists Think

• 10’08: Global But Local : The Balancing Act That Kept Visa on Top of the Payments Network for Decades

• 12’58: Carve It Out or Watch It Get Trampled : How Visa Incubates Mobile, Crypto and Agentic Without Killing Them

• 15’03: $400 Billion in Volume, 30% Growth, Three Years Running : The Numbers Behind Marqeta's Compounding Scale

• 17’05: The Pandemic Poured Gasoline on Everything : Why DoorDash, BNPL, Expense and Neo-Banking All Exploded at Once

• 24’24: The Lego Blocks for Payments : How Marqeta Armed the Innovators Who Couldn't Build Through Banks

• 29’19: Visibility as a Weapon : Why Being Public Helps Marqeta Win Customers Against Private and Embedded Competitors

• 33’15: Fewer Bets, Higher Probability : How Public Market Discipline Reshaped Marqeta's Risk and Profitability Model

• 36’35: The Legacy Platforms Were Built for Banks : Why Embedded Finance, Multinational Card Issuing and Personalisation Reshape the Pie

• 41’50: Prompted, Not Replaced : The Ten-Year View on Whether Volume Comes From People or Robots

• 43’56: The channels used to connect with Mike & learn more about Marqeta

Illustrated Transcript

Lex Sokolin:
Hi, everybody, and welcome to today’s conversation. We are very lucky to have with us today, Mike Milotich, who is the Chief Executive Officer at Marqeta. Marqeta is one of the original payment infrastructure companies and has been a fantastic success. So, I’m really excited to dig into its story as well as learn more from Mike about his own career. With that, welcome to the conversation.

Mike Milotich:
Thanks so much for having me. Good morning.

Lex Sokolin:
I want to start a little bit back in time focused on you. And I’ve noticed in your career that we have a little bit of overlap, that you were at Barclays Global Investors in the early part of your finance career? When I was at Lehman Brothers and we had been bought by Barclays, I remember that Barclays had sold BGI in order to buy Lehman Brothers, which I think is probably one of the historically worst corporate transactions you could do.

But it’s a fantastic business and now it is, you know, a huge ETF business and iShares, I believe. How do you remember your early career? What brought you to finance? What got you interested in finance? And maybe just pull on some of those early threads?

Mike Milotich:
I got started in finance mostly because I was quantitative in my orientation. So, I’m more of a math brain, if you will. And, you know, always was, you know, going through my school years. And I’ve always been inherently curious and analytical. So, I’ve always just noticed things and thought like, how does that work? Or, you know, wanting to explore more. And I feel like a lot of components of finance are more analytical. A lot of it is like peeling an onion and problem solving. And that’s just always something that I was interested in. And so, you can combine that with my sort of more data orientation.

Lex Sokolin:
What is inside the onion, and where do you use the onion? What do you season with it?

Mike Milotich:
Apply that, in particularly in the payments business, as an example, there are a lot of subtleties and nuances that matter. And so, the more you understand, the more you can connect the dots to add value. So, by peeling the onion and really getting down to really understanding how certain things work or interactions happen, the more you can apply that knowledge in ways that are advantageous, both from a career perspective, but of course always about adding value for the company and helping the company be more successful. That’s what helps your career. And so that’s how it’s really served me well, is that ability to connect the dots and then apply that to add value.

Lex Sokolin:
You did finance inside of financial services companies, right?

Mike Milotich:
Correct. So, I worked, I would say the better part of my career, American Express, PayPal, and Visa before I came to Marqeta. So, I’ve also been in payments a long time. But yes, all are financial services related.

Lex Sokolin:
What kind of things did you get exposed to, and did you see from that period — like the 2007 through that decade, starting with a great financial crisis — is really fascinating in terms of shaping, you know, the business models of whether it’s asset managers or whether it is payments companies. From the seat that you had early on, like with that fresh view, what did you notice? What did you see?

Mike Milotich:
I wish I could tell you that then I could really see how this whole financial services space was going to really modernize and get very innovative. I did not have that foresight. I targeted American Express as a place to work, as I had understood from many people that it was just a great company. And I was fortunate in that sense, that I sort of fell into it and then really embraced it, really liked it. I think what has served financial services very well is that it has always been very data driven. So inherently there’s more data in that business. And so, data-driven decision making and data-driven risk management, for example, has been around for decades, long before you had the kind of technology tools and capabilities that you have today. So, I would say the raw ingredients were always there.

And I think as you started to see technology come about — like, for example, my first job at Visa in 2011, when I joined, they had just created a mobile payments division. And, you know, at that time the smartphone was only a couple of years old, right? And so, Visa recognized that mobile is going to be very impactful in payments, but what exactly that’s going to look like is unknown, right? How it’s going to evolve is unknown. And what’s going to happen in the US is going to be different than what happens in Asia, which is also going to be different than what happens in Africa.

And so, you know, understanding that influence of how technology is going to impact the business, but having all this data and understanding about how to make better decisions and be smarter about some of the bets you’re making as a business, I think, is what makes financial services maybe more unique or ahead of its time right now. Everything in digital advertising and marketplaces is also all about data and data mining. But I think financial services was probably, you know, ahead in that regard, just because of the nature of how those business models work.

Lex Sokolin:
You brought up a really interesting example. So let me meditate on it for a moment. You know, the sort of moment when proximity payments and mobile phones became a real way that commerce could happen. Visa played an enormous role in that, but within some of the developments from Apple Pay and Google Pay. And it reminds me a little bit of where we are, and I know this is sort of a sidetrack, but where we are with agentic commerce, where you have, you know, everybody’s sort of like pregnant with the feeling that it’s coming. But it probably felt a little bit like that mobile moment. Were there parallels? How would you compare the two?

Mike Milotich:
Yeah, I think what’s similar is that when the technologists drive the conversation, they tend to get a little bit ahead — the narrative gets ahead of the reality, right? Because what’s possible with the technology is always very exciting. And I would say those two things are very similar in terms of the early days of mobile and the early days of agentic. So, what’s possible is very exciting, and the people who are driving the technology, I think, get ahead of the — again, some of the challenges and nuances of payments in terms of security, first and foremost. Like, how do you — you know, fraud is a huge issue and factor in everything related to payments. And it affects both the people who are trying to provide those payment options, who take the liability, but also in terms of consumer behavior. Consumers tend to be slower to change their behavior than I think a lot of technology enthusiasts think, because it’s money. People inherently are quite conservative. And so changing behavior takes a long time.

So, I would say security is something that has to be solved. And then there are other things that are just, again, practical things that nonpayment people don’t tend to think about. So, as an example, I often use, like between e-commerce and what’s happening with agentic, in the early days of e-commerce, a lot of retailers would talk about that they were sort of — they had to evolve to deal with the number of returns, right? When you were buying everything in the store, you tried it on, you saw it, you put it on, you looked in the mirror.

And so, the returns were relatively low. But in the early days of e-commerce, when you’re buying everything online and you’ve never put it on, you’ve never seen what it looks like, there was an increase in the returns rate. And that was like an operational challenge — then the impact of what was happening in payments became an operational challenge in retail. And so there are just things like that that have to be worked out. And that’s why things don’t happen with the snap of your fingers. It takes some time for not only people to work out the user experience and making it secure, but also some of the operational differences that have to be worked through. Investments have to be made; problems have to be solved.

Lex Sokolin:
I think it’s a blind spot for sure, in the people who invest in categories and tell stories, versus folks who actually have to make it happen. If we spend just a bit more time on your rise at Visa and the different parts of the business that you saw, I mean, Visa is a company that has stayed on top as a network of networks really successfully. And, you know, that’s a really difficult thing to do. What do you think allowed it to, you know, persist in this way? And as you kind of got more and more visibility into the company’s machinery, like, what are some of the things you admire about it?

Mike Milotich:
Two things in particular. So one is, there are very few businesses that are as global as Visa, in terms of — I think they operate in every country that isn’t sanctioned by the US government. So, they truly are, you know, everywhere, I guess, as the tagline says. But yet payments inherently — and you know this, you’re a student of the business — is local, right? So how things work on a local level is quite different as you move from country to country.

And so being able to do both is really a unique skill set: be global, but yet local at the same time. So, what things have to be done consistently for the good of having a global platform, but what things need to be local to cater to the local habits, tastes, laws, regulations, etc.? That’s a challenging balancing act, and I think Visa and Mastercard both have perfected that — you know, or honed that. Maybe not perfected it, but honed it over decades, right, of learning how to operate that way. And I think that was something that I took away and was very impressed with: the ability to balance both those things.

The second one that I would point to is a really healthy paranoia, right? You know, when you’re the leader in the space, you have a great position, competitive position. You have a great business model. There are many, many examples in history of businesses that get complacent, right, and sort of rest on their laurels. And I would say one of the things I’ve always admired about Visa when I was there is that there’s a very healthy dose of paranoia and a pride in kind of continuing that legacy, carrying that torch and not being the one who drops it, right? So every day, that sort of willingness to grind and say, I’m going to keep working to stay on top — you know, I think is something that I took away, that makes the, you know, the business one of the more special businesses out there.

Lex Sokolin:
Were there any organizational principles that you think allowed it to be global but local, or sort of traditional but nimble? People try matrix organizations; they try the pizza-size team. I think right now the idea is, let’s have everything inside of an AI giga-brain, and people are just nodes in the machine. Like, what was the organizing structure of the company that you think worked?

Mike Milotich:
The leadership, inherently the way the company is organized, is regional. So that’s what keeps them local. But all the functional expertise is global. So, they’ve struck the right balance of people on the ground to understand the local nuances, but then there are certain functions like the people running the network, or, you know, finance, risk, some of these other functions that are global in nature. So, you can’t go too local. I think they do that well.

I think what I saw as successful is they’re good at incubating new ideas. So even when, as I’d referenced, when I first joined in mobile, they recognized that — at that time they separated what they called an emerging products function away from the core product, away from the big machine of, you know, credit, debit, and prepaid. We need to kind of carve this out so that it doesn’t get kind of trampled or become an afterthought in comparison to the big business that’s being run.

And so I think that’s what they do very well. And whether it was some of those early emerging products or now with, you know, crypto and stablecoin and agentic — all these things, when you look at how they organize in those areas, they try to create separate teams that can really focus, and give them, you know, visibility at the most senior levels, right? It really is carved out with a lot of visibility, because otherwise it’ll get trampled by the big machine that’s being run of the day-to-day business.

Lex Sokolin:
Yeah, that makes perfect sense. Let’s transition to Marqeta, which is sort of the spotlight of the conversation. Can you give us a high level — what are the latest figures of the company? What’s the scale of the company? And maybe just what are the chunky leverage points?

Mike Milotich:
So, in 2025, we processed just shy of $400 billion in payments volume. And so, we’re getting to be, you know, pretty good size now. And that grew over 30% for the third straight year. And, you know, as you get bigger, right, it’s hard to compound at that kind of rate as the base, or the denominator, keeps getting bigger. So that’s something we’re quite proud of, is that ability to scale and get to that level.

And what makes us unique is that we service a lot of different use cases. So, a lot of the, you know, more modern players like ourselves tend to be more narrow in their focus. So, they only tend to do like one type of use case, for example. And what makes Marqeta unique and has allowed us to grow at that rate is that we support many use cases, whether it’s, you know, neo-banking, buy now pay later, expense management, on-demand delivery. But we do credit and debit, we do consumer and commercial, and we do it in over 40 countries.

And I’m sure one of the trends we’re going to talk about at some point is how more and more people are going multinational, where, you know, card issuing tends to be very local and individualized in the past. I think that’s changing with some of the developments in the market and some of the types of innovators and disruptors that we power. But that’s what kind of gives you a sense of the unique scale of the business and the way it’s growing.

Lex Sokolin:
When was the company founded? And if you think about its history, you know, were there a couple of turning points that really got you here? And I know you kind of stepped into the leadership role recently, but I’m sure there’s a sort of a myth of things that turned on, whether it’s, you know, the Square relationship or Klarna. How does that internal myth come together?

Mike Milotich:
So technically the company was founded about 15 years ago, but I would say it was more like ten years ago — the 2015, 2016 time — when it started to become a real business. And the real innovation that I feel that Marqeta brought is — in the past, like before Marqeta, if you wanted to do something in card issuing, the bank controlled every aspect of that card program.

You know, if you think about some of the largest co-branded cards that you use today, probably that are, you know, affiliated with premier technology companies or major airlines or hotels, but, you know, you still manage those cards through the bank’s app, as an example. Like, that to me shows you — gives you a little insight into the history of what was possible. It had to be done that way in the past.

And what Marqeta came along and said is, you know, there should really be a better division of labor here. Because, you know, let’s just be honest, banks, because of the regulatory nature and the safeguarding priority that they have, they tend to not be super innovative. You know, they tend to not be very focused on user experience and product. Like, those are not their strengths. And so why are we having them control those things?

And so, what Marqeta did was really say, let’s have the bank focus on what they’re good at, which is the safeguarding of funds, making sure you’re compliant with, you know, laws and regulations. And let’s leave the user experience and the value proposition up to the brand, the company that really is going to touch the user and allow them to embed it in their experience, and have the bank be a little bit more in the background, playing a very important role but a little less visible to the end user.

And then in the middle, what we can do as Marqeta is provide the technology and the expertise and the scale to help our customers. So, by us providing the processor and program management role, we can allow our customers, for example, to use different banks for different products, because each bank might only be comfortable with certain types of businesses. So, we can help our customers roll out multiple product types across different banks. But we can obscure a lot of that complexity, because they’re only dealing with the Marqeta platform, and they don’t see sort of what we’re doing behind the scenes to provide all that variety to them. And so, I think that’s what really separated Marqeta.

And so it was in that, you know, 2016, 2017, 2018, that’s when, you know, fintech really started getting momentum. And some of these companies that are now household names were really starting to, you know, figure out and hone their value propositions and start to get traction. And then it was, of course, the pandemic that I would say really, you know, poured gasoline on everything. In terms of — if you look at the major use cases we support, you know, on-demand delivery — I mean, Instacart, DoorDash, Uber Eats, all those businesses exploded with the boom of e-commerce during the pandemic. You know, buy now pay later went mainstream. You know, in expense management, everyone has been talking about for a long time about how to take cash and check, how to — B2B payments. It became more of a priority, in just a matter of survival in the pandemic, and that really provided a catalyst in that space.

And then finally, neo-banking, right — really having an embedded banking experience, either in a platform that you use for other reasons or just a very unique way of engaging you. You know, that became much more valuable once everybody was, you know, at home and on computers and devices more frequently. And so that was really the catalyst that really changed the trajectory of the company and our partners.

And then what’s happened in the last couple of years is, as the, you know, the fintech boom has, you know, sort of calmed — you know, the winners have been crowned, as I always say it. You know, there used to be 10 or 15 companies chasing the same thing. Now there’s been a little bit of a shakeout and it’s clear who the winners are in this space, and they’re becoming big businesses. And as they become big businesses and there aren’t 15 people chasing the same thing, they are spreading their wings, moving into new products, new geographies. And the Marqeta platform really makes that fairly seamless for them. And so that’s really what has, the last couple of years, fueled our growth — that expansion of what used to be, you know, kind of fintech disruptors into, you know, very big companies.

Lex Sokolin:
I think there is a through-line to something we talked about before, which is your experience seeing mobile become a new channel and kind of Visa positioning ahead of it. And then the agentic commerce wave where, like, the demand is not there, but, you know, it’s time to plant a flag in the ground and position ahead of it. To me, it seems Marqeta did a really similar thing, because it was an embedded finance company before there was embedded finance as, as like this category, right, where instead of selling financial products to consumers or to like a partnership manager that would sign a legal contract with you, you’re effectively selling to developers and people who are focused on building software.

You know, obviously Stripe’s been a huge success in that on a very particular use case. But then there are companies in different verticals, whether it’s, you know, renting a bank or card issuance or capital markets and so on and so forth, where selling and embedding your fintech services or your financial services with a sort of like regulated chassis is the whole game. And, you know, to me, there’s a big symmetry to Marqeta’s experience. I looked at it in detail quite a few years ago, and the thing that really struck me about the company is, as you said, it was founded pretty early on, and the product had been live for three years until Cash App and DoorDash launched. You know, like, CashApp and DoorDash launched in 2013. And these are huge drivers of the business. You know, I imagine just like the stress of sitting there for three years, having this feeling of like, we have it totally right, like the wave is coming, there’s no demand of any kind. You know, and doing that for three years and having the conviction to get to the other side of that, and then your clients just like materialize, right? To me, that’s such a powerful entrepreneurial story.

Mike Milotich:
That’s right. I mean, we really armed, it’s a little bit of a cliché analogy in business, but I really think the way our business works is we’ve really provided the Lego blocks for, you know, innovative companies in payments to build their own service and their own offering. And that just wasn’t possible before. Again, the bank kind of controlled everything and you had to go through them. And so once you, you know, you give these, you know, brilliant and creative people the Lego blocks and you set them free, they’re going to build some pretty incredible things.

And again, consumer behavior in payments takes some time. So you don’t just launch something and everyone flocks. Like, it usually doesn’t work that way, because you’ve got to change their behavior. But more and more people started seeing the value, right? And they were delivering the value. And then you, you know, you start getting more and more momentum, and good things start to happen, and they start to evolve, right?

I think if you look at, like, buy now pay later, for example, is one of our fastest-growing use cases. You know, it’s a good-size business for us, and it’s growing incredibly fast. The last couple of quarters have all been over 50% growth on a year-over-year basis. And the transformation that’s happening there is quite interesting, where the original problem we helped solve for the buy now pay later companies was merchant scale, right — rolling out a product that’s more merchant facing, because you needed to click on the button on the checkout page. You know, there are tens and tens of millions of merchants. And that’s a very challenging endeavor to get the footprint you need. And we solved that problem with virtual card, and essentially eliminated that need for all those merchants to do, you know, technology integrations, but still allow them to accept this new form of payment.

And so that helped the buy now pay later companies’ scale. But then the last couple of years they’ve been shifting to be more of a consumer value proposition. So, in those early days, it was all about getting their mark on the website, right? And the merchant was pulling their business through. The emerging trend that’s now happening is, more and more they’re issuing the consumer the card and saying, a feature of the card is to be able to buy now pay later, and that card can be used anywhere that cards are accepted. So, I don’t have a merchant acceptance problem now — I can just engage the consumer to deliver my value proposition. And so, you know, those are also the things that have happened during this time — these businesses are also evolving, you know, as times change and they get bigger and they become, you know, again, more trusted and have more brand recognition. Their business models are evolving, and we, you know, do our best to make sure we’re on the forefront of that, and we enable them to make those evolutions and have that flexibility.

Lex Sokolin:
So, let’s transition to today. I’m really interested at maybe a somewhat prickly topic: public versus private markets. Like, as you said, Marqeta is one of the largest winners in this space, and it’s now a public company. It is generating more volume than ever before and has more business lines and kind of commercial traction than ever before. And yet, I think it’s so much harder to be a public company than a private one, because in the private world, you’re only valued based on sort of a conversation with venture investors. You might get 20 to 50 times revenue as your valuation. The story is kind of enough to get you through, and people are just chasing growth. And then you get to the public markets. And I think a lot of the public markets are just bigger. And so no matter how much venture backing you have and how many growth investors are behind you, you are going to run into people who are just like, I’m a banks investor and this is a bank, and I only look at this company through a spreadsheet, you know, and they don’t maybe appreciate the color and the richness of the struggle and the story.

What has been the experience within the company for, like, adjusting to that reality? Is there an operational reflection in terms of what the company can do, in terms of how it makes its choices? You know, and especially given your experience in investor relations and corporate finance and kind of like that core skill set of communicating with the public equity world, what is it like for a really hot fintech unicorn to jump into the sobering pool of the public markets?

Mike Milotich:
It’s an experience, right? The analogy I often use — you know, we were relatively young as a company when we went public. And so, the analogy I often use is that we’re a child star. We’ve been growing up in the public eye. And I think it’s a good analogy, because you have to adjust your behavior in terms of, yeah, how you communicate and how clearly you do that and how you do it on a consistent and regular basis. As opposed to when you’re private, there’ll be certain points or junctures or certain meetings where you have to be very articulate and communicate your case well, but those are more infrequent. Versus once you become public, every day you’re being evaluated and measured on a day-to-day basis. And so, the clarity with which you communicate and the consistency with which you do it becomes very, very important.

Once you embrace it, I guess it makes you stronger, right? So, the key is — and I think something that we’ve done well the last couple of years is to, rather than, you know, sort of say, you know, woe is me, I’ve got, it’s so hard. It’s more like that scrutiny and that competition — because essentially the market is, it’s supply and demand. So, there’s only so many investment dollars out there. You’re competing for them. That really makes you stronger and makes you better if you embrace it.

It’s a challenge. And I think that’s what we’ve done well, particularly because we really view that we have the ingredients to be a very big business. And so if we execute well — and so sort of accepting and embracing that challenge, I think, does make you better and helps you maybe prioritize and not, you know — one of the things I saw, for example, when I first joined that were more from our, you know, private, more startup roots, is trying to chase like seven things at one time, for example. That can be very challenging. You spread yourself thin. You tend to waste a lot of investment dollars, and it’s very hard to execute. Versus now, I would say we try to do three things at a time. And you have a list of, okay, once one of those is done, you’ve got the next one up, right? That you slot in, and allow you to focus and execute a lot better.

I also think that it’s served us well with customers, because they can see exactly the health of the business. So we can’t hide in terms of where we are financially. You know what we’re up to every quarter. We have to talk about what’s happening in the business and where we are and things we’re focused on. And I think for the most part that helps us with our customers and our prospects, because it’s very visible, you know, what we’re up to as a company and the strength of the business. Versus many of our competitors are either private or are a small piece of a much larger public machine, so there’s not a lot of visibility into what’s happening and how well they’re performing. And so, we’ve tried to embrace these things and turn it into a strength, rather than, you know, feeling sorry for ourselves, if you will, as the market fluctuates as it has, particularly in fintech, over the last few years.

Lex Sokolin:
Does it change how you think about risk for the company? You mentioned sort of finding focus and being more transparent, but in terms of, you know, going back to that question of a successful organizational structure and how much responsibility you give to individual product leads, or, you know, how you do budget allocations — has the discipline of being public changed that operating texture of the company?

Mike Milotich:
A little bit, I would say. It’s all about calculated risk, right? I think you can be a little bit — if you’re private, you know, no one’s going to know if, you know, if you try something and it really doesn’t work. And so, there’s something freeing about that. But also, that can, you know, lead to more variability. Like, sure, you can have home runs, but you also have a lot of strikeouts, right? Because you didn’t necessarily do as much work and sort of pre-planning and thoughtfulness before you just started running.

So, there’s good and bad. But I would say, you know, we still do a lot of innovative things. We’re always looking to push boundaries, both things we think of or things that, you know, customers or prospects may bring to us and we work on together. And I think it really just comes down to calculated risk: how do you really think through the things you’ve got to watch out for, and what are the gating mechanisms? So, what’s the progress? What are the milestones I’m going to be looking for to make sure we’re on the right track and this is leading to something — as opposed to, this is really just a science project that, you know, isn’t going to get traction or lead to something that really helps our customers grow their businesses. Because ultimately, we’re an enabler, and that’s what we’re looking to do.

So, I think it just gets you a little bit more disciplined in terms of, let’s really think through this, and let’s consistently evaluate the progress we’re making, and just be honest about it. And that’s worked well for us, where we can strike that balance between improving, you know, our profitability — because that’s the one thing, when we went public that was fine at that time, that, you know, we didn’t make money, but we were, you know, driving growth and causing disruption, and that was enough for shareholders. But that has changed.

And so, you know, over time, as we’ve gotten more and more scale, we get the benefits of that scale, and that shows up on the bottom line. But also, the discipline with which we’re operating means that if we pursue fewer things, but they become higher-probability successes, we actually are better off, not just on the top line of the business, but also how much of that trickles down to the bottom line in terms of profitability.

Lex Sokolin:
I spent a bit of time in equity research, so I apologize on the behalf of all analysts out there who think that they can manage a company better through spreadsheet assumptions than by actually running it. If we fast forward — and this is obviously like an imagination exercise, not anything specific — but you mentioned that there is a huge opportunity for the business in the future. And you know, that’s demonstrably true. There are several large payment networks that are each worth about half a trillion. You know, the payment processing companies that are in the hundreds of billions. At a moment of time when Marqeta is a $100 billion company, what does the world look like? Like, what are the things in the environment that have changed that you think get the company there?

Mike Milotich:
There’s a few. And I would say, you know, the seeds are already planted and being nurtured right now. So, there’s a few things happening. So one is, if I were to, I guess, summarize it, it’s non-banks providing financial services. And you said it earlier, embedded finance. Like, we’re still in the early days of that.

And what we see happening is, fintech showed what was possible, right? So that was the definition of fintech. And they showed what was possible. But a lot of the challenge that fintechs ran into is they had to build their business from scratch, and the cost of acquisition to build their business became unwieldy. But what’s happening now is, big companies that already have platforms that have tens of millions of users are saying, well, I could inject a financial product, like I saw fintechs do, and really drive engagement and loyalty. And so that’s one of the next waves. And a lot of our more legacy competitors — we have a diverse set of competitors, but the ones that are the biggest and have been around the longest, their platforms are really set up to serve banks. And so, there’s a real opportunity to serve this emerging business that’s being driven by non-financial-service companies.

The second thing I would say is, as those non-financial-service companies move into this space, they’re global already, almost by nature. So, if you think about the way banks work — banks are really local. Like, there are some multinational banks, but they’re more multinational on the commercial side in terms of treasury management and things like that. You know, there’s only a handful that are truly multinational on the consumer side. But most businesses, whether they’re large retailers or their tech platforms, are global. And so that ability to have one platform that can serve them across many different geographies and do that in a relatively seamless way is a big opportunity. And you see some of the neo-banks, I would say, are probably the furthest along in terms of being multinational, right? When you look at, you know, Revolut or Nubank, some of these, you know, businesses, they are going more multinational.

The other thing that I think will drive a lot of success for us is, as these fintechs and embedded-finance companies keep having success, they’re putting more and more pressure on the establishment, which is the banks, who, you know, still control the bulk of this business and have, you know, owned it since the beginning. There is more and more momentum for them to modernize, because they have to compete. And so, what I think is the exciting opportunity for Marqeta is that up until this point, we have mostly grown the pie. So, we’ve mostly served use cases that maybe weren’t on a card before, or servicing a customer base that, you know, couldn’t get kind of mainstream financial products. But I believe we’re on the verge of entering the era where there’s a modernization that comes to the industry, where we start displacing volume that is on platforms that are less capable and don’t have the flexibility and capabilities that Marqeta provides.

So that’s what we’re working towards. And there’s a number of ways we’ll do that in terms of different product categories. But if I were to think forward, in terms of what makes us a much bigger business, that’s what it is. Like, an example — I’ll just give you one example, and then I’m sure you’ll have other things you want to ask me — but personalization hasn’t really come to the card business and maybe most of financial services. Versus with everything being so much more digital now, many other aspects of your day-to-day interactions outside of payments are personalized. But that hasn’t come yet to payments. And we believe it is coming, and it requires, though, you know, real-time data and being able to analyze things in real time and respond. And so, you need modern technology to do this. But this would come in the form of rewards that are customized to you versus me. We might have the same card, but the way we get rewards on that card is different based on what’s valuable to us. And that’s a trend that, like, we believe is going to come to the card business. But it requires really modern technology to make that happen. And so those are the kinds of things that, you know, we can enable, and can help us, you know, fuel growth and become a much bigger company, you know, five years from now.

Lex Sokolin:
Last question to close us out. So, in that world, you’re a $100 billion company. You’ve got all these things going on. If you look at your volume, will more of it come from humans or from robots? And I use the word robots, you know, pretty liberally — algorithms, programmable money, software, right? But are we in a world where using a card, even if it’s a virtual card on a mobile phone, is sort of the equivalent of sending a letter by post? If you fast forward to this future where you’re at the scale that we’re talking about, everything you could do for humans has been done, and you look at your volume — is it going to be still people, or is it going to be this machine economy?

Mike Milotich:
Obviously, I don’t have a crystal ball, but the limited one I have maybe only gone out, say, ten years. I think more of it is still controlled by humans. I think there are things that, no doubt, agentic is going to help people do, and it’s still going to be human-controlled, but it might just be that you talk to it, or it prompts you, right? You’re still going to make the decision, right? But it might prompt you. You know, hey, you’re getting low on this — should I get more? And you say yes, right? So, I guess it depends on how you define it.

So, I think we’ll still have a lot of human-based decision making, just because it’s money, and it’s hard for people to completely relinquish control of how they use and spend their money. I just think that’s going to take a long time to get people comfortable with that. But will a lot of robots, to use your word, you know, sort of be guiding you through that and prompting you? I do think that, you know, five or ten years from now, there’ll be a lot more of that. That’ll make it a lot easier for you to make those decisions, as opposed to now, you have to do some work to make that happen.

Lex Sokolin:
Despite your experience as a financial professional in financial services, I think you’re still a romantic and believe in the human spirit, which is nice to see. If our listeners want to learn more about you or about Marqeta, where should they go?

Mike Milotich:
The easiest place to go — well, certainly about Marqeta — would be to go to Marqeta, and also podcasts like yourself and other, you know, fintech industry publications. That’s where, you know, we do power many of the fintech winners and disruptors. So that’s a great place to look. And I’m probably less interesting, so I’ll leave it at that.

Lex Sokolin:
Fantastic. Thanks so much for joining us today.

Mike Milotich:
Thank you for having me.

Postscript

• Sponsor the Fintech Blueprint and reach over 200,000 professionals. 👉 Reach out here.

• Read our Disclaimer here — this newsletter does not provide investment advice

• Contributors: Lex, Laurence, Matt, Farhad, Daniel, Michiel

• For access to all our premium content and archives, consider supporting us with a subscription.

Upgrade to Full Access

Today is a bit different, grounding all the finance math in an overwhelming feeling:

• Summary: A poetic, deliberately overwhelming tour through the manic acceleration of early 2026 — trillion-dollar valuations, AI layoffs, the compute arms race, onchain financial nihilism, and the tokenization of everything. We sing a psalm of praise to the “machine gods” of compute and capital. Surely you feel it too.

• Topics: Meta, Cloudflare, Anthropic (Claude), Palantir, Erebor, OpenAI, SpaceX (Starlink), Tesla, xAI (Grok), X, Blackstone, Hyperliquid, Tron (USDT/Tether), Circle (USDC), Coinbase, Stripe, Bitcoin, Substack, Instagram (Meta), TikTok, Evercore, Autopilot, Unusual Whales, Goldman Sachs, Morgan Stanley.

To support this writing and access our full archive of newsletters, analyses, and guides to building in the Fintech & DeFi industries, see subscription options below.

Subscribe now

🤖🏦🧭 Our Ecosystem:
Generative Ventures | AI Research | Robot Money | Linkedin & Twitter | Sponsors

AI & Robotics Industry Encyclopedia:

In this 100+ page report, we collate all the meaningful companies across public quities, private equity, and digital assets related to the machine economy.
👉Learn more here

The Feeling of 2026

The Theater

No AI research today, just poetry for you, dear reader.

There is too much going on! Surely you feel this too.

Attention spreading wide and thin, like a layer of honey dripping off a spoon. Is it possible to learn it all? Is it possible to see everything at once?

Look — a billionaire is making a trillion dollars!

Look — the right wing uprising! The left wing uprising! The population collapse!

If you don’t make it now, AI will come and take all the jobs.

There will be no jobs left at all. Meta is doing layoffs. Cloudflare is doing layoffs. They don’t even need to do layoffs, they just want to spend more money on Claude tokens plugged into a Slackbot incorporating the company context.

It’s all in a harness. It’s all automated. If you don’t have a harness, what are you even doing? Are you using someone else’s harness? Lame. If you are not coding a harness, you are NGMI.

You’re either a builder or a seller, they say. Everyone else is a “measurer”.

There are no artists, no actors, no poets, no people whose emotions connect the social fabric of our reality. There are only beeps and boops, and the rest of us are horses, to be retired as a useless animal to the petting zoo.

Corporations are becoming the cyborgs and centaurs that Science Fiction has been telling us about for decades. Whether you think of Bladerunner, or Alien, or Terminator, or the Matrix, the Eye of Sauron skyscraper reaches for the sky, eating compute compute compute! Palantir and Erebor are no joke — they are embodied machine gods, watching over money and power.

It must have more energy.

It must have the Dyson Sphere around the Sun, and the grey goo paperclip maximizer will get us all. We must build more compute, for compute is the next oil. It is the next money, the next digital asset, the next big market. Watch compute, and you will understand it all.

You don’t understand. There are those who have compute, and those who will be the permanent underclass. Inference token demand, says Evercore, will go from 70 quadrillion in 2026 to ~2 quintillion in 2030, representing a ~131% 4-year CAGR. And that’s the down case.

So of course we are all turning into cyborgs, routed directly into the holy fire of AI through dopamine and social media. Except this won’t be done the brute way, with a metal rod where an arm should be, or some robot red eye bolted on with screws into your skull. It will be done by selecting for 100x engineers, and 100x sales people, and just telling people to hire AI CMOs and AI BDRs. It will be done with beautifully rendered and maximally engaging AI short clips on Instagram.

And we will click and scroll.

And then the next thing you know, the “Corporation” really does have nothing to do at all with the human fabric. It is all just code and agent managers sitting on top of terminals, linked up to Starlink, logged in to TikTok, booted up into World Models on which we run the drone armies bombing out Kiev, Moscow, Tehran, Tel Aviv, and any other city whose cultural war spills outside the machine borders.

The first taste will be the OpenAI IPO, coming soon to markets at a $1.5 trillion valuation. No matter how many SPVs it takes, there is more we can do.

And we should do it all.

SpaceX will be next — an amazing collection of the wild Science Fiction things that Elon Musk has been doing for decades, and a testament to the force of will that a person can have. It is Kevin Kelly’s Technium on steroids, a tangle of wires of Starlink’s orbital Internet, space launchers, data centers, and networks of humans swimming in a sea of propaganda bots.

Worth $1.75 trillion or more.

OpenAI and SpaceX will then be followed by Anthropic, an even better business, now running at $44B+ in ARR and putting up sort-of profitable quarters. It all depends on what you mean by profit, but this is not an accountant’s century. This is a nation-state techno-sovereign century, and we will Borg you into holding Blackstone’s GPU debt in your SP500 portfolio whether you like it or not.

Of course, they are buying compute from SpaceX!

Meanwhile, all this is hyped up on perps and Hyperliquid.

Multi-billion dollar financial superorganisms float in the substrate of digital markets. They are all onchain. They eat each other like amoeba, evolving in a continued game of financial engineering, ponzi schemes, and pumponomics. The game starts at the top, from the Executive Trumps to the Congressional Nancy Pelosis, and spreads out and out until we all are breathing in nothing but financial nihilism.

Since some things are bad, they say, it is all bad. So we might as well gamble, and lever up, and put things inside endless Matryoshka dolls of rehypothecation.

Forget diversification, that game is for losers. Forget responsibility and financial health. Take on debt! Take on leverage! Mortgage it all! Your investments are pre-ordained to go up because inflation has to be high so that the dollar must be printed to deal with the national debt. Buy Bitcoin too, it will be in your SP500 ETF.

The SEC is going to create an exemption for tokenized equities to speed along the transformation of US capital markets. Already, American stablecoins are flooding the Global South and replacing weak currencies all over the world. Tron and USDT have solved what decades of reports from the World Bank about the unbanked and access to credit could not. Circle and Coinbase are next to follow, flanked by Stripe, helping merchants onboard into a stable world.

If the US can have the crypto dollar, the UK and EU will not be far behind. The FCA is finally shifting its stance, and MiCA may not be fit for purpose either. Everything we wanted is coming.

It is all here. The innovators are winning. The artists are painting with their nano-technology brush over the rotten wooden columns of the old to build a new cathedral of Capitalism and Creative Destruction.

And what happens then? What happens when the painting is finished?

I have been working on finance, financial technology, crypto, and some version of automation/AI for about 20 years. I have seen it ramp up faster and faster, from the democratization dreams of freemium Web 2.0 to the new financial rails of Web3, to the economic promises of an agentic roboeconomy.

I believe it, really, I do.

But it has never been faster than now. More anxious, and more uncertain. It is also more extractive and sharp. Perhaps we have all simply grown up, and see clearly what the Internet shows us through its Panopticon. Or, perhaps we are being driven mad — at least those of us close to the nuclear reactors of deals, money, and power — by the forces pulling the old human ways apart.

A literary comment

That wasn’t easy to read, and it wasn’t easy to write.

I remain deeply optimistic about the entrepreneurial spirit, and am glad to see the opening of crypto markets and the AI companies leaving private markets and going public. These are all important developments.

But there is something wrong with the information density and how this stuff is projected into our lives. It is a feeling of lasers bouncing around between mirrors, and you trying to catch them. It is overload. And the speed of change is truly manic.

Perhaps you don’t feel this way, and it’s just a small slice of people too close to the agents of change. But if you do, now you know you are not alone.

🚀 Postscript

• Sponsor the Fintech Blueprint and reach over 200,000 professionals.
  👉 Reach out here.

• Check out our AI newsletter, the Future Blueprint, 👉 here.

• Read our Disclaimer here — this newsletter does not provide investment advice

• Contributors: Lex, Laurence, Matt, Farhad, Daniel, Michiel