Hi Fintech Architects,
In this episode, Lex chats with Evan Malanga — Chief Revenue Officer of Yuma, a subsidiary of Digital Currency Group focused on growing the Bittensor ecosystem. They discuss how Bittensor's $6 billion protocol incentivises AI builders worldwide through token emissions across 128 competing subnets, and why the network has produced real commercial outputs — including a 72 billion parameter model trained on-chain and a coding agent rivalling Claude at a fraction of the cost. Evan explains Yuma's role as the institutional gateway to Bittensor through its validator, accelerator, and asset management products, and they explore why the concentration of AI in OpenAI and Anthropic is a systemic risk, and whether Bittensor's future extends beyond AI into a broader coordination engine for decentralised work.
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:
Bittensor has crossed from experimentation into shipping benchmark-competitive work at a fraction of centralized cost. Three recent proof points: Templar (subnet 3) completed the largest decentralized pre-training run of a 72B parameter model using only the network’s token incentives. Ridges, an AI agent platform, is hitting 88–90% on software engineering benchmarks, on par with Claude-class agents at ~5x cheaper, built by a 3-to-5-person team under $10M of token emissions. Score (subnet 44) is doing computer vision 200x faster than centralized counterparts. Small distributed teams are producing outputs competitive with frontier labs without raising venture capital or hiring staff.
Dynamic TAO restructured emissions from validator-curated to market-curated, making each subnet its own tradeable asset. Previously, dominant validators assigned weights that determined how the 7,200 daily TAO emission flowed across subnets. Under Dynamic TAO, each of the 128 subnets has its own token denominated in TAO, and any holder can buy or sell into specific subnets, pricing them like a market rather than a committee vote. Subnet owners, miners, and validators earn fees in the respective subnet token. Distribution has settled into a power law: the top ten subnets hold ~80% of market cap. This is the move that turned Bittensor from “decentralized AI protocol” into a financial hyperstructure with hundreds of tokenized work markets layered on top.
The economics for subnet owners are genuinely unusual — hundreds of millions in annual incentives, fully subsidized labor, no fundraising. A subnet owner gets access to up to ~256 miners globally competing to satisfy their problem statement, with miner compensation paid by protocol emissions rather than the subnet owner. At current TAO prices, annual incentives across the network run into hundreds of millions; at higher prices, this approaches $1B/year up for grabs. No hiring, no benefits, no recruiting, the network runs as a continuous adversarial competition where validators rank miner outputs. This is the mechanical answer to “why would an AI researcher choose Bittensor over Silicon Valley”, and explains why researchers at Meta and Google reportedly mine Bittensor on nights and weekends, with top miners on subnets like Ridges earning ~$30,000/day.
Background
Before becoming CRO at Yuma, Evan Malanga served as a platoon leader and battalion officer in the U.S. Army's 75th Ranger Regiment. After completing his MBA at Columbia Business School, he moved into private equity at Gotham Consulting Partners before joining ConsenSys as a founding team member of its digital securities division, one of the earliest efforts to tokenise real-world assets on Ethereum. He went on to become Senior Director of Business Development at Securitize, where he led institutional partnerships during the early days of security token offerings.
In 2021 he joined Digital Currency Group as Director of Strategy & Operations, rising to Vice President of Investment Strategy, where he helped shape DCG's positioning across its $20 billion subsidiary portfolio including Grayscale, Foundry, and CoinDesk, before leading the incubation of Yuma as DCG's decentralised AI subsidiary focused on Bittensor.
👑Related coverage👑
Topics:
Yuma, Bittensor, Digital Currency Group, DCG, OpenAI, Anthropic, Foundry, Templar, Ridges, Bitcoin, Meta, Google, BlackRock, JPMorgan, Decentralized AI, Crypto, Blockchain, AI, Tokenomics, Decentralized Science, DeSci, AI Agents, Computer Vision, Proof of Work, Tokenization, Real World Assets, RWA, Machine Economy
Timestamps
1’09: The World Wide Web of Intelligence : How Bittensor Turns AI Into Open Competition
9’48: Decentralized AI or Financial Hyperstructure : Unpacking Bittensor's Tokenomics and the Shift to Dynamic TAO
15’04: 256 Miners, Zero Payroll : How Bittensor Subsidizes the Labor Behind Every Subnet
18’03: The Olympics of AI : How Subnet Competitions Replace Bitcoin's Proof of Work
20’09: The Grayscale Playbook for Bittensor : How Yuma Is Building the Institutional On-Ramp
23’19: AI Is the Wedge, Not the Ceiling : Bittensor's 3-to-5-Year Path to Coordinating All Work
28’03: Right but Early : Why the Vision for Decentralized AI May Take 15 Years to Realize
30’52: Decentralized Science as the Next Wedge : Why DeSci Could Be Bittensor's Most Underrated Use Case
34’10: $30,000/Day Mining on Nights and Weekends : Why Meta and Google Researchers Are Quietly on Bittensor
35’56: The channels used to connect with Evan & learn more about Yuma and Bittensor
Illustrated Transcript
Lex Sokolin:
Hi everybody. Welcome to today’s conversation. I’m super excited to have with us today, Evan Malanga who is the Chief Revenue Officer for Yuma. Yuma is a really interesting company. It’s a subsidiary of Digital Currency Group, of course, which is one of the pioneers in crypto based investment management. But Yuma in particular is a decentralized AI company affiliated with Bittensor, and we’re going to dive into what that means. So, with that, Evan, welcome to the conversation.
Evan Malanga:
Lex, thanks for having me. My pleasure.
Lex Sokolin:
I kind of just want to get to the conclusion, which is to ask you what is Bittensor and what is Yuma and why are you working on it?
Evan Malanga:
We can spend a lot of time on all of those things. So, let’s start with what is Bittensor. Bittensor is a protocol. We like to describe it as the World Wide Web of Intelligence, its decentralized, open-source incentive mechanism that enables market participants from all over the world in a permissionless environment to contribute or build towards artificial intelligence projects. You can kind of hear sort of the rhyme with Bitcoin in a lot of ways. It’s, you know, it’s non-sovereign, it’s immutable. Hard cap supply, decentralized, global, etc. And instead of doing proof of work in the Bitcoin sense, which is random number generation, market participants are establishing competitions and competing across a variety of use cases to build across the AI value chain, which I think is fascinating and that’s why I’m working in it.
Lex Sokolin:
We’ve talked in the past here about decentralized AI and kind of the value proposition there, and we certainly written a bunch about it in the blueprint. And then out of my fund in Generative Ventures, we invest in the machine economy and decentralized AI. And, you know, I’ll start with the sharp question, which is, has decentralized AI worked right? Because we’re in this world where, like all of the value of, like every single market, from infrastructure investment to the price of components, to ram to the price of energy, to like sovereign debt.
Literally every single market is going to capitalize OpenAI and Anthropic and maybe a bit of Elon Musk’s ambitions. My hope was that decentralized AI was going to be this kind of open-source counterweight where privacy, decentralization, all these things matter. Are we in that world and what has happened over the last few years in Bittensor?
Evan Malanga:
So, I would say, like philosophically, I am very much aligned with you. And I think the point you just made around OpenAI and Anthropic, I mean, that is the risk here is that the world across all markets will leverage two companies and the decision making of a few individuals at those companies to determine, like what AI model parameters, outputs, training data, etc. influence our everyday decision making, be it in personal or business use. I mean, look at in recent weeks, I mean, the US Department of War going to battle between these two companies. And then, you know, you could think about all the different use cases around deepfakes, elections and things of the like all over the world that, you know, having an open-source counterweight really, really matters.
And so, I would say, you know, kind of over the last two years that I’ve really been focused in this, we’ve gone from sort of the promise of the ethos. I think we saw a lot of different projects in the decentralized AI space raise money about two years ago, that we’re still kind of waiting to see what those projects bring to market when they’re to TG events occur, but specifically in the Bittensor world. You know, this protocol. If I take a step back…
Lex Sokolin:
Just to be clear, I appreciate, like, the vote of confidence that these projects, both that they have a product to bring to market and that their TG will happen. I as a crypto VC, I really enjoy those assumptions. Thank you.
Evan Malanga:
I think you’re reinforcing the point I’m trying to make is that Bittensor has a massive first mover advantage. The founders first met and started tinkering on this in 2016. I think Bittensor has already tried and failed, and learned the lessons that many of these other projects are trying to set out do so there are multiple cycles ahead of the rest of the market in many ways.
And so what I would say two years ago would kind of be like the initial discovery, the initial excitement, that ChatGPT moment where everyone was like, okay, AI is here now, two years later, we’ve got themes around rebuilding the American industrial base, reshoring, you know, this need for energy compute, data centers, small nuclear reactors, right. Like there’s this new industrial revolution happening around AI. And I think in the Bittensor world, we’re growing up. We’re moving away from what I would describe as the experimentation phase into the tipping point of maturation. And so, to give you like a few succinct highlights in the last few weeks, one subnet called Templar completed the largest decentralized pre-training run of a 72 billion parameter model. I mean, fully leveraging the token incentive mechanism of a blockchain on the Bittensor network. That’s subnet three. I mean, that’s massive and no one’s really talking about it yet, but it’s starting to break into the dialogue. Ridges, which is a AI agent platform, is hitting like 88, 90% of the software engineering benchmarks.
It is effectively on par with like a Claude type agent for 5X cheaper the cost and was built with under $10 million of token emissions on the Bittensor blockchain. Right. So, here’s a small 3-to-5-person team being able to do this at a fraction of the cost. Fully decentralized across the globe amongst miner’s validators in the subnet owners. Another example would be score subnet 44. They’re doing computer vision over 200 times faster than centralized counterparts. Specifically, they’re doing a variety of different verticals. But where they started off was sports specifically soccer. Doing like post-game play analysis for managers, coaches to think through game strategy and stuff like that, but would take a lot of people weeks if not days, if not weeks, is taking the team a handful of minutes and now has natural language prompting to ask questions around, you know, what occurred in the game and how to prep for like, you know, the next match, so to speak. So, there is a lot of real-world use cases, a lot of tangible businesses being built, generating revenue with real customers across the board Bittensor ecosystem. And that’s just, you know, 3 or 4 upwards of 128 subnets going to 256 subnets.
Lex Sokolin:
These are kind of cool spot checks of capability I want to rewind a little bit, just to kind of set the groundwork for a bit tenser for an audience that might not be familiar with it. So, Bittensor is a token-based project. It is around 3 billion, give or take, in terms of market cap, or between five and a half and 6 billion in fully diluted values. So, there’s still about, give or take, a little bit less than half of emissions that are going to happen. It’s AI themed. And I mean, one of the directions I want to talk about is like, is Bittensor decentralized AI or is it like this financial hyper structure? And so, I’ll just describe what the financial hyper structure looks like. And then I’d love to get an update because it’s been a bit since I’ve touched the tokenomics. They’ve gotten more complicated in the beginning. You’ve got the emissions curve in the same way that you have an emissions curve for Bitcoin, and people can get rewards of these tokens.
Obviously, they’ve got no value unless they have value. And today they have $6 billion of FTV of value. And then the emissions that come out are then directed to different projects which are called subnets that you had just mentioned. And they’re directed through this mechanism where effectively other participants in network are able to, using their existing holdings, kind of decide which things are worth funding, kind of on a periodic basis through the Yuma consensus, which I’d love to hear about. And then people can get effectively millions of dollars of token rewards for doing work. That sort of aligns with the decentralized AI mission. And then since then, I think the emissions have gotten a lot more programmatic and attached to business outcomes. So, to me, that’s like a financial product. It’s less of an AI product. Can you guide us as to what the current token infrastructure looks like, and then what’s the connection here to AI? Other than sort of like what the subnets decide they want to do or not do.
Evan Malanga:
To take it one step back before I dive into this. It’s important to know that when this protocol was created and they first launched the token similar to Bitcoin, there was no venture capital round. There was no like, you know, large investment with, you know, vesting schedules and things like that. Right. Like the founders just launched the token. In order to accumulate TAO, you had to provide work to the network and earn it through the emissions. Right. And so that sort of community aspect of people just showing up here, naturally, organic growth was really important to our thesis. In addition to the parallels to Bitcoin. 21 million tokens hard caps apply for your halving schedule. The first halving just occurred in December. So, you’re right. We’ve just passed 50% of the supply have been emitted. And so, what has changed over time, kind of diving into your question more specifically, is what we went was from emissions determined by the validators, the validators would assign effectively some sort of weighting scheme. It varied validator to validator to determine the 77,200 tokens admitted the way.
Lex Sokolin:
How did one become a validator in the first place? And is Yuma a validator?
Evan Malanga:
Yuma is a validator. To become the validator. You have to create a wallet, you register in the network as a validator, but then you have to provide from subnet to subnet the validation function. It varies. So, it’s not it’s not uniform across the network. Each subnet has different validation requirements. And you have to provide that hardware, that infrastructure to validate the work of the miners on each respective subnet. So today there’s 128. So, it’s basically like validating 128 different protocols concurrently. And so, there’s you know there’s a couple dozen validators in the network I’d say roughly like the top ten, top 12 have the majority of the dominant in terms of assets staked. We’re about number three, number four, depending on the day of the week in terms of TVL on the network. And so, you know, the change in Tokamacs you’re referring to is called dynamic time. So, prior to that the validators would have basically assess these weights.
And the community realized and the foundation realized that, you know, this was this wasn’t sufficiently decentralized, and we needed to move to more of a market-based approach. And so, in order to do that, to allow any token holder anywhere in the world provide some level of input through their assets as to where emissions should flow. We created Dynamic TAO and effectively, what that is, is each of the 128 subnets has its own subnet token denominated in TAO, and market participants can buy or sell. Any token holder can kind of come in permissionless trade their TAO for a respective subnet token. The subnet owners, the miners, and the validators earn their fees on that respective subnet in that subnet token. And so effectively, what you have is what you’re seeing is more of a Curato distribution, as you would imagine, where market participants are ascribing value to you, call it. You know, the top ten-ish roughly have 80% of the market cap versus, you know, the long tail, 20%, 80% of the rest have, you know, the rest remaining 20% of market cap.
And so, it’s further decentralized the protocol into the hands of token holders, to your point of like this seems more of like a financial instrument than an AI instrument. It does introduce complexity for sure, to a new entrant coming to Bittensor. Part of the reason we created Yuma Asset Management, to simplify that entry point process for investors and token holders, to get sort of a simplified one click trade approach to getting exposure to the subnets. But overall, I would say this is allowed other subnets that may have gone unnoticed to really build their businesses and rocket themselves into the top ten because they don’t have to deal with, you know, pitching a handful of teams. They can make their announcements publicly. They can manage their own communities publicly, be it through X or discord both, and the market’s reacting accordingly to various business updates. Be it technical or commercial.
Lex Sokolin:
Looking at that financial hyper structure again, like to me it feels I’m sure you won’t like this a bit like transitioning to like a virtual model, right? So, a virtual model is you’ve got the main token for virtual and then you’ve got an agent launcher. But the agent launcher has like infinite agents, like you can have as many as you want, and they can be as thin or as thick, sort of in terms of their functionality. And then each one is tokenized. And then in order to kind of support the creation of that agent and the trading of that agent and the market infrastructure like the virtual token is required. And so, you kind of have the platform and then you have speculation on the components.
Now, in the case of Bittensor, you have a limited number of subnets. And it’s actually quite expensive to become a subnet, and once you become a subnet, you’re eligible to receive distributions from token rewards. But there’s also stuff about your own token floating and then creating markets between using TAO and your own subnet token. I guess what kind of profile is the right profile for somebody who wants to participate as either a subnet? If I’m an AI researcher and I’m like, this is the best, I want to do it. What are the ways that they would engage?
Evan Malanga:
Yeah. So, I would say, first of all, you don’t always have to be a subnet owner to participate in the protocols if you’re an AI researcher. The attractive angle here, right, is that you can come into this network without going out to Silicon Valley and having to fundraise a business, hire employees or anything like that. You get companies like Yuma, there are others that provide funding to get on, so we’ll pay the subnet registration fee. You can also buy one from the protocol directly. It de-registered an existing subnet. That’s sort of the design. And there’s talk of the cap extending from 128 up to 256. I don’t know when that will occur, but it is being discussed in the community. But effectively what you get is a handful of things. Number one is you can tap in to a few hundred million dollars of annual incentives, right. So that’s paid to the subnet owners. It’s paid to the miners. It’s paid to the validators. Right. So as asset prices and a dollar, you know, TAO to USD basis move with the markets.
You know, TAO at 300 versus TAO at, you know, 500 to 1000 whatever. You know that starts encroaching in like $1 billion of annual incentives up for grabs every year. Right? So there’s meaningful dollars at stake if you compare that to some of the more centralized kind of web2, it’s actually quite comparable. Second is you get access immediately. If you’re a subnet owner and you define your problem statement, you get access to, you know, 256 miners anywhere in the world competing Against each other to earn as much of your specific token, and you don’t have to pay them right. It’s fully subsidized by the network. It’s fully incentivized by the network. You don’t have to go hire them. You don’t have to pay them or do anything. Recruit them, provide them employee benefits or anything like that. It’s fully decentralized and it’s continuous, right. So, you get this continuous iterative model where you can constantly improve whatever it is you’re doing. So, most subnet owners, you know, kind of get punched in the face a little bit when they first launch.
It’s naturally kind of a exploitive type network where, you know, another term would be like a GAN, an adversarial type network by design to have the best competitors, you know, sort of rise to the top. And so, you’re getting the best outputs for whatever your prompt is. And those miners are getting rewarded, and the validators serve as sort of the judge. Right. Sort of ranking the miners against the prompt statement and allocating tokens accordingly. And so, you get all of this in a totally permissionless environment. You also are starting to see network effects across Bittensor, where other subnets are tapping into the work of adjacent subnets to build upon to succinctly describe it. I mean, you get access to hundreds of millions, if not billions of dollars of annual missions, continuous improving system access to hundreds of miners all over the world, fully subsidized by the protocol, and immense and growing network effects between you and other subnet owners as well.
Lex Sokolin:
Can you elaborate on how the sort of proof of work like mechanism is actually evaluated, like this evaluation of work done against the problem statement? Like technically, how does that work?
Evan Malanga:
Yeah. So, this is slightly where it departs against kind of the Bitcoin analogy. The example we use is you kind of have to think of the subnets as like the Olympics right. Everyone’s competing but no sport is the same. So, if you’re let’s use the Winter Olympics as an example like a cross-country race is different than figure skating, right? Like how you win, how you get the gold medal, and what is required of you as a competitor is totally different. And so, you can think of subnet owners as like the Olympic Committee. They’re setting out the competition. So, you’ve got training at home. You’ve got data scraping, data validation. You’ve got computer vision, AI agents, prediction markets, kind of science type things like protein folding, drug compounding. So, you have you have all these different use cases. And within that the competition is not necessarily linear. Some of these competitions are winner take all. Some of them are more designed to be more of a Pareto distribution. There’s different sort of designs within the incentive mechanism in of itself.
But your miners are your competitors. They’re your athletes and the validators are the judges. So, when you think about like if I’m asking a subnet, if a subnet is asking miners to produce social media data within a certain time window or a certain topic, the validators are just verifying is that data? Does that meet the prompt, and how quickly were those queries returned to the subnet owner? It’s a tough question to answer because it’s not so uniform across the network, but the I think the Olympics analogy is a helpful framework to understand.
Lex Sokolin:
Let’s talk about Yuma for a moment. In my mind, thinking about Yuma as sort of similar to thinking about a Bitcoin miner in 2012. We’ve had a couple of years. There’s been returns. We know what the infrastructure is, but these things can grow into tremendous size. But the business model is kind of still evolving and it’s new for a lot of people. How do you think about it and what was the origin of creating this entity?
Evan Malanga:
I think kind of going back to the sort of the DCG Bittensor origin story. I mean, this was a, I think 21 or 2022. We made our first investment buying tokens OTC from the foundation and at the time our other wholly owned subsidiary, foundry, was running all of our infrastructure, and foundry has the largest bitcoin mining pool in the world. They did equipment financing for mining machines and stuff like that, so it naturally fit within that business at first. And so they spun up a validator and started sort of validating on the network, kind of fast forward to 2024, as we were thinking through kind of the strategy of what we wanted to incubate next Bittensor was something we already had high conviction in and was like a natural starting point of let’s, let’s look there and see where you know, what we can do, what we can play. And the short answer really is, at the time, we realized there was a lot of barriers to entry to get on the network as a subnet owner. And so, running a subnet acceleration business made a lot of sense by providing the TAO to get teams on the network.
And that’s sort of where we started to find initial product market fit. We started to see a lot of demand of subnet teams coming to us for capital. The validator was a natural kind of complement to that business. And so, we combined those two efforts to form Yuma and Yuma’s objective is to. Is to grow the Bittensor ecosystem and increase network adoption. And so, we’re doing that in a variety of different ways. As I mentioned, you know, we offer an accelerator to lower barriers to entry for summit teams to get on the network. The validator. And we have a partnerships effort with other crypto providers, you know, custodians, wallets, exchanges to get institutional staking available to more sophisticated platforms versus using sort of the command line interface, which is what it was originally a couple of years ago. We self-mined. So, we have a we have a one-person mining team that’s, you know, doing some proprietary mining across the network. But we don’t have a broader sort of externally facing mining business at the moment.
And then we launched when Dynamic TAO happened, we launched Yuma Asset Management, which is, you know, we have two fun products, a subnet composite and a large cap product, which effectively are like, you know, kind of you can kind of akin to like the S&P 500 of subnets where it’s, you know, market cap weighted rebalances on a regular basis to give investors to reduce the barriers to entry for investors to get exposure to the subnets. So, it’s a simplified product. And you in addition to not only the price appreciation you get, you get the benefit of the staking rewards as well within that. And so we’ve seen a lot of success across all four business lines. But you know the nexus was really about growing. The ecosystem is like how can we be a force for good? How can we continue to support this protocol and its growth trajectory and get more teams, more builders, more investors excited about it? And, you know, a lot of that playbook comes from what grayscale did with Bitcoin and applying it here with Bittensor with Yuma.
Lex Sokolin:
If you were to sketch out the future of the bittensor protocol and Yuma, you know, DCG is famous for being first to a real large scale Bitcoin trust product that defined a a good seven years probably of the crypto industry is almost like the heart of institutional investment. What do you think things look like for bittensor and for Yuma from here.
Evan Malanga:
For Bittensor. We’ve been you know, the community talks about this. I talk about it. We’ve talked about the days like AI is the main use case. But if you really think about what this protocol has designed is really like a global coordination engine. And so like although AI is sort of rooted in the founder’s core DNA from their career experiences prior to starting the protocol, and it’s attracted to AI builders at first is I really think if you look 3 to 5 years out, I mean, Bittensor could be coordinating all types of work, all types of use cases. It doesn’t have to just be AI. I mean, you may be an existing business that has some R&D challenges and you spin up a subnet, you run some prompts, you kind of run some modeling, and then maybe you don’t need any more and you shut it down, but it provides value for you.
Or maybe you’re, you know, a startup that’s looking to kind of create something new and novel that never existed. And, you know, TensorFlow enables you to do so in a very cost-efficient way. Right? So, AI, I think is kind of where we’re finding the wedge currently. But I think this can be much bigger than, than AI specifically. And I think the role of Yuma within that is, is to continue to be that that institutional entry point. We want to be the go-to phone call for anyone looking to build, invest, get sharp and get smart about this. You know, we published reports. We just launched our second state of Bittensor. We’re publishing regular kind of data. What we’re seeing in sort of the asset management products. And so, we want to be we want to be a data resource for parties interested all over the world about what is going on and continue to support the growth of the network.
Lex Sokolin:
We talked before the call that you and I actually share quite a bit of, let’s say, institutional history between the Columbia MBA, the time of Consensys. I mean, you were also a securitize. You’ve had this background of seeing where things are going, right? And kind of positioning ahead of the curve. And I think in many ways Bittensor is trying to do that. Like it’s imagining a world where the trillion dollars of value isn’t going to the Musk organization or the Sam Altman organization, but is rather going to all of the holders of a particular token that want to guide the development of these profoundly fruitful technologies. There is this danger. And I think this is a bit of just my own thinking recently of figuring out where the value accrual is in these plays and how to position for them and how to find them. Right. So like if we think back to the early consensus days or the securitized days where 2019, 2018, 2020, all this work is being done in terms of pioneering research and proofs of concepts with every single bank around digital assets and central bank digital currencies at the time, you know, like working with the payment networks and issuance of private credit, like all the ideas that we see today were happening back then.
But then you fast forward into 2026, and I think the biggest winners of tokenization, as far as I can tell to date, are like Circle at 50 billion and Tether at 200 billion or whatever it is, right? Whereas a lot of the other companies of that time, I wouldn’t say got left behind, but are looking at smaller outcomes. And then similarly, like you look at the pioneering work that DCG is doing, and now a lot of the large institutions have come into the space. And like whenever we talk about Bitcoin as an investment vehicle, you’ve got kind of like Larry Fink and Jamie Dimon as the poster stars for that, which to me is like obscene, but nonetheless is sort of what it is. And now we’re in this third wave that you’re seeing around AI and the value expansion there. And if we look at kind of like investment returns at least in 25 and in the beginning of 26 for AI. And I’ve sort of set this up a little bit before. But if you really look at the value chain, most of the sort of value accrual is going to energy companies.
It’s going to component companies, to the chip manufacturers and to the folks who train models. And doing that in sort of like an extreme way. And I guess just in terms of soul searching and thinking about the future, how do you square this? Like being right, but being early and building exposure, but maybe doing it in a way that’s philosophical or, you know, focused on ideals versus how these things actually end up playing out, how the winners actually end up growing into these behemoths. Like, any thoughts on just that story and then any ideas on how to navigate it?
Evan Malanga:
Yeah. No, I mean, the story really resonates. I mean, we were talking before the call started, but the industry uses the term real world assets, or RWA, but back then we call them STOs or security token offerings in like, you know, we were doing a lot of novel things like tokenizing a hotel and whatnot. And, you know, you kind of needed that moment to happen first before we figured out how to tokenize treasuries and, you know, and whatnot as the sort of evolution has matured over time.
And I think the same thing will happen in these markets, too. I mean, just as much as there’s been shake outs in crypto, albeit they seem to be more newsworthy, they seem to be shorter, like every 2 to 4 years than, say, other industries like there will still be shake outs in, you know, the regular web2 AI industry. And you know, there’s a lot of teams raising a lot of dollars. And that value will not accrue to every single one of those teams. And I think the markets parallel each other. Just maybe the timeline is, is a little bit longer versus a little bit shorter. And that said, I think generally speaking, there’s the vision has always been there. It’s just the distance to realizing that vision may not be tomorrow, like most people think. You know, Bitcoin was, you know, 2009. I mean, we’re, you know, 15 years into this and we finally have, as you mentioned, like Larry Fink and Jamie Dimon talking about this.
It may be 15 years before Elon Musk and Sam Altman are talking about Bittensor. That’s totally feasible. And there will be peaks and valleys in between. And there will be teams that come and go. But I think, you know, there will be things that happen in society, in the world that will continue to drive more desire for open-source AI. The token-based incentive is a great structure to bring people into the ecosystem. And moreover, I think we’re going to see a lot of incidences with AI kind of impacting people’s lives in a negative way. I mean, I’m still a bull for the technology, but, you know, if we’re leveraging AI for, say, medical decisions or therapeutic decisions or whatever the use case you may have. Or maybe it’s, you know, legal analysis and the output is wrong. You know, that maybe that boils up to a decision made at, you know, one of these two companies, right. And so, people will want inevitably over time, there to be, you know, greater competition and more transparency when it comes to this type of stuff. And so, I think I think it’s a, it is a matter of time. But yeah, it does it does feel very early. So, we’ve got the vision, as you said.
Lex Sokolin:
When you think about other things on the horizon, are there any other green shoots that you’re seeing beyond kind of the story of creating an AI alternative? Like, are there any weird and novel and interesting things that are popping up for you that you’re starting to get kind of like your second sense on?
Evan Malanga:
One of the things that has kind of been a bit of a start and stop within this ecosystem and probably maybe crypto. Crypto more broadly is kind of the I think the industry lingo term is called DeSci or decentralized science, but what I really do think is quite interesting. Some teams have tried and failed. Some teams still exist within the network. But you know, leveraging this as a way to, you know, do something you couldn’t otherwise do in sort of the real world, like if you wanted to, you know, compound DNA sequences or drugs or proteins or whatever.
Right. Like in Bittensor, you can continuously, every 12 seconds issue a new block where you’re having miners respond to queries and just like very quickly building up massive, massive data sets on things that would otherwise be like very manual, very labor intensive. And I think I think that’s really unique. We haven’t kind of figured out the business model around it yet, which is why some teams have struggled. But I think that kind of similar to like kind of like the hotel analogy I used earlier is kind of like we need that to come and, you know, we need to fail and learn the lesson. But when it comes around again. I think it’ll be immensely valuable. And, you know, teams be a decentralized or centralized players may want to leverage that type of data and that type of experimentation for whatever, you know, business purposes they may have.
Lex Sokolin:
Yeah, I think we’re also like just to riff on that. We’re seeing a lot of open-source AI papers coming out of the top, researchers who sort of, you know, there’s been like this reshuffling in the world because Web2 is funny actually, to see, like how people’s roles have changed just even in the last 20 years. Right. Like, you used to have finance people at the front of the house and they’re making all the money, and then you’ve got it. People in the back in the basement running like the servers for Wall Street banks. And obviously that’s flipped completely where the technologists and the developers became huge superstars. And the finance people are now like doing special projects in finance and operations inside of their companies. But I feel like over the last few years, we’ve had another switch from the Web2 model, where the people that had the most traction and visibility were like these empathetic product leaders, or go to market people who had intuition for where demand was. And then between both crypto and AI, I think that has flipped completely. So, like the people who are extremely deep technically in either cryptography or machine learning or, you know, neural network architecture and then transformers and so on, like the heavy technical people are getting recruited by Facebook for like whatever it is, $100 million contract, right? So, like the value has finally gone back to like the highly technical creative people.
And I feel like a bunch of them made enough money now that they just get to research what they want. And the stuff that’s coming out is, you know, how do we get a rapidly improving recursive mathematics researcher or protein folding researcher or like, you know, biotech stuff. And we’re just sitting in front of this completely insane explosion of technology that we couldn’t even imagine.
Evan Malanga:
One of the one of the learnings we’ve had is, is there are a lot of people sitting at the Meta’s and the Googles of the world that are mining Bittensor on their nights and weekends. And you’re right, like if you are that technical individual, that researcher, you have this opportunity cost. If I can do something and maybe get hired by meta for $20 million and with, like, you know, an insane compensation package. But if I’m also the top miner on, say, ridges or shoots, I can be making $30,000 a day kind of thing. So, there’s there are parallels. It’s just it’s just not it’s just not well known enough yet. And it’s still very much in sort of the cypherpunks kind of crypto World that it’s slowly starting to break out into more of the mainstream.
Lex Sokolin:
We’re definitely in science fiction land, and it’s one of these things where the more you know the stranger it feels and is and I think biosensor, it can be a really confusing project if you’re coming at it from the outside. But at the same time, it has been a project that has maintained its value in a pretty bad bear market and through it and then recovered, which, you know, is not the case for a lot of other decentralized AI protocols that went up really fast and then melted. And I think that speaks to a large amount of different activity that’s on biosensor as well as the capital depth of the players involved. So, you know, for anyone listening, I would definitely recommend that you go deeper and check out what the protocol does as well as what Yuma does. So, Evan, if our listeners wanted to learn more. Where should they go?
Evan Malanga:
I’d say the best starting point would go to Yuma. Go to our website. We have the state of Bittensor volume one and volume two. They’re volume two we released about ten days ago. It’s kind of got the latest and greatest coverage of the protocol. Follow us on X. We’re posting kind of updates market color that we’re seeing on a regular basis as well. Those are two great places to start. And if you want to go deeper, fill out the contact us form and we can figure out a way to work together.
Lex Sokolin:
Fantastic. Thank you so much for joining me today.
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
For access to all our premium content and archives, consider supporting us with a subscription.
