Solana 创始人：加密货币即将像互联网改变其他一切一样改变金融 | EP#204

The fact that we now have stablecoin legislation, and that people are projecting $1 trillion to $10 trillion worth of digital dollars being minted over the next 5 years, is going to massively accelerate things.

Why does your company exist? And second, why do we need it?

Solana’s there to really enable 1 billion people to go fully interconnected. Bitcoin is store of value, Ethereum is settlement, and Solana is execution. I simply wasn’t interested in settlement or store of value because they’re not the kind of engineering problems that I’m interested in solving.

So, is that the future: a single machine layer for all markets everywhere?

There is no computer science reason why it can’t exist. It’s purely an engineering problem, and we’re on our way to solving it as fast as we can.

Toly, how long before this science-fiction future could come into existence?

Now, that's a moon shot, ladies and gentlemen. Everybody, welcome to Moonshots. Here are my moonshot mates, Dave London. Hey, Dave. Hey. Alex Swizner Gross. Hey, Peter. Hey. And Salim Ismail.

We have a special guest today. We’re going to be diving into Solana. Anatoly Yakovenko is here—software engineer, entrepreneur, co-founder, and CEO of Solana Labs, which today is the sixth-largest coin, now worth over $100 billion.

Congratulations on that.

We appreciate it. Yeah, yeah.

$100 billion here, $100 billion there. You’re on your way to $1 trillion.

You have a bachelor’s degree in computer science from the University of Illinois Urbana-Champaign and began your career at Qualcomm. An amazing company, right? I had a chance to watch their domination.

Solana has emerged as the leader in decentralized finance and blockchain, generating—what is that?—$2.2 billion in annual revenues between 2024 and 2025. That’s extraordinary. It’s the cool one around MIT, too. I can tell you that firsthand. I’ll give you some quotes later, but yes, you are known, and you are cool. That’s awesome.

Today, I really want to hit on a few things: crypto essentials, sort of Solana 101, and what makes it different from Bitcoin and Ethereum. We’re going to dive into Solana as a payment system, the everything coin, and then really the convergence of crypto and AI, which is going to cause this explosion in the global economy.

Let me begin with a question that is on my mind, Toly. Given our incredible rush toward AGI, every dollar in the ecosystem is being sucked into this black hole of computronium we’re building across the planet. We’ve got AI, we’ve got agents, we have Solana, and we have stablecoins. What’s the future of money going to look like? Do you think it’s going to be recognizable in the next 10 years?

The 2 things that I see converging are that the cost of intelligence is dropping and markets require intelligence. Because it’s now cheaper to have intelligence analyze all the signal in the world, you can now have a lot more markets.

Public, permissionless blockchains like Solana allow you to create markets permissionlessly for whatever random thing it is. You saw this with prediction markets, with Polymarket and Kalshi taking off, but weird, cool experiments like futarchy—where you have decision markets for every decision that a fund or a company can take—can now exist.

As intelligence gets cheaper, you have more markets that are viable, and you start to see this exponential explosion of everything being decided through market forces. I think it’s like the quote I made before the podcast: the ant is not aware of the intelligence of the anthill. I can’t fit all these markets in my head, or understand the outcome of all of this, but my gut is that this is probably the most optimal direction for society to move forward and make decisions.

The more market-based it is, and the more intelligence you have to make those decisions correctly, the better. The forcing function of losing money is a good way to course-correct when you have bad intelligence. Hopefully, it’s a good thing, but I can’t fit it all in my head. It’s beyond my comprehension.

The speed of change is awesome right now. Alex and I talked about this as the economy 3.0 of the future. Before we started recording, Toly and I were lamenting the sad case of Truth Terminal, the autonomous AI agent that’s being forced to mint its own memecoins just to survive in this really harsh world for AI agents right now.

Toly, I’d be curious: in principle, new layer 1s, including Solana, offer the premise for AI agents just to survive. If humans, for the most part, were banked, we would have the ability to engage in a human economy. But if you’re a baby AI agent just trying to find your way in this world, and you can’t open a bank account or interact as a first-class citizen, what do you do? What’s Toly’s handbook? What’s your guide for a baby AI? Can you tell a good joke?

Then you can survive. I think that is the final litmus test for true AGI. Can you entertain a person? Not even a complex person—can you keep a toddler entertained?

How do you economically survive if you’re a baby AGI and you want to be autonomous? I think memecoins are a weird Keynesian beauty contest for attention. It could work. If you can keep people paying attention to you by being entertaining, then you can probably pay for the inference to do that.

That’s the main business model that I see right now for these poor baby AGIs. So, maybe as an assignment to you, the creator of Solana, we need a better business model for these baby AGIs to survive other than minting memecoins.

Alex, can you imagine an AI, an AGI, coming to you and saying, “Listen, would you please adopt me? I need a human shield. I need a human to fend for me in the banking system. I need a credit card”?

I think, in some sense, Peter, that’s the corporation model that we have right now, with AIs using corporations as embodiments for themselves as economic actors.

Every week, my team and I study the top 10 technology metatrends that will transform industries over the decade ahead. I cover trends ranging from humanoid robotics, AGI, and quantum computing to transport, energy, longevity, and more. There's no fluff. Only the most important stuff that matters, that impacts our lives, our companies, and our careers. If you want me to share these metatrends with you, I write a newsletter twice a week, sending it out as a short 2-minute read via email. And if you want to discover the most important metatrends 10 years before anyone else, this report's for you. Readers include founders and CEOs from the world's most disruptive companies and entrepreneurs building the world's most disruptive tech. It's not for you if you don't want to be informed about what's coming, why it matters, and how you can benefit from it. To subscribe for free, go to demandis.com/metatrends to gain access to the trends 10 years before anyone else. All right, now back to this episode.

Why don’t we kick ourselves off with a little bit of the Solana vision? Our buddy Mike Saylor is the Bitcoin promoter of all promoters. But the original vision of Bitcoin was as a transactional tool. Now it’s become a store of wealth—a huge store of wealth—but it’s way too slow to be the transactional engine of the future.

Alex’s baby AI is in no way going to sit there at the cash register for 20, 30, or 40 years of AI time trying to buy something. Ethereum came after that, but now Solana—I heard on your last podcast—is 1,000 times faster than the Ethereum chain. So, it’s keeping up with AI, basically, right?

Yeah, that’s the goal. A very simple way to think of it is: Bitcoin is store of value, Ethereum is settlement, and Solana is execution.

I simply wasn’t interested in settlement or store of value because they’re not the kinds of engineering problems that I’m interested in solving. Execution is not, I don’t think, a computer science problem where we need theoretical solutions. It’s just a fun engineering problem. How do you move bits as fast as possible around the world? This is what I like to get my hands dirty with, and I’ve always gravitated toward that.

I’d love to hear a little bit of the history behind the proof-of-history concept that you came up with, Toly, because I think that’s such a fascinating, unique approach to blockchains. Also, while we’re doing this, I’m unboxing a Solana Seeker phone live.

Nice.

I have a Solana Cypher phone that I’ve been wanting to unbox for a while, so I’m curious to hear about that.

First of all, I’m not a researcher. I’m an engineer. I spent most of my career at Qualcomm. I got there right after college in 2003 and was really interested in optimizing how fast you can move bits through memory.

I was a performance nerd. I would have dreams where I had debugger windows between different co-processors because I would stare at them for so many weeks, just trying to debug problems and things like that.

I was definitely aware of Bitcoin when it came out, and Ethereum as well. As an engineer, you often overlook the social aspect of these things. You look at it and think, “It’s not going to scale. It’s not going to work,” but you miss the transformative change that these things could provide because they’re so new and weird.

In 2017, I was working on a side project. I always had a side project. This one was building deep-learning boxes for transformers and mining crypto in the background because you could pay for the GPUs this way as a way to offset the capex.

Oh, wow.

I was just doing this for fun with a friend of mine, another Qualcomm nerd that I do side projects with. We had 2 coffees and a beer at Café Soleil in San Francisco. It’s a combination that usually doesn’t sit with anyone; it didn’t sit with me, so I was up till 4:00 in the morning.

We were discussing proof of work and this idea: could we build a single-threaded mining system that was just a totally different concept? I had this eureka moment at 4:00 a.m. that there’s a way to measure a different physical constant than entropy, which is what proof of work does for Sybil resistance. There’s a way to actually measure the passage of time in a way that’s hard to fake, and this is a recursive cryptographic hash function like SHA-256.

You run it over itself, and you run this process and sample it, and you get a data structure that you can say, “Well, it’s incredibly hard to beat TSMC by more than a factor of 2 at how fast you can make a processor. So, the amount of time that somebody took to generate this data structure is at least X.” Obviously, you can cheat and go a little faster, but you cannot go arbitrarily faster. You’re basically limited by where technology is today.

That was this eureka moment in my head, because even though I wasn’t working at protocols—Qualcomm, just being at a cellular-protocol company, you just know how the stuff works—and one of the first cellular protocols that anyone ever built was called time-division multiple access.

Yes, TDMA. Yes.

Yeah. If you remember your physics classes, if 2 radio towers transmit over the same time at the same frequency, you get noise. So, the first thing that they gave them was a clock to alternate, and that’s how you get TDMA.

In Bitcoin and proof of work, you have a similar problem. If you produce a block at the same time in 2 different parts of the world, you get a collision. The network is in this noisy state, and information is not passing through. The difficulty adjustment in Bitcoin is very similar to how ALOHA has a random backoff to retransmit in a random access radio protocol.

You’re trying to prevent this collision from occupying the same channel, so you stretch it out to make sure the probability of a collision is very small. But because you’re doing that, you’re lowering the channel efficiency. You can’t send as many bits through.

As soon as I had this thought—“Oh, I have this clock that nobody needs to trust as a third-party provider”—I could do the same trick in maximizing the number of bits I could shove through the network. My back-of-the-envelope estimate was, “Oh, this is at least 10,000 times faster than Bitcoin.” So, that was the idea: I built TDMA for blockchain.

What’s funny is that I don’t know—I was bad at selling the idea to venture capital, or maybe I was good enough. But the idea that what you’re really solving is this channel-efficiency problem was hard for me to communicate in those early days. I was more focused on, “Oh, this is a whole new consensus algorithm,” yada yada, really focusing on the implementation side more than what it unlocks.

Years later, I think we’re in year 7 for Solana and year 20 for crypto in general, and we now see that the next-generation consensus algorithms don’t actually have to depend on the clock the way proof of history does, but they do solve this problem of channel efficiency.

We got so successful that I was able to hire a best-in-class, bleeding-edge consensus team out of ETH Zurich. They’re throwing away proof of history and all the code that gets built up in the early days, but they clearly understand the problem that it solved. This is the problem you solve; this is why it’s important and why that’s a good thing, but you don’t actually need all this complexity.

Now I get to see my baby retired in a way. Just so everybody knows, ETH Zurich is the MIT of Europe. It has nothing to do with Ethereum.

Yeah, it’s a total coincidence that it’s called ETH, but it has nothing to do with Ethereum. That’s such a beautiful founding story, Toly, and in part, I never made the connection between TDMA and, just in general, the Qualcomm view of wireless and aiming for high throughput.

I see that source code for earlier versions from Solana Labs is still preserved for posterity on GitHub. I’m curious: if you were to mentally run the clock forward, what does the perfect layer 1 look like to you? Where do you see all of this going if you could fast-forward 10 or 15 years? What is it solving?

So, I think if you’re solving execution, what you’re trying to do is have as many markets around the world synchronized as much as possible. There are several challenges there. You look at something like NASDAQ or NYSE: it’s a centralized piece of memory; literally, it exists in a single physical location.

That seems like the fastest way to build something. You can build a sub-microsecond matching engine and run really, really fast markets there. But the problem is that if you have an event in Singapore where a container ship full of iPhones sinks outside your window, that information still has to travel at the speed of light through fiber from Singapore to New York before it’s in the market.

A perfect layer 1 is something that can do both. The way that we envision it at Solana is that you actually have concurrent block producers making blocks at the same time: one in Singapore, one in New York. Your latency to the block producer is as short as possible, and they’re located where all the signal is—in the most important markets.

You have concurrent transaction ingestion, and your latency to stick this data into the chain so markets can take action on it is as low as possible. This is a communication-physics problem. Again, the analogy that’s kind of dumb but works is that you’re going from TDMA to CDMA, where you have concurrent channels that you can use simultaneously.

You can shove all this information into this single, giant state machine that is churning through it as fast as we can synchronize around the world.

So, is that the future—a single machine layer for all markets everywhere, with everything living on top of that?

Yes. If you imagine science-fiction finance 20 or 50 years from now, that’s what it looks like. There’s no computer-science academic reason why it can’t exist. It’s purely an engineering problem, and we’re on our way to solve it as fast as we can.

Toly, if you could walk us through concretely, what is it in your mind? What does that future look like? Does it look like every object everywhere? We talk on the pod all the time about tiling the Earth with data centers and with compute.

No, no, no.

Okay. We agree. We agree. We agree. Okay, fine. I’m first person singular. I talk all the time about tiling the Earth with compute. But in your mind, does this vision concretely involve embedding something like SHA, or a hash function for proof-of-history generation, in every object everywhere?

You don’t need any of that. You can just have next-generation Byzantine fault-tolerant consensus, like Alpine Glow—there are a bunch of other options—but you can have concurrent nodes ingesting signal anywhere in the world where there’s valuable signal.

What’s kind of interesting for a proof-of-stake network is that people will move stake to those places so they can run those block producers more often and therefore be the center that ingests that signal into the chain. That’s the most profitable thing you can do, because the way that these proof-of-stake networks make money is that you have markets on-chain.

There is value at risk in those markets, and the faster you get data to adjust those markets, the faster you can make money. Your opportunity cost of being late, of being second to take that trade, is effectively the entire profit of that trade.

You’re now incentivized to start colocating next to the signal, whether that’s Singapore, London, New York, Los Angeles, or wherever in the future. You can move block production next to where that signal is produced that impacts markets. You ingest the data as fast as you can into this single global data structure.

It’s actually not that big, because markets themselves, trades, and all this other stuff are not a five, you know, a 4K video streamed in real time from a person in China to a person in Los Angeles. That’s what Qualcomm built: mind-blowingly complex, low-latency, high-throughput stuff. Markets and trades and all these other things are actually a relatively small amount of memory and a small number of messages in comparison.

Yeah, the NFL is building all the infrastructure for us, like that. But then how long before this science-fiction future could come into existence? Is there a tipping point at which, when enough people are on that platform, on that layer, it doesn’t make sense for NASDAQ not to be there? It doesn’t make sense for everything else not to be there?

Yeah, I think you’ll still have very localized markets, because there are advantages to being colocated and having that light cone around microstructure and the queues.

People talking about light cones. That’s great.

Yeah, in that little server room in NASDAQ, there’s actually a lot of value in having that part run.

But to build a single global layer for all this stuff to synchronize, I think that’s the challenge that we want to take on. How fast is it going to happen? I think the fact that we now have stablecoin legislation, and that people are projecting $1 trillion to $10 trillion worth of digital dollars being minted over the next 5 years, is going to massively accelerate things. Those dollars are going to get minted for all sorts of trade purposes and settlement between, you know, like, in-country and basically globally. Once those dollars are there, it’s just a very simple interface to interact with them. Even if they’re not, the price to move dollars from Ethereum to Solana is a million times cheaper than to move it between any 2 banks.

We had Jeremy Allaire on the pod 2 weeks ago, so we’re discussing this. Salim, what are you thinking right now?

It feels to me like one of the most incredible applications here would be the whole DeFi world. Can you speak to what you guys are doing there?

Yeah, so I think what our smart contract platforms do is implement some kind of escrow mechanism, where you can escrow money and then conditionally release it based on some action in the future. That could be Turing-complete computation or whatever. DeFi is basically the use case for being able to conditionally place money at risk based on an oracle, a market, or whatever future signal.

I think it effectively applies to almost everything that businesses do, right? You have to borrow money from somebody. That requires escrow and risk calculations that are running all the time and things like that. The challenge here is that, in the US, we have a really amazing financial system. It was mostly built after the railroad boom and the scams there, before World War II, and a little bit after World War II, as it was exported to the Western world.

It was all built before the internet. It functions amazingly because, through a lot of trial and error, regulators actually constructed processes and systems that are very robust, but they’re all human-based. This human action is what forces 2-day settlement. It’s very hard to cut that down to 1 day, 4 hours, 10 minutes, or 10 milliseconds because people are in the loop.

Blockchain was built—even Bitcoin, with its 10-minute blocks, was built—well after the internet. It can rely on cryptography and the fact that I don’t even know if NATO can actually reliably partition the internet anymore. We live in a super-connected world, and we have cryptography that can mathematically guarantee the correctness of certain actions.

You know, just to reinforce your point, I was at a public-company board meeting all morning, and we were talking about something we might be interested in acquiring. One of the board members said, “Well, maybe they’ll just go public.” The other board members said, “There hasn’t been an IPO below a $3 billion valuation in like a decade. You can’t even afford the legal friction. It’s so high.”

Wow, how’s that going to work in the world of AI? The friction is way too high for that system, like you said, built after the railroad—

Texas, baby.

—and before World War II.

Texas exchanges are coming.

Yeah, but there are 50,000 meme coins launched today.

Yeah, and there hasn’t been a single IPO in—I think the number of IPOs has shrunk to the lowest levels since the ’70s or something like that. Yeah, yeah, yeah. It’s—

I’d like to pull on that point, if I might. I think stablecoins are amazing, right? The Collison brothers call them “room-temperature superconductors for finance,” in principle driving international money-transfer efficiency to infinity.

But what, in your mind, is the killer app for Layer 2 and Layer 3 that expands the wealth of humanity versus just driving efficiencies in terms of money transfer? The DAOs, for a while, were going to create an entirely new class of economic actors, with mixed success. What’s the big transformative outcome that we get that radically expands human wealth?

I think what you should start seeing is the cost of finance basically dropping to its actual value. The challenge here is, you look at the Figment IPO, and I think they lost something like $3 billion in that process in a single transaction. That’s 10% of the market cap of the company, which is kind of crazy.

If you have $1 trillion to $10 trillion in stablecoins, you have incredibly deep on-chain markets. They could literally just use a smart contract to direct-list. If they use a third party to actually create that market, that party has to provide value comparable to what they’re getting for it.

They could maybe be the person they use—an analyst or whatever—to dive deep into that company and create that nice set of information that everyone else consumes, so they can participate with safety or whatever in that market. But none of the listing or access to capital—all of this stuff—is fully on-chain. There’s zero cost to actually get access to it.

All you’re trying to do then is get above the noise floor in terms of signal: “Yeah, you should allocate capital here because of X, Y, Z.” Providing that value is important. How much you charge for it should be at the most competitive price ever.

I think this is where we see that finance right now is taking out a huge chunk of GDP, and it’s a tax. It’s not actually generating as much value as it’s consuming.

In the gears. Sand in the gears. The statistic I’ve seen is that, over the last decade, 40% of corporate American profits went to the financial-services sector. It’s an unbelievable sucking sound there.

But pulling the thread on that, if I may: If the desired end state in your mind is basically driving profits in financial services to zero, A, is there anything beyond that? Or, if blockchain technologies in general—not just Solana’s Layer 1 specifically—are able to suck all the profit out of financial services, do you declare victory? Is that sort of the end, and you move on to something new? Or is there something even larger than just making financial services profitless?

I think the end result of that is you basically have talent anywhere in the world that can acquire capital from anywhere in the world. That unlocks human potential. I have a really great idea; I’m an engineer in Ukraine or whatever. I can now get funding from people in China, the US, or Australia.

I’m not relying on the trust model that a SAFE is based on in YC—the fact that YC does due diligence on this particular person, and then you can invest in them in this kind of flimsy contract, right? That, I think, is awesome. The fact that it works and reliably produces great results is a testament to how important finance is to innovation.

We need to eliminate all those barriers so you can actually have as many founders in the world starting companies, getting to that profitable state, and creating value.

The velocity of money is going through the roof, right? It’s accessibility and velocity finding a new level and just becoming a fuel for acceleration of the economy.

Can you give us your vision on regulation? If you look at Mercor as an example, here’s a company 2 years old, getting close to $1 billion of revenue, that unleashed a whole new class of employee in the AI world, but is still paying them through traditional banking means. Obviously, that needs to move on to Solana and be frictionless.

You’ve had an incredible journey through different regulatory regimes, and in just your 7-year history, it’s been the craziest roller coaster. That’s just in the US. Look at every jurisdiction in the world. How’s that going to unfold as this over-the-top economy rolls out?

Paul Atkins and Hester Peirce have been awesome, and David Sacks has been awesome. You can think of the whole securities law as having been created in the United States when your neighbor says, “I have this railroad certificate. Give me some money for it.” You get it, and that one—it’s a real certificate, that railroad company exists, and you have all the information to actually make that decision.

A lot of stuff can go wrong with that, right? Even if your neighbor is not lying to you, they could literally have a fake piece of paper. Or the company doesn’t exist, or the company lied about everything it’s doing, and there is no actual railroad. Those scams actually happened and fueled a lot of railroad construction and a lot of bad financial decisions in the late 19th century.

And a lot of regulatory paperwork.

Exactly. What’s cool now is that the difference a blockchain creates is kind of like the little SSL lock in e-commerce in the ’90s. I can transfer you a token, and you can cryptographically verify that this token goes back to the issuer with the issuer’s public keys that signed off on their financial statements, all the stuff that they published, their ASWAN, etc.

You know that you’re actually receiving what they’re claiming it to be: their equity. When you’re paying for it, all those guarantees that are solved by regulation are now just solved with math and data, just like the SSL lock tells you that when I put in my credit card number here, some intermediary sending this data is not going to steal my credit card.

So, a lot of stuff that humans do kind of goes away. Regulators have a hard time getting up to speed because a lot of stuff simply works, and there's no reason to change stuff that works when there's a lot of financial risk at stake. But the fact that the whole crypto industry, outside of any support from regulation, has been able to grow so dramatically to such a large market cap—with Bitcoin, Ethereum, and Solana leading it—is evidence that this stuff actually solves real-world problems.

The reason why somebody in Singapore can go start a protocol around trading coins, make money, and earn ROIs is because people can trust the settlement and execution of those assets, and they're willing to put money at risk. Because the software minimizes the risk, they can maximize the amount of money they actually put at risk in these systems. So it's happening anyway. It's happening with or without the United States.

I think with regulation on board now, and with the Stablecoin Act coming out, you're just going to see this accelerating. A lot of these services and intermediaries along the stack are resistant to change, but there are so many of them, and some are always looking for an opportunity to move up and down the vertical and expand.

Maybe dwelling for a moment on the regulatory side: in a paper contract—you mentioned the Y Combinator SAFE earlier—you have a few parties, and there's a whole societal apparatus built around paper contracts. You have court systems, rule of law, and regulations. With Solana or another smart contract, you have, to a first-order approximation, none of that.

So I'm curious: where do you see this going? Do you think rule of law basically moves on-chain in some sense, with bodies to adjudicate disputes also on-chain? How do disputes work in a future where, hypothetically, everything's on-chain?

I think you're minimizing the number of intermediaries necessary to do the right thing. If somebody raises money and lies about the purpose of raising those funds, that's fraud. Whether they do it on a public blockchain or not, they're still liable. It doesn't matter whether they do it with a SAFE or not.

What changes is that, if you do it on-chain, I don't have to rely on some transfer agent, broker, or whatever to facilitate moving these contracts around, or to certify that this contract actually belongs to this Y Combinator company. You can do all that verification yourself.

All the boring stuff that people don't think about when they think about finance—all those little service layers that take a few bips off the top—goes away, and you're dealing directly with the company as the issuer. They could obviously use some third-party service provider to implement the software to manage all of these things. There are a bunch of them. Squads on Solana is a great example of formally verified multisig contracts for managing governance and all this stuff.

But at the end of the day, if a company claims it's doing X and you buy this thing, the result that enforces whether that company is lying or not is going to be the local jurisdiction where it's incorporated. It's still just as liable.

I totally get it. As a stepping stone toward where Alex is going, let's talk about the actual SAFE note itself. Just to take this hypothetical transaction: I want to invest in your company; you're in Singapore. Here's my money; it's on-chain. What about the SAFE note itself? Do you just store it and hash it, then put the hash on-chain? How do you deal with the actual terms of the SAFE note, or do you not touch that?

I mean, this is where you want to draw the abstraction. There are efforts to actually use the ledger as the cap table. Whatever's on the ledger, or however these SAFE notes are allocated and distributed, those public keys and private keys—that's the actual cap table.

Then you can build all the cap-table management software on top of that because it's a public data structure. You can manipulate it and move it around. Whatever rules you want around clawing back shares, or only allowing transactions if the company agrees, you can encode all of those in a smart contract reliably.

All that stuff is just code. It's a pain in the butt because it's a bunch of database, gnarly code, but it's all doable. This would be the best way to do it because then you're getting rid of all these other layers—transfer agents, brokers, et cetera. All those people are gone.

Not just that, but I don't want to beat this Y Combinator SAFE note to death. If you look at the actual terms of the note, because we do these every day, they're not settled in the courts. They go straight to JAMS or to some third party because the courts would be years before they decided what they wanted to do.

So it's already—not on-chain, but certainly out of the courts and out of the federal and state systems because it's just too slow. I think bankruptcy could be one of those things that could be dramatically optimized.

It took 2 years to resolve the FTX bankruptcy. That was a massive bankruptcy in crypto, but FTX is not a DeFi company; it's a centralized exchange. They just bought and sold tokens like any other centralized financial system. When they collapse, it's incredibly messy to unwind all of that, go through bankruptcy law, and figure out who owns what.

In reality, if it were a single ledger, even if it were a permissioned one, you would immediately see that the amount of money going out is more than the amount of money coming in. That would be obvious immediately. So you wouldn't even get to the stage where you have bankruptcy.

But if you did, Aave has liquidations, which is effectively the bankruptcy process when you borrow and can't repay. Those are programmatically encoded and run on every block—every 12 seconds on Ethereum. Kamino does it every 400 milliseconds on Solana.

So you would never even get to the stage where you have to spend 2 years figuring out who owns what. It would get liquidated and processed immediately. I think that's a really critical part of finance. If you can invest knowing that, in case of a bankruptcy, there's a deterministic process that's run immediately, everyone gets the best of what they could get out of that process within a minute, it makes investing a lot more viable—especially across jurisdictions where you may not have the ability to enforce it in the courts, like in Ukraine or whatever.

No, I think those FTX investors were getting 50 cents on the dollar, 60 cents on the dollar, because of the insane amount of money lost in that friction, just in that FTX case. Let's go to Salim. Salim, ask a really intelligent question that I'll understand, okay?

Well, I want to make 2 points, and then I'll ask you a question. The 1st point I want to make is that there's a really important distinction around digital and crypto: the value of crypto is not the fact that it's digital; it's the fact that it's programmable.

What Toly talks about—when you can do settlements all programmatically—it manages it all. It's all done. You don't have to think about it. That takes it away from the human layer, and it takes it away from all sorts of other things.

I think it's worth taking the time, for our readers, to just step back and understand the Byzantine generals problem, because that's actually sitting at the core of a lot of this. I'm happy to get into that question, but can I take a minute and describe that, Peter?

Sure. Yeah, of course.

Okay. I remember—you're the 1st person ever to tell me about it. This is such a fun, foundational thing. For viewers who aren't familiar with Bitcoin, blockchains, Solana, et cetera, the core innovation at the heart of blockchains is what's called the Byzantine generals problem.

It's actually the story of Constantinople in the 15th century. There were 8 generals circling the city, trying to coordinate a siege. They were sending messages around that network: Who's going to go 1st? What time should we attack? How are we going to get in?

The problem they had was that 1 out of the 8 generals was a traitor and could send the wrong information, lose the element of surprise, and blow the whole operation. In computer-science terms, that's become known as the Byzantine generals problem. How do you send a trusted, secure, authenticated message over a network when you don't trust the network? That's a really hard problem.

For 40 years, computer-science practitioners have been trying to solve that problem—until the blockchain. Because of the consensus mechanism and the synchronization across multiple ledgers, you can now know that if I send Toly a message, he has a 100% guarantee that it came from me, wasn't double-spent, can't be revoked, and wasn't tampered with along the way.

That's a magical thing in the digital world. This provides an authentication layer, and a validity and validation layer, that allows all of this other stuff to take place. All of the layers that we talked about in the settlement process are there to validate, secure, and ratify that this stuff wasn't screwed up along the way.

Now that all happens as part of the infrastructure, and that makes it unbelievably powerful. The broader implications of decentralizing authentication are absolutely profound. For folks who aren't aware of this, go check out this particular problem and the nuances around it.

I have a question around that specifically.

The double-spending problem. Add the double-spending problem in there, too, because that was a perfect summary.

You can't get into that either, which is: can you stop somebody from double-spending in 2 different places and doubling up every time? The question I have for you, Toly, is that early on there was a lot of criticism of Solana for its centralization. Over the last couple of years, you guys have solved a lot of that, and I think it'd be really great for you to explain what you've been doing and the architectural changes that have taken place, because now it's really solid and robust and on its way to Ethereum-level decentralization.

I think there's a lot of debate. If you're on Crypto Twitter, people fight about what decentralization actually means. It just depends on who you ask: whatever coin they own is decentralized, and whatever coin they don't own is not. But the way that I've always approached it is that I started programming in the '90s, and I'm a huge open-source Linux fan.

It's awesome for me to be able to analyze the software that I'm running on my computer, know exactly what's happening, know why it's broken, and know why it's not. So we've always looked at permissionlessness as the core part of decentralization: it doesn't matter if the system is open-source, but can I participate in every part of the stack without needing a third party to approve me?

Can I run a validator? Can I make blocks? Can I transact on it? Can I deploy code on it? But can I also own this, have a copy of that state, and be able to recover the entire network if that kind of failure exists? Solana has always focused on that aspect of it: every part of the system that anyone can run is permissionless.

I think that is incredibly critical for this idea of a single layer for finance, because no matter what, you're still going to be dealing with people. If you have the science-fiction, futuristic layer of finance for all of the world's execution, it's still going to run in France, England, and Spain, and none of those guys are ever going to really trust each other 100%.

The banks in France will need to be able to participate in every part of the stack. So will the banks in England, and so will the banks in New York. Even if they're allies or whatever, because they're people, they want to have control, so they will need to have access to every part of the system.

To make it really decentralized, you have to allow for adversarial nodes that are either misconfigured or actually deliberately adversarial, like the Byzantine generals problem that you described, where they're maliciously signing the wrong message, trying to double-spend, and trying to create chaos in the network. The protocol has to be robust enough to handle that.

You can work it out on paper. You can say we have all these proofs and formal verification that it actually is robust with Byzantine actors, and then, in implementation, things blow up. I think the Solana journey has been that we're trying to solve all of these problems at the same time—both performance and decentralization—and doing it as a startup, shipping as fast as we can.

There's a lot of trial and error, or I would say growing pains, or blood, sweat, and tears, that went into making Solana robust. It's incredible.

Yeah, incredible what you pulled off. Here's the challenge: what percentage of the world—or let's look at the United States—do you believe understands crypto at all, if you had to guess?

A tenth of 1% of the population? I think anyone who finished calculus can understand it—basically understand it. Now, the difference is between who can understand it and who does understand it.

One of the questions is the idea of Solana as the everything coin that I'm using. Let's define what that means for us. What would Solana as the everything coin mean, and what would it take? Yeah, go ahead.

The coin itself that runs the network has only one purpose: to prevent spam in the network. Fundamentally, the only problem that it can solve is decentralized spam. As part of solving the Byzantine generals problem, where an adversarial node can send infinite messages, there's a cost to sending messages. That's the coin.

But to use it for anything else simply because it's tradable or can be easily sent, it exists as any other coin in the network, and you can use it. We encourage people to go use dollars for commerce. If you're buying and selling stuff as a merchant, go use USDC, PayPal USD, or any other digital currency.

To us, it doesn't matter what you use the underlying token for beyond providing us with cryptoeconomic guarantees that adversaries pay some fixed cost for spam. What's been surprising to me—and this has really been the case when you start a startup—is that not only do you have no idea what product-market fit is going to look like, but as soon as you have markets and money in escrow and at risk, the opportunity cost of being late to access those markets is so high.

The fact that this underlying token is what prevents spam in the network means that it can actually capture substantial value. The $2 billion or whatever that you guys mentioned captured over the last years on Solana is because there's an opportunity cost to being late to send a trade, and you're willing to pay the highest amount to be first in the queue to execute the trade.

Under the hood, the only thing that the token is doing is preventing spam. So the fact that it actually loops back and allows value capture for this thing was not planned and was totally unexpected. It was almost discovered through trial and error, after we got such massive congestion in the network from NFT trading.

That was the only way to solve the problem, and it was kind of a lightbulb moment: 1, this is a classic database-hotspot problem. Why didn't I think of this when we started? And then, of course, 2: it actually works for money.

Was the craze around NFTs and meme coins a surprise to you?

Yeah. When we started, our tagline was “blockchain at Nasdaq speed.” The idea was that we're an execution layer that can run an arbitrary number of markets and keep them all in a single state machine to maximize capital efficiency and minimize arbitrage opportunities. We thought that was really valuable.

The fact that meme coins took off was really surprising, because those are assets with no value at all, but they have a price. If you look back at the history of the internet, it makes sense. As soon as you have any kind of shared state that people have—even if I played Ultima Online as a kid—people would start trading gold.

I built all these silly scripts that would automatically mine wood and resources and post them. Ultima Online wood has as much digital value as a meme coin, a smart coin. There's nothing backing it, but people still want to trade and consume it: 1, just for fun, but 2, because you now have this shared state with some economics.

For whatever reason, as soon as you have any kind of shared state, people create markets and start trading random things.

That was Richard Garriott's company. Yeah, Lord British. Brock Pierce, our buddy. They want to be able to trade that gold and wood between different video games. How old are your kids?

10, 6, and 2 and a half.

Okay. I got 2 14-year-old boys, and so does Salim—a 14-year-old boy. So it's interesting, right? The entire video game world is teaching an entire generation about the value of digital assets in extraordinary fashion.

Yeah. Isn't it funny that the video games are where the GPU came from, which is now driving all of the AI, and it's where crypto came from: trading gold for wood or whatever within Minecraft?

Maybe to take the counterpoint to that: before modern video games, we had board games, credit systems, and sort of fake-money systems. So it's not, in some sense, that new.

But I guess maybe a question for Toly, pivoting off an adjacent question to what Peter asked you earlier about how many people do understand crypto, by which I assume Peter's referring to cryptocurrency and sort of layer 1s and layer 2s. How many should? Because I think you were making the point earlier, almost like lamports/SOL. This should be sort of under the hood, and it should just make everything frictionless.

I think of it as the number of people you invite to your wedding, which is roughly 200. That’s your core group of people. At least 1 or 2 of them understand Linux, and 1 or 2 of them understand crypto. That’s enough, because you can invite them to your wedding and trust them. They’ll explain it to you and tell you, “Okay, use it like this. This is how you minimize your risk,” and so on.

So, I think the number of people who actually need to understand crypto in the world is 1 in 200—1 per social group. Your extended trust circle. I think we’re probably there. Yeah, we’re probably there.

Striking. What I understand you to be saying is that this is not something retail investors should even be paying attention to, which is contrary to a lot of messaging out there that everyone should be paying attention to. So, what I think I hear you saying is that this is actually just infrastructure under the hood to build the next generation of financial services.

I think it is very much a B2B system. There are parts where it touches the consumer. When we built our phone, we did a presale, and we had credit card or stablecoin as an option. Half of the purchases came via stablecoin without any incentives, because the friction for somebody in Southeast Asia to use a stablecoin is actually lower than using their local credit card for an international purchase.

So consumers will figure it out, and you have demand. If you have products that are available over stablecoins, people just use those rails on their own. But the benefit to us as a merchant was very obvious. We saved 2% on the gross sale amount, which is like 3 or 4 engineering salaries—easily—just on that one product.

As a merchant, you immediately see, “Oh, if I have this rail, I don’t have to pay the 2% fee. Yeah, I’m going to use it. Why wouldn’t I?” The money is in the bank instantly. So, I think a huge improvement that will move dollars and have people take action is going to be more on the business end. Consumers will ultimately see the benefit of that.

A lot of the work now is going to be for Web3 services. Peter and I talk about going from deceptive to disruptive. When something becomes exponential, there always has to be a huge 10x change in usability. Coinbase makes Bitcoin easy to buy. Solana is making it easy to transact.

The usability layer has to be there, and many Web3 services—like DEX liquidity pools—are still incredibly difficult to use. The complexity of going through those and funding a pool is ridiculous. I have to have a hardcore crypto sherpa standing next to me to make sure I don’t screw it up. It’s insane.

Little by little, those will improve in terms of usability, and then they’ll start becoming a very powerful tool for everybody. Alex, I’ll take your pregnant pause as a question.

Yeah, no. I’m trying to wrap my mind around this. This is actually a question both for Toly and Salim. I want to put you guys both on the spot and ask: What do you think? Let’s fast-forward to the victory state.

We’ve driven the future of financial services to these highly vaunted superconductors for finance. We’ve driven transaction costs down to near zero from credit card fees of 2% to 3%. We’ve achieved victory. What do you view as the singular killer app—the concrete state that we’ve unlocked in the future—that creates radical wealth for all of humankind? What does that look like concretely?

It’s hard to say that accurate price efficiency globally is that big of a deal to a person. Do you care if Starbucks gets the best price for its coffee beans, and the person making those gets the best price? These are very abstract things that, end to end, my parents are not going to understand in a way that really matters to them.

But GDP will move faster. We will make fewer errors in finance. We will grow faster as a world, and fewer and fewer people will be in poverty. I think that is very substantial.

Yeah, concretely, though—

But let me ask you a question. Right now, what differentiates America in so many ways is its access to capital. Entrepreneurs throughout the U.S. can put forward an idea, and rich individuals can access capital. One of the questions I have is: Is this going to increase the speed of entrepreneurial creativity, financing, and company financing, where it just supercharges the economy at a speed?

If I had time, I would rewrite economics for the future, because economics are so fundamentally broken and so last century. So, is this about reinventing economics 2.0 or 3.0? If you look at all the companies worth over 1 trillion, or all the assets, they’re all U.S. companies or Bitcoin. And maybe the Saudi oil company, right?

Yeah.

So, the only other place for finance that can compete with the United States is the internet, and it’s happening on the internet through crypto. You’re now seeing the kind of finance and growth that happened in the United States being replicated on the internet through crypto rails. This is the only place that I think could compete with the United States.

At a gut level, I think it is effectively helping the United States export its influence, because the internet is truly an extension of all the values that we care about in the United States. We’re so deeply plugged into it.

I can give a real-life anecdote that may highlight some of this, Peter, and maybe answer your question, Alex. If you look at the NFT world, where they started doing art on the internet, you create a collection of 10,000 things—10,000 apes—and start selling and minting them. You can have a community building around those.

All those, by the way, were de facto DAOs because they wouldn’t have a big purpose, and they would use crypto-economics to incentivize the community to mint. The first people who got to mint got them for very cheap, and then if you minted later, you paid a little bit more, and so on.

What happened then was that you had a bunch of breakages in the system. There were some rug pulls where people collected a lot of money, and the collection would basically die on the vine. People are now doing turnarounds where they’re buying old collections that have really good art and repurposing them. There’s a whole M&A thing going on in the NFT world.

What’s fascinating about this is that it’s art in a different form, because it’s programmatic art. The experience of the art is not something normal, where you would look at a painting on a wall. You’re engaging with it in a digital mode.

With Bored Apes, for example, they gave owners the ability to license the intellectual property of their characters to people and television shows, and people started doing that. That created a new wave of innovation. Then we had CyberKongz, which created coins within the collection. If you own a CyberKongz NFT, it issued bananas to you, which traded as a secondary utility token.

But Salim, I want to challenge you on this.

Hold on. Let me—

Imagine doing all of this with lawyers.

The broader point I’m making is—

Or imagine not doing it at all.

Let me make my bigger point. My bigger point is that, as you get through this, all of the stuff that’s happening represents a collective level of innovation at scale that’s moving faster than almost the AI world. It moved faster than anything I’ve ever seen before. It is absolutely profound.

When people did an initial set of rug pulls, now when people issue NFTs or tokens, you have to hold the token for 6 months before you can sell it, and it evens out the curves. There are all sorts of things happening at unbelievable speed. The ecosystem is learning very, very fast.

There’s something profound happening and incredibly creative. What’s even more interesting is that it’s democratized, so anybody with a great idea can show up, do something, test it out, and see where it goes. That adds to the collective pool of ingenuity.

Why are these examples important? I think it’s because when I send money into this contract, I have enough guarantees to know what happens in a catastrophic event. So I can participate in this, even if it’s stupid, even if I’m getting bananas out of this—

How many bananas do you own, Salim?

Right. The fact that I can transfer, transact, and sell these bananas means that my risk is outlined. That’s very different from if I had to do this blindly over eBay, sending somebody a cashier’s check to buy their wooden item in Ultima Online. That’s a very different type of risk.

The chain eliminated a whole bunch of risk that was previously impossible to eliminate. Because of this, a lot of this stuff is going to accelerate at the tail end. The things that now have the option to use lawyers and go through the Figma IPO process through an investment bank now have an alternative.

So the price through the traditional processes is going to collapse down to what’s available on-chain. Right now, you have this massive adverse-selection problem where, if I am Figma, I don’t need to go get on-chain capital.

I can go through the investment banking process. I've been building this company for 10 years. All my lawyers are telling me to minimize risk. You're just going to do an IPO the traditional way, and you shouldn't think about any of these costs because that's the last thing that you should try to innovate on.

That's hard for a CEO to make that decision. But that's going to shift. I think that's going to shift massively and quickly as you get to this $1 trillion stablecoin amount, and then $10 trillion is going to be way past the shift, I think.

Now you're getting to the meat of it. Throw some meat to Alex here. He'll eat it like a pit bull.

Peter's question was, “Okay, I want to raise $5 million for my company. Why is this so hard?” But the assumption underneath that is that your company needs either some physical iron or something, but it doesn't anymore. What does it need? Well, all it needs is labor, for the most part, for any of these virtual companies.

Why am I raising $5 million? Why aren't I raising some virtual thing and paying those people? If those people end up being in Venezuela, Ethiopia, and Ukraine, they don't want US dollars anyway. And if I put a value on their time, which I have to do legally in the US but don't have to do across borders, they don't want to be valued at something that's taxed.

The economy of AI just solved every math problem known to man. What's that worth? It's worth a lot. I'm not going to pay for it valued by a 409A valuation from an accounting firm in the US. That whole economy of trading AI-generated things with one another, and even with human labor, is going to be completely outside the world of normal valuations and normal payroll taxes, because you don't have to value it in US dollars if you don't want to. That's where the dam is going to break.

Even if you value them in US dollars, I actually think most markets will go through US dollars. The fact that you have 2 alternative paths— I can use a SAFE, go to a YC company, and go through that process, or I can buy a token in an ICO.

The difference with a token is that I have immediate access to secondary markets. Both events are just as risky. You're taking massive risk, and basically it's, “Is this founder a jackass or not?” It doesn't matter if it's a YC founder who's a jackass or an ICO founder who's a jackass. You're going to get screwed either way.

But the guarantees you're getting out of the SAFE are competing with the fact that I have secondary markets and this reliable execution and transfer. So which one are you going to take as an investor? You may not actually be able to get any money out in the catastrophic event out of the SAFE, or any value out of that event either way. The success upside is the same in both.

I think most governments think that, regardless of all the transactions you do on-chain, sooner or later you're going to want to turn that into either real estate or labor in my country. You're going to come back through my regulatory framework, through my Treasury Department, through my SEC, to turn it into something that you can use to enjoy your life.

I think increasingly that is not true. It's irrelevant, because what you really wanted is either compute or virtual entertainment or whatever, and you can buy it outside of that regulatory framework with your Solana. That part of the economy today is not a big deal.

I think that dam's going to break very quickly, and then that part of the economy operates across borders frictionlessly, operates in milliseconds, and operates entirely in the virtual universe. That economy will grow so much faster than the physical economies that it'll go from a rounding error to dominant in just a few years. That's my guess.

If you're in the US, you pay taxes in dollars, marked to market in US dollars. So even if you make your profit and sell, you have to mark to market and cover your dollar exposure. Otherwise, you're running a debt to the place that can collect it with an aircraft carrier full of F-35s or whatever.

I think, Toly, that gets to my original question, which is that government and governance mostly don't live on-chain right now. To the extent that Peter, Dave, and, to some extent, Salim are aspirationally hoping that we're going to live in an utterly frictionless economy, it seems the elephant in the room is that all of this physical government apparatus, the court system, jails, and methods for adjudication are all off-chain.

Alex, let me take it there, because this is the conversation we had with Jeremy Allaire from Circle. I was asking him, “When do you think we're going to see the first fully on-chain corporation, where contracts, payments, treasury, governance, and even all of the agents employed by that company are on the blockchain? And we have the explosion of a new corporate structure that's operating at light speed compared to everything else?”

Do you imagine we're going to have that? When could we have that?

I think you can look at some of the cool experiments in this. MetaDAO would be one, because of futarchy, decision-market-based DAOs, where effectively every decision that this DAO makes, or this group that holds the coin, uses a futarchy mechanism. Should we go invest in this, or should we pay these engineers Y?

The way these decision markets work is that you're basically saying, “If the market decides yes, then I'm willing to buy your MetaDAO tokens at price Y.” I'm willing to pay more for Apple stock if they go and build an iPhone 20 or whatever.

This is, I would say, the closest thing you're going to get to full on-chain governance and corporate control over funds and assets.

Peter, I'd like to try to answer your question, but then forward a subquestion to Toly, if I might. I think the answer to your question relies on a subquestion: When will we see, at least within the US, the first state government approve a new type of corporation that is an on-chain autonomous corporation?

It would require an act of a state government at minimum. So my subquestion, Toly: Texas, Florida, or Wyoming?

Wyoming's been very advanced in crypto adoption and acceptance. They actually issued a stablecoin where, as you interpret the law, because they are sovereign, they're neither a person nor a corporation. They've issued a stablecoin under their own regulatory scheme, which is interesting to think about.

But what about Puerto Rico? Puerto Rico's inside the realm of the F-35s circling around, yet completely independent from the federal government. That's potentially a birthplace for this whole thing.

Let's have Alex take this home. The subquestion then would be: You have a hypothetical preference for Wyoming, maybe Puerto Rico. When does this happen? When do we see, in your prediction, Toly, the first state-level government in the United States enshrine in statute the ability for an autonomous corporation—which, by the way, could be an AI as a person as well? It doesn't necessarily have to be an on-chain entity.

I'm not 100% sure, but I thought there were DAO-friendly bills already being passed in Wyoming, specifically for creating these kinds of corporate structures that assign control, or parts of the control, of the fiduciary duties to a DAO.

Yes, there are. We looked into this. Wyoming does allow that. The problem is you're still under the corporate and federal SEC issues.

But I'll give you this: In Panama, and in a couple of other places, you can have foundations that create DAOs. Those are fully decentralized autonomous organizations, and everything can be run from the DAO. The DAO owns everything, so it's almost like a trust with digitized, crypto-enabled transactions going through it.

You actually don't need these things. I think the key part that you need out of the federal government and laws is that, if you participate in the smart contract, your liabilities are not commingled with everyone else who participates.

This either needs to be passed through Congress through the market structure bill or figured out in the courts. Once that's true, the fact that you can now participate in these DAOs and things like that effectively becomes code is law. If the government says that your liabilities are limited, and you're not part of this cohort responsible for everyone's actions in this mechanism, then the only thing that you have is the code.

You are solely relying on the code to enforce all the decisions of that thing. If you go into it and lose money, tough, right? You've actually taken on the risk, and there's nobody for you to go after anymore.

I don't think we need laws to pass to allow it. What we need is laws or courts to figure out that you don't have liabilities if you participate in these systems, and that's about it.

I think, Toly, you've put it well: They exist just by existing. You've put your finger on, I think, the core issue for the future of this entire space, which is: What does the future of law as code look like? I think there are maybe 2 directions—at least 2 directions—I can imagine it going in.

One is, I think, what you're articulating, which is that in the future, state plus federal law plus a whole bunch of regulations get encoded in something that looks like future Layer 1 code. It's basically software. There is another future where laws and regulations remain pure natural language, but we have a whole constellation of AI agents interpreting them—basically AI lawyers and AI regulators. Do you have a sense of which of those 2 futures, or maybe door number 3, we're going to find ourselves in?

I'll answer that question. We start with number 2, which is that you have AI agents dealing with natural-language law, and then, over time, we have to develop completely new forms of mechanisms and governance structures to deal with it. Well, those will be on-chain. What about you, Anatoly?

I think humans in the loop, and kind of how our human egos are designed, means that you will have—I think The Big Short has this awesome scene where the higher up the chain of command you go, the dumber the person, the less they understand about what's happening. The same applies to podcast hosts, by the way.

Because of human nature, politics, leadership, and all these things, you're going to see the most sophisticated understanding of how this stuff works at the lowest layer. Then, as it goes higher up—to the court system, to politicians, to regulators—the more broad strokes they have to control it.

My view is that I don't think you're going to have AI agents in that spot because of human ego. I think humans really, really want big organizations under them.

So, Anatoly, I just have to jump on that because I think it's the coolest thing in the world that you, as an individual engineer, came into the country 14 years ago, whatever, settled in San Francisco, and invented a hundred-billion-dollar thing out of thin air. You just created it out of thin air.

Then, to figure out how it interacts with the government, a guy, David Sacks, who has nothing to do with the government, ends up in the role kind of overnight, and now you guys figure it out together. Show me any other country on the planet that could ever do that.

But it's exactly what you said from The Big Short: you get to any other jurisdiction, and the factor in the way makes it impossible to interact. It's just the coolest story.

Hold on. Hold on. Hold on. This isn't quite fair. If you take Switzerland, which has been crypto-friendly for longer than almost anybody, they passed legislation quite a long time ago, and a huge number of the crypto founders moved there because it was so crypto-friendly.

Japan is friendly. Now there's an increasing list of countries—Malta, Liechtenstein, et cetera. Countries are jumping on board, and a lot of these—Binance, et cetera—were created in places where they were crypto-friendly first, and then they worried about the U.S. later.

The U.S. has now jumped on the bandwagon in a big way, which is fantastic, but it's a little bit too little, too late for some things. I think in other areas it's going to really take off.

You know, Anatoly, when you started Solana, agentic AI was not even in the conversation. It was 4 or 5 years out, right? Just transformers. And I can imagine that AI agents are going to be the single biggest transactors of Solana. Can you speak to that?

Yeah. I think the fact that you can synthesize a lot of signal around the world through AI means that the cost per unit of intelligence goes down. What that means is that you have an army of analysts that can understand all of the signal and synthesize it into a buy-or-sell signal, right?

You can now create a lot more markets, and I think this is where stuff like futarchy can potentially become scalable to large organizations. Why I think this is kind of happening already is, if you look at me or Vitalik, we're obviously not business people. I'm not a broad business person who understands how to do this.

You play a business person on TV.

Yeah, exactly. We're engineers.

Way too candid and honest, I think.

He's a researcher. I'm an engineer. I understand how memory works and stuff like this, but simply from this trustless coordination of blockchain, the fact that you could own a part of SOL and actually run the systems from virtually anywhere and participate in them—that actually creates enough incentive alignment for everyone in the ecosystem to move it toward a common goal.

Even if each participant is rationally optimizing their own P&L under the hood, they're maximally trying to profit, sometimes in a zero-sum way over other participants. But because it's all cryptographically glued together into this one-chain engine where, if there's a bunch of markets, it all makes money, that forces everyone to move in the same direction.

I think that is an example of this class of new organizations that are not quite corporations and not quite a single-person small business, but an internet-scalable system that's glued together with cryptography and can all move together for a common benefit. That's a very, very unique, cool thing.

As you've had AI explode, and the ability to analyze a whole bunch of signal into an action becomes cheaper and cheaper, I think the number of markets that can support this decision-making is going to blow up. Hopefully they all run on Solana, and we make more money from making sure there's no spam. But I think that's awesome.

How many AI agents do you imagine are going to be operating a decade from now?

I don't think the question's going to be meaningful a decade from now because I think “agents” is a very 2025-era term. I think we'll look back 10 years from now and laugh at the premise of the question.

I think you have a continuous information funnel to GPUs that's just constantly looping. Yeah, if the question were, “How many terawatts or petawatts of intelligence will we have a decade from now?” We currently measure AI in terms of energy, right? We talk about gigawatt data centers and so forth.

For the moment. For the moment. For the moment. Are we going to start a new layer of measurement, which is a financial layer of measurement, in terms of how much capital is going to be transacting on these systems? The most valuable bits of information are bits of information that carry financial value with them, right?

But that's circular, though. That's circular to say that what's most valuable is most financialized. Ideally—and Anatoly, keep me honest on this—I would hope 10 years from now we're in a state where we've not just solved math, but we've solved economics.

We have a rigorous science of what wealth is. We have the beginnings of it today, arguably, but if you put 10 economists in a room and ask them, “What is real wealth?” you'll get maybe 20 different answers.

I would hope that, a decade from now, we have a notion of what real wealth is—not just monetized wealth, but actual wealth—and then we could actually trade real wealth. To my knowledge, we don't know what real wealth is.

Probably just energy, just jewels. Maybe, but maybe not.

I'll give you an alternative, too, which is transistor flips or FLOPs. What? FLOP isn't right, but it's going to be some metric of compute that's more foundational than the dollar or any concept of currency.

Right now, we're all trained that wealth is dollars, and the foundation metric is dollars, but that's going to break very quickly. It'll be something either like power—electrical power—but I think more likely some metric of compute.

I'd like to challenge all of this. We've had, for the last few hundred years, the main mode of discourse in the world has been business, commerce, and money, right? We run the world on that. Then we moved away from feudal systems, and now the power is in the power of money, which is more liquid and free-flowing. It'll fund good ideas, et cetera, et cetera.

But over the last few decades, we're moving from money to information. A startup today is much more interested in collecting data about its users and then monetizing it later. Facebook maybe has done the best job of taking social data and monetizing it, and it's fungible—you can go back and forth.

Over time, though, the information becomes the harder bit, and the monetary side becomes less relevant. I would think that, over time, you'll end up in a Star Trek world where you don't have any money, you don't have any commerce, and you're just doing things.

The cost of things becomes so meaningless that it doesn't matter. Commerce—the idea of commerce—doesn't matter. So that would be my challenge.

I'll challenge the challenge and say that this very much feels like a very October 2025-era discussion, where we've seen large parts of the West deindustrialized for a couple of decades, and we're just taking it as an axiomatic truth: “Oh, well, we're deindustrialized, so of course it's just about the bits; it's not about the atoms.”

But actually, the atoms are incredibly important. I would argue that, if I had to choose my favorite denomination of real wealth, it's going to be something embedded in the physical world.

I've argued in the past for something about future freedom of action. If we could quantify the future freedom of action of humanity—in cubits or even classical bits—that's the closest I can come up with for a real definition of wealth.

But I’d be curious to know: what’s your best definition of wealth?

Okay, I’ll give a hot take. I think it’s Shannon’s law: how much information we can process. Our channel capacity to process information is real wealth. It’ll be a measure of information. Yeah, I agree it’ll be measured in units of information.

Time out. I’m taking this in a different direction. One of the things I’m concerned about right now, given the speed at which we’re getting wealth aggregation into the large hyperscalers, is that while not in the long term, in the short term I think there’s going to be job dislocation. There are a lot of people really concerned about civic unrest.

We had Balaji on the podcast talking about, “I would not want to be a tech entrepreneur in the Bay Area in 2 or 3 years.” You guys remember that? I’ve had a few of my friends in the VC world getting death threats right now. It’s crazy. I mean, really, really crazy.

My question is: how can we potentially see Solana and the crypto world help ease this tension that we’re going to have from this discontinuity? Because in the long run, totally, all I speak about and write books about is this world of abundance where we’re demonetizing and democratizing access to food, water, energy, health care, and education. It’s the interim state—the next 2 to 8 years—that I’m most concerned about. What are your thoughts there?

I actually have a very opposite view of Balaji here, probably because my parents came to the United States from the USSR, literally with $50 per person, in ’92. I think what we’re seeing with AI is a smaller transformation in terms of labor than the steam engine. The number of people who actually became unemployed because of the steam engine was dramatically greater, and those were all fighting-age men.

I think the likely outcome is that we just get better at what we do, jobs become easier and less risky, and you can scale up. The amount of wealth that you create for the world increases, poverty decreases, and people simply work less. The difference between the kind of work that my dad had to do as a civil engineer in the USSR versus here was night and day in terms of risk. That was just moving across the Atlantic Ocean into a more mature economy.

I think the world is only going to get wealthier, and we’re blessed to live in this age. I’m very optimistic. It’s the most exciting time to be alive.

OpenAI is not shrinking its headcount, and this is the most advanced AI company. They’re not hiring fewer people; they’re hiring really smart people and giving them the best tools to accelerate that company. That’s only going to happen everywhere.

I think you’re going to end up with a lot of demand for people who understand how AI works, where it makes mistakes, and can course-correct it and keep it on track. I think that’s probably going to be a super-valuable skill, and a lot of that comes from experience.

My experience with AI tools is that I can keep half an eye open watching Claude stream its code and know that it’s doing something right or wrong, whereas a junior engineer has to actually analyze every commit and everything. I think you’ll see the exact same kind of systems that you saw at the advent of databases and computers.

Everyone who wasn’t an accountant didn’t just become homeless. They’re super-smart people with a college education who have incredible access to networks and can go and retrain a lot more easily than I think everyone else can.

What’s your take on this, Salim?

Well, I think if we have a kind of breakdown in society like you’re talking about, what ends up happening is that because it’s so easy to build a crypto system today, you could create a local community DAO that circulates tokens just among themselves. You can spend those tokens on certain things and very quickly boot up a local economy.

Notice that most monetary transactions—80%—are local. You’re buying supplies, you’re getting your haircut, whatever. Not the amount of money, but the number of transactions. Crypto becomes really powerful for replacing that.

Sorry—it’s a strange example for you to use. Can you hear me?

It is a strange example for me to use, but this becomes really powerful to replace it. This is why I believe we’re going to move to very granular environments where people will create small communities that self-support. This supports Balaji’s network-state idea.

We just need to keep the internet going and everything’s fine. If that breaks down, then you have major issues. But I remember this wonderful story in the ’70s, when there was a huge central-bank strike in Ireland and no checks were getting cleared. No checks were getting cleared, and for 6 months the Irish just kept going. They kept passing checks around, saying, “When the central bank comes back online, it’ll settle up.”

Six months later, the central bank said, “Nobody’s even noticed we’re on strike. We might as well come back.” Then they cleared all the checks, and business went back to usual. Human beings are incredibly resilient at figuring things out, and I think when we get into major issues like that, we’ll figure it out.

I’m hugely biased because, to me, Toly is the most perfect example of the person you want to be. After he’s done with this podcast, he’s going to go debug some code he’s building. We’re discussing what this AI philosopher who’s never done squat over here thinks of it. Your perspective is exactly what you want, especially when you come from a foreign country.

I had the same conversation with Thomas Peterffy, who was born in Hungary in a basement as the Soviet tanks were coming in. He can’t connect with his own kids who grew up here, and they’re like, “Oh, we’re destroying the world. Dad, you’re polluting by taking a dump.” He’s like, “What are you talking about? We live in the best time in world history and in the most free and fair place in the world, and you’re not taking advantage of it.”

What Toly is doing is actually building the future platform for all of society, and it’s going to create the abundance that Peter is talking about. To me, your perspective is exactly the right one. I’m not even vaguely connected to the ultra-pessimistic perspective. It makes no sense to me, but everything you just said perfectly resonates with me. It’s exactly the right view. Anyway, that’s my rant.

Yeah. I am curious about something. We’ve just seen Kazakhstan launch a stablecoin on Solana. How do projects like that get you excited? What projects are getting you excited right now about the use of the technology you created? You must wake up in the morning and say, “Holy—that’s amazing, what someone just did with Solana.”

I think crypto is going through technology phases similar to the internet. You have the punks create the first version, then you have the hoodies scale it up, and now the suits are moving in. A lot of the things happening are driven by suits, which is great, but I’m less connected to that and more connected to the low-level experiments.

If people really want to be deep in crypto, go check out futarchy and decision markets, and how to run an online collective that makes real financial decisions but is fully market-based. That minimizes risk for investors, participants, and so on.

I think those kinds of things are really cool, and to me it’s also a physics problem. Can we run a market for the top 10 million important decisions that people make in the world every day?

Just to drill into that, that would be cool, right?

Yeah, just to drill into that for a second: what Anatoly is talking about is futarchy, which allows you to use prediction markets to do DAO governance. One of the big issues with DAOs was how you manage governance, because it’s like being in a town-hall meeting with everybody shouting loudly.

It’s not just DAOs, Salim.

No, no, that’s right. It’s not quite a prediction market. It’s literally like if Apple said, “I propose Apple build this VR device,” and I’m willing to buy your Apple stock for a higher price than it’s currently trading because I’m so bullish on the idea that the managers need to pursue.

I’m willing to increase your value if you disagree with me. So this, in fact, forces a financial stake for any decision to be made in the DAO, which is different from saying, “Let’s all vote with our shares and say, ‘Yeah, we approve this,’” or, “We do not approve it.”

You literally have to put your money where your mouth is for any decision that the DAO takes. It’s an amazing, really cool idea, and they’ve run a couple of ICOs that have gotten over $150 million in commitments, which is pretty crazy.

I asked you a question, but I didn’t hear the answer. In terms of super-exciting projects that you’re looking forward to on Solana, is there anything you can share with us?

Obviously, Solana Mobile.

Show it to us. Tell us about the Seeker.

We have 2 goals. One, because I spent most of my career at Qualcomm, it was immediately obvious to me that you can have the full cryptographic guarantees that a hardware wallet provides in the phone form factor.

In fact, all the technology to implement a TrustZone and keep your secret keys stored in an enclave was built years before Bitcoin for DRM, of all things. It was to keep people from stealing Netflix or whatever. Phones implement all this technology to prevent any kind of data snooping and maintain encryption all the way from the deepest secure enclave to the display.

So it was just obvious to me: “Why don’t we embed hardware wallets directly into the phone?” That was the idea. We can make the Seeker as good from a security perspective as a cold wallet—you get cold-wallet-level guarantees in your hardware device in terms of the kind of security that you get—but with the user experience of Apple Pay, so you can have a hot wallet that is both secure and great for consumers. That was part of it.

The other part of it is that you have these very mature companies like Google and Apple that have a 20% rake on all digital spend, at least through their ecosystems. There’s no reason for it; it’s just a bug in capitalism that doesn’t make sense. Why isn’t it converging down to the lowest cost? To me, it’s an opportunity to use crypto as a wedge because developers don’t want to pay these fees, so we can use different economics where you pay, we can earn, I think, as good of returns as Google or Apple, but through effectively trading transaction fees that are volume-based.

Is there a prediction market? Is there a Solana YouTube replacement, like an Odyssey equivalent on Solana, kind of?

There are people trying to build crypto-based content creation. None of those, I think, have really taken off or proven out. I think that’s a harder problem than software and software experiences. The sheer amount of content that you need to generate to get something within a 10% margin of error as entertaining as YouTube or TikTok is just astronomical.

But I think with AI and Sora, you’re going to see that kind of change. It’s just—

That’s actually the crux of my question. Does Sora 2, or whatever is next, change the math or not? Because, yeah, you’re totally right.

You can see it in 5 years: this entertainment stream that I’m getting—the infinite stream of entertainment—is going to be AI-generated. That’s kind of a scary thought, because I think humans in the loop in what keeps us entertained is probably an important part of our shared cultural experience, but we’ll see what happens.

An amazing future ahead. Gentlemen, I’m jealous of your Seeker, Salim. I’m starting to move all my crypto onto it. Toly, my whole community wants to move our NFT collection onto Solana.

Oh, awesome.

I’ll ping you about that. Any closing thoughts, Salim?

And do I get one?

Yeah, please. Yeah, go for it. Of course you do.

I have one that I think the audience might really care about. So, Toly, if the world moves all these transactions to Solana—you know, because it’s 1,000 times faster than Ethereum, which is 1,000 times faster than Bitcoin—so it’s like the engine for AI. The fundamental use of the Solana token is anti-spam, right? It’s a way to pay for your transaction to get settled.

Is there any way to translate the value of Solana, without giving investment advice, from the transaction volume of Solana to the value of the Solana token? What’s the math that connects those 2 things?

I think the transaction volume is less important, and what’s more important is the opportunity cost of being late. The network actually makes more value. As a block producer, you need SOL to stake to be a block producer, because if you had no Sybil resistance, you could create infinite blocks, and that would effectively spam the network to death.

So you have some Sybil mechanism to prevent infinite block producers. You have some percentage of SOL that you stake that gives you X amount of the percentage of blocks you can make. When you’re making blocks, people are paying you to be first in the block, to be first to get access to this trade. The amount that they’re willing to pay you is based on the opportunity cost of being first.

Right. Right. Has anyone put together a blog or a white paper or something that explains what the math is?

Yeah, Blockworks probably has done the best job in terms of compiling all of this data and analyzing it into more traditional categories: this is revenue, and these are network costs. If you go to Blockworks, you can look at Solana and Ethereum and do comparisons.

This is true for all proof-of-stake networks, because the Sybil mechanism for proof-of-stake is the token itself. If you have X percentage stake, you can literally ask: Do I put a percentage of my portfolio into Treasury bills that are risk-free, or do I risk some of it to run a block producer and get block tips, effectively, for including transactions?

It doesn’t actually matter that you’re getting tips in SOL or USDC, or somebody gives you a sack of potatoes, because the fact that you have to stake X amount to get access to that revenue is how you can do your Kelly-optimized allocation. This ties to The Intelligent Investor. It’s a very traditional, boring approach to doing analysis here.

You can’t do this for Bitcoin, because the Sybil mechanism for Bitcoin is energy, and I don’t know if it’ll ever change or anything like that. I think Bitcoin is its own special snowflake. I’ve struggled to come up with a standard model to analyze it, but for proof-of-stake networks, I think you can really put them in The Intelligent Investor box, do analysis, and make your own decisions.

All right. Alex, I want to give you a chance to ask a question. Then I have a question, and I’ll go around the table to ask everybody. I’ll ask a fun question, not a serious question.

So, Toly, let’s project forward. Say humanity does, in the end, take apart our solar system to build the Dyson swarm, and we have lots of computronium. What will be the medium of commerce in a Dyson-swarm future for humanity? Do you think this is kind of like the central-planning communism problem? Is it computationally feasible to solve it just mathematically without markets?

At relatively high latencies, right? You’re bound by light-speed latencies just like we all are. I think this is the question: How many qubits do we have to be able to solve this massive linear-algebra problem? Can you allocate the resources to everything?

You might not need commerce. That might be the end of capitalism, and you might only have it simply for human entertainment. But I think personal freedoms are far more important than efficiency in a lot of ways, so I would go against it. I would be very much against it.

I think it’s very important for people to have purpose and compete for tokens, whatever they are—bananas, a meme coin, or whatever.

Here’s my question for the group. I’m going to follow on Alex’s question from earlier, which is: What’s your definition of wealth in the future? Salim, what do you think? It’s very different looking back historically at the kings and queens and pharaohs. It was how many slaves you owned and your ability to have access to agriculture. That was sort of wealth. What do you think it is in the future here?

I would think it’s a combination of time and health span.

Time and health span. Okay. Dave?

I think it’s a no-brainer that it’s purely tied to compute. I was asking a class at MIT the other day: If I offered you $10,000 cash—here it is—or a GPU that’s worth $30,000, how many of you would take the GPU? They’re like, “Are you crazy? I’ll take the $10,000 cash.”

But in the near-term future, compute can be immediately turned into cash. Compute is the universal thing. When you have AI agents who are the laborers of the world, the compute—your number of workers—is the amount of compute that you have.

Your ability to make yourself happy, whether it’s controlling your Figure robot, cleaning your house, or building something virtual, is all bounded by the amount of compute you have access to. That becomes the universe. It also determines your health. If you put your compute toward analyzing your scans, it determines whether or not it finds your cancer. So it becomes universal.

All right. Compute is your answer, Alex?

I think we’re suffering from the cliché of the blind philosophers who are touching different parts of the elephant and are all overconfident that the part of the elephant they’re feeling is what an elephant feels like.

I’d argue for a more general definition that generalizes all of those definitions. I would argue that real wealth will be measured, to first order, as future freedom of action, which generalizes compute and generalizes physical resources.

It can be measured in units of bits, so it’s an information-theoretic definition, but it’s not just about compute. It’s about the ability—some might call it empowerment, but that’s a specialized term—to take the course of action you want in the future, not just in the present.

I knew I should not have asked Alex to say that to me. I didn’t get it. But if you go to alexwg.org and read his paper on the topic closely, you’ll come away saying, “Oh my God, he’s totally right.”

Yeah, I knew I should have gone with what I was going to say: degrees of freedom.

Yeah. Degrees of freedom. Yeah, it’s the same thing.

Yeah, you guys—I should have gone before Alex went. I’ll answer, and then I think it’s the ability to fulfill your desires, your purpose. Compute is part of it, but nanotechnology is going to be fundamental as well—a manipulation of the physical universe. So it’s not just compute in that regard.

And you’re going to go thinking in scarcity terms. You’ve got to jump forward to abundance terms and think that all that matters is time and health span.

Hello? I think, Salim, you’re thinking in meat-body terms. Think in post-biological terms. Come on.

All right. Anyway, totally. So, Toly, your answer here?

Degrees of freedom.

Degrees of freedom. All right. Also, I think people will never feel satisfied because there’s somebody else who has more degrees of freedom. This is the human condition: always striving for something else. The hedonic treadmill spins faster and faster, doesn’t it?

Yep.

Yeah, I think you got that right, because you can have a lot of compute and still be beaten like a dog every day by some government. So you guys are right.

All right, we’ve reached a conclusion here. Toly, where do people find you on the worldwide web?

On X, @aeyakovenko. Follow me on X. I have hot takes, sometimes boring takes. I don’t know.

Love it. Love it. Yeah, thank you for the work that you’re doing. Thank you for the future that you’re enabling for so many globally around the world. Grateful. The only time we’re excited in today is tomorrow, and it’s going to be a doozy of a decade ahead. Moon shot may So love you all. Thank you for today.

Every week my team and I study the top 10 technology metatrends that will transform industries over the decade ahead. I cover trends ranging from humanoid robotics, AGI, and quantum computing to transport, energy, longevity, and more. There's no fluff, only the most important stuff that matters that impacts our lives, our companies, and our careers. If you want me to share these metatrends with you, I write a newsletter twice a week sending it out as a short 2-minute read via email. And if you want to discover the most important metatrends 10 years before anyone else, this report's for you. Readers include founders and CEOs from the world's most disruptive companies and entrepreneurs building the world's most disruptive tech. It's not for you if you don't want to be informed about what's coming, why it matters, and how you can benefit from it. To subscribe for free, go to diamandis.com/metatrends to gain access to the trends 10 years before anyone else. All right, now back to this episode. [Music]