加密货币正在重建金融系统——机会遍地

Yeah, Bryan. I mean, it's just crazy. The world has changed so much since we first got into crypto, and so has LayerZero.

The difference between our early days of BD and doing BD with SushiSwap to now spending 100% of my time with Fidelity, with BlackRock, with NYSE, with all these different groups—it's just night and day different. Totally different world. It's been really fun and enjoyable.

I mean, it's a sign that we've made it.

For sure.

Guys, welcome back to a very special episode of ThousandX. We have the founder and head of LayerZero, Bryan, on the pod today. Thank you for joining.

Of course. Thank you so much for having me.

There's been so much discussion recently about the tokenization of assets and the new frontier of crypto. We're sort of entering this era where things are actually happening. I've followed you for quite some time, Bryan, and I know that you've been a builder in this space and that you're pushing forward on LayerZero.

1. Poker, Bitcoin, And Selling Models To Billy Beane

But my first question, for the audience that doesn't know, is: What is LayerZero, and how did it even start? Tell us where you came from, how you found crypto, and then how you decided to build LayerZero.

Sure. Yeah. Where I came from is always a bit interesting. I grew up in a town of 900 people in rural New Hampshire—as remote as you can get. My grandmother had a 100-acre farm. Fortunately, I got a computer when I was young, so I was just a nerd with a computer for most of my life. I went to school for computer science.

I ended up dropping out to play poker professionally, which was a very unpopular decision in my household. But it worked out pretty well. I did that full-time for 8 years, traveled to 80 countries, and did all that.

Then in 2011, online poker got banned in the United States. So I found Bitcoin super early in 2011, because that was how all of us in poker woke up one morning. There were DOJ seals on every website: your career is gone, all your money is frozen. It was just over immediately.

That gives you a healthy skepticism of government, which I think was probably a good thing and definitely aligned with the early crypto ethos. And, 2, everybody within poker started using Bitcoin as a way to deposit to any of the sites that remained.

So I found crypto super early. I started my first company in 2011, sold it in 2013, and went all-in on crypto in 2013–2014. I had racks of miners in my garage mining Bitcoin. I was completely obsessed.

Then, at the end of 2014 and into 2015, the whole industry just died. So I saw DeepMind's Atari agent demo and got super motivated by reinforcement learning. I built a bunch of machine-learning models for fun. I showed them to a friend, and he said, "Oh, I have this group of MIT PhDs who worked on something similar."

I'm a college dropout, so that's pretty intimidating, but I said, "Sure." I showed them what I built, and they were like, "This is amazing. We have somebody you have to talk to." When I got on the phone, that person was Billy Beane—the founder of Moneyball and general manager of the Oakland A's.

I ended up selling those models to a bunch of the professional baseball teams.

What did those models actually do?

They were primarily predictive models. What I focused on was pitcher-versus-batter matchups in baseball. It was basically taking this really unique data set—because baseball gives you about 50 data points on every pitch—and trying to find when a pitcher's release point and all of these different variables would start to degrade before you would visually know it, and when you should actually take out your pitcher.

Then you have to weigh the state of the game and the current skill of the pitcher against the bullpen and your upcoming schedule. You want to optimize for wins overall. It was interesting and super fun.

Wait, that's really cool. Do you know if any of your models are still being used?

I don't know if they're still being used. No. But I know the one team we worked really closely with went from the bottom of the league to taking 2nd the next year, losing the World Series.

Oh, wow. Are you a big baseball fan? I assume you are.

I wasn't. I actually hated the sport. But it had such a rich data set. It seemed like a good and interesting way to apply it.

Well, as a huge baseball fan myself, that kind of hurts to hear, but that's amazing.

It's grown on me a little bit over time. At the time, I would have loved to be doing football or basketball or something else, but the data was so rich. For the NBA, if you wanted data, it was like $500,000 a year to get SportVU or any of the player-tracking data, and MLB was just free. It was free and open, which made it an awesome entry point.

Yeah, it's honestly the best sport in the world for stats nerds. It's crazy, for sure, because all the people involved in baseball are very forward-thinking as well.

There's also a really long, published history of state-of-the-art models along the way, so it's very easy to benchmark yourself. It's much harder in most other sports to benchmark where you actually are in the prediction space.

So that was super interesting. I did that from 2015 into 2016. I took 100% of my net worth and put it into Bitcoin in mid-to-late 2016. I started another company, then sold that company 2 years later for $50-ish million, and then basically just built for 4 years, trying and failing at a bunch of stuff.

I got fortunate because Bitcoin went from $700 to $20,000 during that period, and all of this stuff was great. We were just building and trying to figure out what we were doing.

We ended up doing AI research, basically, and showed it to Noam Brown. Noam, at the time, was a really amazing guy. He had left the Fed, was doing his Carnegie Mellon PhD, and had just won the science breakthrough of the year for his work.

He was really impressed with what we'd done, and we co-published with Noam. We got cited by DeepMind and had this really cool path on the AI research side. Again, as a college dropout, DeepMind citing my work later was awesome.

Noam has gone on now. He's the inventor of test-time versus train-time trade-offs, the author of o1 and o3 at OpenAI, and one of the world's leading AI researchers.

2. MEV, Broken Bridges, And Building LayerZero

But we did that and then basically started messing around. We got a bunch of people asking us to make trading bots between the first early DEXs. All this was happening at the same time. As we were doing that, we realized nobody had any clue what they were doing.

So we started doing MEV. This was me and my 2 co-founders, whom I brought out here. We were building a bunch of technology through a bunch of different lenses. It was MEV before MEV was really a thing.

You were just trying to submit transactions as fast as humanly possible into blocks. One day, you started to notice that people were ahead of you in positioning with zero gas paid. It was like, "Oh, actually, the validators themselves are now colluding against us, and there's this whole structure happening."

We were like, "All right, well, that's out the window. We don't know how to compete with that."

How long did it take before that got competitive? How long were you running that?

There were 3 groups. I know one of the groups actually ended up being a cohort of Jump Trading. There were 3 groups that were really competitive at the time, and then it got into the cabal, and they got formalized into MEV.

So, about 1.5 years on that. Then we started seeing more volume. We started seeing Binance Smart Chain launch, and it had more volume and more users than Ethereum.

That was really uncommon. Having been in crypto from so early, you saw everything launch in 2013 and 2014, and everything launched in early 2017. It was all vapor. Nothing got used. Nobody actually used any of this stuff.

That was really interesting, and we started asking ourselves, "What do you do with an environment that's fast and cheap but you don't really trust?"

You treat it entirely as ephemeral, and then could you roll the result back to Ethereum—sort of keep state there? There were rollups, pre-rollups, and we just tinkered. We built a toy game for ourselves to mess around.

As we were building, we realized there was no way to actually trigger events between the 2 chains. We just assumed it was a solved problem. So we looked at bridges and were really shocked. We would never put millions, let alone billions, of dollars into the state of the bridging world back then.

And so then we tried to make a better bridge. We thought, “Okay, this is a solvable problem. Let’s just do this.” We were going to do it entirely anonymously and just launch it completely randomly.

As we were building, we realized there was no way to trigger events across the chain. It wasn’t until then that we realized the generalizable problem was actually communication between these two things. You need to be able to build anything on top.

I guess all of this is a long-winded answer to, “What does LayerZero do?” LayerZero is the communication layer between chains. That’s when we formalized building LayerZero, and then building on top of this became a question of how you normalize a protocol for how chains talk to each other.

Once you have that fabric, and you go from an isolated execution environment to a broader internet of value, or connected chains, what can you do with that? Fast-forward to today: We’ve got about $150 billion worth of assets built on top of us. We just crossed $300 billion moved over our lifetime.

We’re moving roughly $10 billion to $15 billion a month, which is more than Western Union, more than Wise, and more than most of the traditional payment processors, as you would think about them. It’s started to become really meaningful, and it’s been a fun, interesting journey for sure.

Yeah. So BNB itself was an EVM-compatible chain, right? There were a lot of these other EVM-compatible chains coming out at the time, but are you able to also work with things that aren’t built on the EVM? I mean, you can move from Ethereum to Solana, from Solana to somewhere else. I guess that’s a tough technical problem, right?

Very much so. Anything that has enough of a virtual machine and enough smart-contract logic, you can work with. We’re on a lot of non-EVMs today: Solana, Aptos, Sui, TON, Canton, and all of these networks.

Where you run into problems is with something like Bitcoin. It doesn’t have enough smart-contract logic for you to allow applications to do arbitrary things. You’d have to end up writing a static validator set. You’d become more like a wrapped Bitcoin or something. You could do that, but for us, one of the interesting things that we built—which is very unpopular, or even now, very nonstandard—is that the entire protocol is completely immutable.

We can’t change it, even if we wanted to. We put up the largest bug bounty in the world when we launched it—a $15 million bug bounty—and it has lived ever since. The non-EVMs are really tricky. EVMs are easy now, right? You have the thing, and it’s like, “Okay, you just reproduce it.”

Non-EVMs are a really big task because it’s “measure twice, cut once.” It has to be absolutely perfect. So yes, we do support them. I think we cover about 170 chains today, including most major networks, but each new non-EVM is still a really big undertaking.

Right. Of those—I mean, 170 chains is a lot—but I don’t think any individual is transacting over more than 5 to 10 chains, unless you’re a crazy power user or a trader. Where’s most of the money flowing right now?

Most of it is still in EVMs, mostly within the L2 ecosystem. What we see is that we’ve moved a lot from being very crypto-native early on, and now we’re working with basically all of the largest institutions in the world.

Ethereum and Solana are the default. Everybody wants to be on those two. After that, Arbitrum comes very closely behind, and then you have a broader ecosystem that expands out from there in concentric circles.

It’s still largely concentrated there. I haven’t checked recently, but last time I checked, about 50% of our volume still came just within Ethereum and the L2 ecosystem. So it’s still primarily there.

That’s the core premise of LayerZero, then: You believe—I mean, maybe you don’t, but I’ll just say you believe—that the future is a ton of different chains potentially transacting in areas that are specialized for them, or that have specific use cases for the companies that want to use them. Your goal is to build a platform that, in the future, will connect all of these regardless. Is that accurate, or do you think we’re going to converge on a few chains in the future?

There’s always going to be more than one. For us, it didn’t matter if it was 2 or 2,000. You still need the connective tissue.

Our big belief was that general computing systems are not how you do anything. If you build modern software today, you have microservices for everything. You’re hyper-optimizing on every possible vector to try to get the most performance out of whatever it is that you’re doing.

Our strong belief is that you’re going to have environments that make really strong trade-offs on a given vector, and that a generalized system can’t compete around that. I think you see that even today; this happens all the time. That thesis has really held true.

I do think there will be—I’m more confident now—maybe 5 to 30, not 5,000, long term, unless they look different, at least in general environments. I think there will be some consolidation there.

The biggest thesis was that there were going to be multiple environments, assets were going to want to be everywhere, and you needed to provide a way for that to become seamless. If it isn’t seamless, none of this is going to work at scale.

If you remember stablecoins from about 5 years ago, when they were emerging, you would go to any chain and there would be 10 versions of USDT or USDC. There was a Wormhole-wrapped version, an Axelar-wrapped version, and all of these different versions of the same stablecoin.

I think you’ve seen consolidation. You go to a chain, and it’s just USDT or USDC. There’s a canonical version for that chain and that asset.

I think you’re going to see the same thing across tokenized equities and every other piece of tokenization. You need all of this to play nicely because you don’t want to have tokenized stock version 1 and not be able to move it over here, while another version uses a different standard. It’s just a total nightmare.

So I think we have very strong conviction in that consolidation.

Were you involved in those discussions? The whole idea of standardization is quite difficult. You have to coordinate a ton of different actors and get them to agree on something. I assume everybody wants to get there, but what were those conversations like, if you were involved with them?

3. Winning The Standards War: Tether, PayPal, Ondo

For us, it was never really about that. I find it very interesting. We have 90% to 91% market share right now for interoperability, so we’ve sort of consolidated around OFT, this token standard.

Along the way, there were 10 different working groups of industry participants trying to create standards. Everyone was fighting against this along the way. Our view has always been that you want to work with the absolute best partners in the world, go super deep, and help them build something that meaningfully moves their business forward. If you do that, we’ve found that you just win over time.

Tether was a really big one for us and drove a massive amount of consolidation. Tether, Ethena, Ether.fi—there are all of these assets that we started to work with. They really fueled growth in volume and acceptance of all of these things.

From there, that became a point of leverage in every ecosystem. Now you just can’t have competing standards. It adds too much friction over time.

I think the industry itself has pushed it forward, which is my preferred way. My very first job ever was working on IETF standards for voice-over-IP systems. After dealing with that for a very long time, you don’t want to be there if you don’t have to be.

Standardization is great, but I think it’s much better when it can come from a commercial lens and force the fold, versus trying to spend a year to multiple years getting the industry to consolidate on something ahead of time.

I’m curious. You obviously have to work with a ton of different teams—you’re a focal point here. I’m curious if you’re willing to share which teams have been the best to work with, or which integrations you’ve enjoyed the most.

Yeah. For us, it always— I think the risk you run is getting trapped in POC hell forever. What you don't want is for everybody to be loosely interested. Sometimes they're interested just because they want to talk to you, or they want to learn, but they don't actually want to do something. Or they don't know what they want to do, and they're like, “Tell us what we should be building.”

I think the best people are the people who very intentionally know what they want to do. Tether is a great example because, at the time, they had saturated all of the primary chains. You're talking about $160 billion to $170 billion. They were on every major chain, and they thought there was very little value in expanding across the network.

Lorenzo and these guys—we pulled together for USDT0—when the idea was, “Hey, as they're thinking about launching this, what if you just do some of the longer-tail chain stuff that you're not thinking about going to right now, that you don't think has any immense value? Let's build that together and see what that would look like?”

From launch, within the first year, they grew AUM by about $10 billion, right? And $10 billion of AUM growth just on the chains you thought had no value—hundreds of millions of dollars directly to the bottom line. I think that became a really strong proving point. Then you started to see other legacy versions. The USDT that was on Arbitrum got migrated to USDT0, and you've seen this happen more and more.

PayPal's PYUSD is built on top of us. They were just on Ethereum and Solana, but disconnected, and there were about $300 million of AUM. Now they're at $4 billion or $5 billion, and they've connected the 2 chains, expanded the networks, et cetera. I think what you want is people with a very clear vision of what they want to do. Then it's very easy for us to show, “Here's how you can accomplish that,” and look at the business outcomes that matter on the other side.

For us, it's typically been about going really deep. With ether.fi, we built sync pools for them. They had this problem where they wanted to be able to restake on the Layer 2s, but basically, you would have this 7-day window to come back and stake because you're missing yield for 7 days. You have all this issue of just how to get this to happen for all the ETH that is living there, so we built sync pools for them to be able to do this.

We've worked really closely with a bunch of teams, and the ones we have are doing billions to many billions of dollars a month. The teams that are the best to work with are the ones who want to go the deepest. So, Frax, USDT, Ethena, ether.fi, PayPal, all the Paxos assets, USDG on Robinhood—it has been amazing. We built all their tokenized equities, and we helped build Nexus for them.

Really, you'll find the teams who want to lean in and actually build stuff. How much it's used in production is night and day versus everybody else who's just like, “We want to launch something. Let's just see how it goes.”

That's good. Thank you for not saying USDC. It's going to help out my Circle short. But I won't make you comment on that.

I'm curious—you guys are obviously building out a ton of new stuff, which I want to get into, like Atlas and the new blockchain that you guys are launching. But before I get there, because you're focused—maybe you're focused—on building new products, do you feel like the bridging protocol is in a good place right now? Are you going to continue to try to improve it, or is it just like, “This is good. We've actually reached a point where we have a very usable product. We're in a really good place, and now we can turn our attention elsewhere”?

For example, one person commented that Ethereum transactions still take 3 minutes to complete on LayerZero. Are there areas to improve, or is it like, “We're on the blockchain; we're focused on our settlement product now”?

Yeah. For us, we're totally agnostic as a technology. Ethereum transactions can take 1 second to complete, but you're taking some finality risk. It's very easy to integrate; every asset issuer basically defines this for themselves. You'll see some asset issuers, when you're leaving certain chains, take 30 minutes, and others take 30 seconds. Ultimately, you're pricing finality risk.

There's always stuff evolving in how you do verification. ZK has been a huge area of research around this, as well as how you deal with finality on certain chains and guarantee stronger sets of finality. All that stuff is evolving, but if you look at the core messaging protocol, it's pretty set in stone. We're at more than 90% market share, and we're moving more than $10 billion a month. It's going to continue to improve, but I think what people want now from the product side is different.

You see this with intents, which are getting a lot more volume these days in general. How do you batch that, take a secure message, and basically say, “I'm just going to price the finality risk. I'm going to deliver this in 400 milliseconds on the other chain. I'm just going to price it in”?

LayerZero is built for single messages. You can move $800 million, almost $1 billion, in a single transaction. It's incredibly efficient to do that. But if you want to go from USDT on Chain A, swap into an asset on Chain B, and do that incredibly quickly, a bunch of the infrastructure isn't structured for that.

The products that our customers are looking for are the ability to do swaps in size—$50 million to hundreds of millions of dollars—cross-asset swaps across stablecoins and across a bunch of different things. That's a big piece of it. How do you solve the regulatory side? We had to go through NYDFS approval for PayPal to be able to do that. How are you able to provide that to a bunch of these assets?

The third is that you see groups like FOMO doing this incredibly well in revenue. It's just very quick: any-asset swaps, being able to land transactions and submit and provide fast and immediate transaction sets to the customer base downstream.

All of those things from the product side are things we think about a lot. How can you give better execution? How can you structure liquidity or execution across multiple assets? How can you actually do that in size, and what does it look like? It's very easy to do it for a $1,000 customer order. It's much harder to do it for $100 million slugs between stablecoins.

We spend a lot of time thinking about all of this stuff, but that's really about how you provide better service to the customer base building on top of us. The pure messaging is pretty solid.

I mean, crypto is so much easier to use today than it was even 2 years ago. It's amazing. You mentioned FOMO. There are all these other apps coming out that let you transact cross-chain. Even when you open MetaMask today versus 4 years ago, you get to see all of your assets across chains.

I remember before, you had to click through and change the RPC to make sure your assets were still in the right places. It was such a big pain in the ass to even track your assets across chains, let alone actually transact with them.

Bryan, I have so many questions for you, but one thing that stood out to me was this concept of pricing finality. That leads me to believe—has there ever been a problem that you've run into with finality? Maybe you can tell us a story there.

People like me, for example, and a few others, often think this is silly. Nothing bad is going to happen anyway. Just let the asset move over. Ethereum's not going to roll back the transaction. The assets are going to be fine. Who cares if it takes 3 minutes? Just do it in 30 seconds, and the finality will come. But have you ever run into an actual issue with it?

4. When Chains Roll Back

Yeah. It's getting a lot better over time. Reorgs are generally much less frequent and much smaller, but there was a time when Polygon reorged 200-plus blocks every other day. We've seen ZK chains roll back hundreds of thousands of blocks. There was just a really large reorg on one of the EVM chains that had a security incident. So—

Can we get into that? What happened there? I think that's interesting.

Yeah. One of the large DeFi protocols got hacked on Cronos. Basically, this is something you see happen—it's happened a bunch of times—but they had to make the decision: Do you allow the hack to exist, or do you roll the chain back to a point pre-hack?

Obviously, there are a bunch of downstream implications when you think about what assets got bridged off and what all of that looks like, because when you're rolling this back, you've already committed transactions on the other side.

So, it definitely happens. Luckily, now most of the highest-volume things being transacted are stablecoins, and stablecoins have a layer of enforcement. They can freeze assets; they can do things that you can’t do with totally permissionless assets. I think there has been some ability to rectify some of the issues, freeze seized funds as they hit other chains, and, before they hit other chains, blacklist certain wallets and do all of these things.

I do think we probably see a chain have a meaningful rollback every 3 to 4 months. So, it definitely still matters, and it definitely still happens, and people are still pricing in bypassing that to provide a better customer experience. When you move to a world where Aptos and a bunch of other projects are starting to do some of this, you’re basically shifting the risk from the protocol or the asset as a whole to a specific counterparty—a solver or person filling the thing on the other side who’s bearing the risk.

That makes a ton of sense. I mean, it’s good. You guys have done a reasonably good job at avoiding some of these pitfalls. Obviously, avoiding all of them is quite difficult in crypto. You guys are running billions of dollars of volume through your protocol, but what people really care about today is: are you making money?

You guys have a token associated with the protocol. All of the top assets right now, I think, have a path to making money for token holders or shareholders, or however you want to call them. What is your plan for making sure that ZRO maintains value?

Yep. I think this is worth having the conversation after we talk about Atlas and Zero, because I think it’ll all converge together. So, let’s definitely do it, and I’ll walk you through all of the pieces of that stack and where it all comes together.

Okay. So, why don’t we just get into this now? We talked about the cross-chain protocol. Originally, you mentioned that you were building a blockchain itself. You’re adding to the thousands of chains that exist. I’m curious why you decided to do that instead of just allowing other people to connect other blockchains. Why build your own?

Yeah, so we didn’t want to. I will say, we definitely had no intention of setting out to launch a blockchain—not even on our radar. Almost 4 years ago now, my co-founder and our CTO, Ryan Zarick, was really upset with the state of Ethereum and the state of Layer 2s in general, and the fact that Ethereum’s roadmap was pushing everything to Layer 2s. Basically, all of them were just these upgradable contracts that were positioned in a very different way.

5. Two Million Transactions Per Second

He was literally complaining to me every day, and I was so sick of it. I was like, “All right, dude, you’re so freaking smart. What would you do?” He drew out this architecture, and we debated it for multiple days: What about this? What about this? How can I break it? How can I break it? At the end, I was completely sold. It was very clear that this was the way everything was going to play out, and this was the end state for how a bunch of this stuff would look.

There were a bunch of things we were directionally right about. The things we were wrong about were that we thought ZK would be off the shelf. We thought ZK was going to be totally commoditized, that the technology was going to move super fast, and that you would just be able to pull it off the shelf. That was not at all the case. Even now, we just announced everything we’re doing, and I think Ethereum’s roadmap for doing even close to the same thing is 10 to 15 years away. We’re just so far away from what we thought it would be.

I would say the first couple of years were just us having this picture of what we wanted to do. Then we had this breakthrough with QMDB, or Quick Merkle Database. Our chief architect went on paternity leave—we had a baby—and he had some time. He was obsessing over this problem, and we came up with this new database structure, QMDB.

It was many, many orders of magnitude beyond the state of the art at the time. When we had this, the question was, “Okay, now this becomes very interesting.” Downstream from that, we published a bunch of different papers. The CliffsNotes of what we got at the end of doing this were that we thought maybe 10,000 transactions per second was something you could do. We ended up at around 2 million transactions per second, or even 1 million transactions per second with all of the overhead of the EVM. We were blown away by that.

I would say there were 4 major breakthroughs along the way. That’s when we really started getting serious. We found quirks in almost every major non-EVM we’ve ever integrated. I think we’ve gotten deeper into the actual VMs of every other blockchain and every other approach than almost any other group in the world. We actually needed to integrate the endpoint into all of this, understand the trade-offs that people are making, and understand how to do this.

We got to this system and said, “This is actually really amazing.” QMDB is now being rolled out in every major high-performance system in the world. Tempo is rolling this out through Commonware. Anybody who’s trying to do performant blockchains is effectively using this structure that we invented.

Then it became, “Okay, what do you actually do with Uniswap?” Uniswap certainly doesn’t need 2 million transactions per second. That’s not really helpful. We started going down this path with a bunch of our partners about what actually needs this. What really needs it is finance and payments, and those are the only 2 things that matter.

When we talk about Zero and what we’re launching, we’re launching a ZK-first blockchain built from the ground up—probably the first one truly built from the ground up in a very, very long time—at multiple millions of transactions per second and incredibly low latency. On top of that, the only things we really care about are finance: markets and payments. That’s it.

On the market side, we’re building Atlas, which we’ve announced. We announced the blockchain and some of our big partners.

And Atlas is built on top of the ZRO blockchain, but does the protocol that you built before—the messaging protocol—have any connection to ZRO, to the Zero blockchain, other than just connecting it to the rest of the world? So, is it just connecting to the rest of the world, and then it’s the interface for how you communicate within the blockchain?

One of the unique things about Zero is that, on top of it, if you think about Ethereum, there’s a beacon chain that exists, and there’s a bunch of globally distributed validators. On top of that, they can support a single application, and that application is the EVM.

The EVM is the application that is rolling state down to the beacon chain, which is holding state. With Zero, because of the ZK stuff, which is really just compression, and SV, which is this other algorithm that we’ve invented, you can basically now hold multiple of these things.

Rather than having one general-purpose environment, you can build something completely from the ground up. When you talk about Atlas, Atlas is a trading environment meant to scale to many millions of transactions per second at roughly 10-millisecond block times. That basically gets rolled down to Zero.

LayerZero, the protocol, is just connecting it to the rest of the world. All the assets will be there. The interface is how each of these zones or environments communicate with each other, but otherwise they’re not totally connected.

6. Atlas, And TradFi's $100M Checks

Okay, understood. I’m trying to wrap my head around Atlas itself. First of all, who’s the end user of Atlas? Who’s actually interacting with it, and what’s getting done? Would Hyperliquid be built on top of it? Is Atlas itself a system that you’re going to be trading on?

Yeah, Atlas itself is the exchange. It is the settlement, clearing, matching, and risk-credit layer, all in a single stack. Atlas is the environment for how trades are ingested, how they’re executed, how they’re held, and all of these things.

On top of that, we’ve structured it so that we’re not going to have our own front end. We’re not launching a public-facing exchange, but what this became is technology. The partners we announced on February 10 are DTCC, ICE—the parent company of the New York Stock Exchange—Citadel, and so on. They’re really TradFi-heavy partners.

For us, we see there being 2 lenses on top of Atlas: one being open Atlas publicly to the world.

And so this is open and permissionless markets, broadly available to anybody. Then you have institutional markets, which are really—as you think about many of the largest exchanges in the United States, as you think about tokenization as a trend, as you think about this trend happening—access-controlled. It's going to be permissioned. It's typically KYC'd. It's being opened up to a set of registered broker-dealers, although those rules have just changed slightly, so now it can be opened up to a broader set.

If you want to think about it as domestic and non-domestic, or open and institutional, that's how you have this split emerging. But what's going to happen is, hypothetically, you'd have ICE launch basically ICE Markets. They would launch this on top of Atlas itself.

Okay, understood. I guess one question is: What did you mean when you said you have partnerships? What were those conversations? Did they say that they're actually going to be building on top of the platform, or is it just that they're invested?

Yep. Some of this we've announced, and some of this we haven't announced. I'll be as specific as I can without breaching anything. We've talked about all the announcements. A bunch of groups invested, sort of, $50-ish million—anywhere from $10 million to $100 million, basically. So, they invested.

There were so many that put in $100 million into this.

So, yes. Then $50 million is the next largest, then $30 million. So, yes.

Okay, okay. Does the name rhyme with “men”?

Some of this we've announced publicly, and some of it we have not announced publicly. Really, really large institutions are putting in very meaningful checks and buying tokens. Some had a token-and-equity split, but primarily tokens for the majority of them. And then the question is: What did we want with all of these partners? One thing I think we really focused on—

Are these tokens locked?

Yes, for almost all of them. Some of them have vesting schedules, and some of them are locked for a certain period of time. So it depends on the group.

So what did we want? One thing we knew was that we're technologists at heart. I don't have a background in finance; I haven't spent my entire life in finance. When we were building LayerZero, the best example is probably when we were raising our rounds. We raised $225,000 from friends and family. We raised $6.4 million at a $50 million valuation, and our only focus in that round was, “I want the largest or second-largest stakeholder of every major DeFi protocol.”

I wanted the people who could influence the conversations that we needed to have. Then we raised at $1 billion and $3 billion. In those rounds, I was very clear: They were co-led by a16z and Sequoia, and I said, “Listen, I don't want you to tell me how to build. We know exactly what we're building and how to build it. What I don't know is how to build a world-class company.”

I think that's really paid off in aligning with the groups that have the domain knowledge and aligning with the best partners. For us, what we wanted out of this is that we know the core technology can do something for markets that can't be done right now—in terms of latency profile, throughput, everything. You're talking about going from thousands or 10,000 transactions per second to millions of transactions per second at a completely different latency profile.

What we don't know is how all of you are thinking about this, how you're actually bringing it to production, and so on. That's what we wanted. We wanted the knowledge from Citadel, from DTCC, and so on.

What these groups wanted primarily was to understand where all of this lives. That's evolved right now. What does the landscape actually look like? Are perps coming to the U.S. as an institutional product? How does this evolve on the tokenization side? What does it actually mean to have tokenized trading? Is this trend going to continue?

Obviously, in the 18 months since we started a lot of these conversations, I don't think anybody questions that anymore, in terms of the volumes that are actually being done for a lot of this stuff relative to the broader commodities pool. And so that was the initial group.

We have 4 or 5 of the largest exchanges in the world actively deployed in production, testing on testnet, and rolling out toward a real production release. You will see—and there is still a bunch of uncertainty. Clarity would have made it a lot easier. The CFTC and SEC are giving some rules around—

Were the SEC rules helpful for you?

The SEC rules were helpful; the CFTC's were also. Knowing what licenses you need, knowing whether an FCM can go directly, and knowing whether every single participant who's offering markets needs to have a full DCO/DCM—what does all of this actually look like? There's a bunch of clarity on the licensing side of things.

Which I think has been very helpful. There's still a little bit of contradiction, I should say, between some of the prior no-actions that were given and the new set of SEC guidance that were given, in terms of—

Where you'll need an ATS or what this looks like. So I will say it's all moving in a very positive direction from the regulatory-clarity side, which is amazing—huge, huge, huge progress from that side.

So 4 to 5 of the largest exchanges in the world are all actively live right now, rolling out markets on testnet and doing all this stuff. Multiple of those have committed to going into public markets. We'll see this transition into public markets on Atlas—or, I should say, into permissioned markets in production, sort of public mainnet. The timeline for that is just going to be whenever they have 100% complete regulatory certainty.

I think what you're going to see is that, early on, we have 20 front ends right now. Remember, Atlas is the core. It is the trading engine itself; we're not running our own front end. You have 20 front ends that are basically building to be able to offer this up and to be the front end or the touchpoint for the consumer, or for anybody as they're interacting with this.

You have 20-plus front ends deploying on Open Atlas and testing now, and then the large exchanges. I think you'll see Open Atlas markets come broad and immediate, and then you will see a constant stream of progress on the institutional front, including multiple of the largest U.S.-based institutions coming to market.

Without naming any names specifically or giving anything too much more specific than that, it is not, “Let's follow along,” or, “This seems cool; let's put our name on it.” It is deployed on testnet right now, actively testing and asking when and how they can take this into production.

So many more questions for you, but first of all, the people that bought the tokens—the investors—are they also invested in the equity of your company?

There are some splits. A bunch took token only; some took token and equity. One group—no, actually, they ended up taking both token and equity. Everybody has at least token exposure. Some have exposure to both, but the largest 2 checks were token only.

Okay, interesting. I think maybe let's lead into the next question now that we've talked about Atlas. You have this entire infrastructure built. Where does the ZRO token fit into the ecosystem? There's been a lot of talk about when the fee switch is going to come on. It's overwhelmingly positive, but quorum never gets reached. What's the plan here?

7. The Fee Switch, And Burning ZERO

Yep. I'll try to make it super clear now. LayerZero, the protocol messaging protocol, has the fee switch. Every vote is overwhelmingly positive, so it seems like that will be turned on. I've mentally said I think within the next 18 months; you're sort of looking at that time frame for it.

For ZRO, it's very clear: ZRO is the native gas token of the chain. All priority fees, all block space, everything in terms of gas is ZRO directly.

Atlas itself has its embedded set of trading fees. A bunch of that goes to the front end that's offering this up, and then 75% of the remainder goes directly to buying and burning ZRO. You should think about this as, if the fee schedule matches what is in the market today, this should be about, call it, 1.2 to 2 basis points of every dollar traded on the platform going directly to buying and burning ZRO.

And that's still going to be cheaper than the traditional financial system because of the other cost savings, or what's the comparison?

Yes. Again, it depends on what you're looking at. If you're talking about BTC perps—more open, broad, traditional perps—your blended average rate across most venues is somewhere on the order of about 3 basis points. If you're talking about forex or equities, that fee schedule might look very different, but volumes may end up being drastically higher.

It depends on where you're looking, but I think calling it 1.2 to 2 basis points is sort of our mental model right now.

It depends a little bit on how some of the institutional markets are going to roll out and where that fee lands, but 75% of all fees generated by Atlas itself are used directly to buy and burn ZRO. So that is very clear-cut.

That's pretty good. And Atlas itself—I want to talk about how it's going to impact a viewer here. When I log on to Fidelity or Robinhood and make a trade, it does feel pretty seamless for me right now. There's not a ton of badness, let's say, with this experience.

Other than the fact that when I sell a stock, it takes 2 days for the cash to settle in my account. That's quite annoying. But what is Atlas going to do to make those traditional experiences better, if anything? Are you ever going to be the backend to Fidelity, the backend to a brokerage? Is that the way the world's going?

So yes, definitely. Multiple names—similar names, et cetera—are actively involved in this question. If you look at stablecoins, it's very interesting to see that domestic stablecoin use has trailed international stablecoin use by almost a decade now. Honestly, in terms of real traction, only now are you starting to see traction. You're starting to see the banks get worried about this.

This is why tokenized bank deposits are being pushed so heavily: what stablecoins might do to the deposit base of banks, what this actually looks like, and where tokenization moves. Whereas internationally, you have 500 million-plus users and $300 billion worth of stablecoins. They've grown 60x in the last 6 years, alongside RWAs, which have grown 2,000x in the last 6 years.

8. Tokenizing $100 Trillion

I think you're going to see a bunch of stuff happen internationally that has started to put pressure on the domestic side. Take DTCC: DTCC has a direct mandate to tokenize $100 trillion worth of assets. It was a completely unhinged mandate when you think about it. It's like, okay, actually, a huge amount of the asset base is going to be tokenized. There is movement in this way, and then the question is, what does it actually look like?

I think when you talk to the institutions, there are 2 things that are immediately interesting right now. One is perps as a product: what perps as an instrument have done internationally and the volumes they're generating now have become very interesting. And then, what does perps as a product within the United States look like? Where does that live? What does that look like?

And then the second is, is there room to change the existing model broadly—to touch more internationally? You look at something like ICE: ICE has 6 clearinghouses that span around the globe, and customers need to keep funds in all 6 of them. If you're talking about a single stack that does settlement, clearing, and matching—and crypto prime brokerage alone is a $50 billion business—the cost structure starts to become pretty hard to ignore for most groups.

You look at perps as a product right now and how much some of the venues are making relative to traditional exchanges. They're doing basis points of the overall volume but making double-digit percentages of the net profit of the trades. So I think there's a bunch of stuff that is interesting, in flight, and happening.

To answer the question directly: are brokerages going to start to open this up? Are you actually going to have trades opening up tokenized products, perps, et cetera, on these rails to your normal mainstream brokers? It is 100% yes, that is going to happen. Now, will the user ever actually see it and notice? We'll see, right? How much of that just gets abstracted away? But 100% that is happening.

Okay, that's good. I want to also talk about what you see the future of this industry looking like. We've seen a tremendous amount of tokenized assets come on-chain. Most of these are equities that already trade elsewhere, that are put in some sort of SPV or tokenized, and then issued on the chain.

In the conversations that you're having, do you think it's likely or possible that, in the future, equity—like, if a company issues equity—it'll just immediately be on-chain, like in 10, 15, 20, 30 years? Are all IPOs going to end up on-chain, or is this just something to appeal to a niche set of an audience that wants their assets on-chain, and there's always going to be these 2 parallel systems running?

Yeah. So I think on a long enough time horizon, yes, definitely. I think all of this started with: What does an institution—what do any of these groups want? All of tokenization today is basically just a distribution engine. There was a pool of capital that was sufficiently large, and it was like, okay, well, now it's a multiple-hundreds-of-billions-to-trillions-of-dollars asset pool.

If I'm selling a high-yield private-equity fund, or if I'm selling whatever it may be, I'm very happy to just distribute that in the system. I think stablecoins, again, have crossed the Rubicon in terms of Lindy adoption. You look at Visa's stablecoin volumes. You look at what Stripe is doing. You look at all this stuff. This is just getting integrated more and more into traditional flows in a way that is slowly but very steadily eating away at existing systems and legacy systems.

So again, where we are in tokenization today—excluding stablecoins—you're trailing by 5-plus years from where we are in the rest of the industry. But I think very clearly it is going to start happening, and I think if it is a catalyst that pushes, again, a consolidation of some of these components that are separate today, that don't need to be, and an ability to distribute to a more international audience in a way that the regulatory framework evolves around, I think all of these things are very, very strong drivers.

So I think, will an IPO happen on-chain? I think on-chain and off-chain—there just won't be a difference anymore, right? I think the rails are just, again, slowly being replumbed in a ton of this stuff, in the way that stablecoins themselves are completely starting to cannibalize some of the existing rail sets. I think you're going to see them more and more.

What we haven't talked about yet is the payment side of Zero and some of the interesting stuff there, but I think you're going to see it in a way that just can't exist elsewhere. So there's a bunch of really interesting stuff happening.

And when do you think you're going to go live with these products?

So we have publicly said fall this year, so it gives us a very narrow window over the next 2 months.

It is fall right now.

So, in fall. Yeah, we're right there. Again, testnet is live; you're talking right now on Atlas on testnet. Right now, we have a stable version at 200,000 transactions per second at sub-1-millisecond latency. We have a smaller version at 10,000 transactions per second. That's 400-microsecond latency.

And then we have a research prototype that's at 28-microsecond P50 latency, with only 40-microsecond P99 latency. Institutions have been hugely impressed. The most common feedback we've gotten is, "We didn't think this was possible," and, "Oh, wow, we have never been able to do this."

We have a bunch of them who took their existing workflows. This can be for how you distribute data around tokenized equities, and it can actually be just the trading engine itself on-chain—just CLOB on-chain. In both cases, they're hundreds to thousands of times more performant than what they had before. So I think we're in a great state right now.

We're talking about whether partners want to open up testnet to the public so people can come and play with it. But right now, all the partners and all the front ends—again, 20-plus front ends and a bunch of the largest institutions—they're all on live testing. From a technology perspective, I feel great. I feel like we're pretty much there.

From a release perspective, it's just aligning go-to-market, getting all the partners together, and making sure all the small pieces are in place. Launching a chain is not easy. It's a lot different than launching a product. You have to make it incredibly easy to get assets in, and you have to make it incredibly easy to do all of these things.

And then you have custodians and everything upstream for all of the partners that you want, which you have to sort out. There's a bunch of that going on to make sure you have the best possible launch.

What are your plans for getting people to come on the chain day 1? Because I assume you don't just want institutions on it. You want the average person—you want the people listening to this podcast. You want the people on Twitter to come on as well.

Of course. I think one of the benefits is we've got about $150 billion worth of assets today on top of it. It's very easy to open that up to the world and connect to Zero. So I think getting assets in will be quite easy.

I think when you're talking about your larger institutional groups, for any user, it's going to be seamless. It can be a wallet; you're going to bridge, and it's going to be easy. You'll just deposit, and it will be a non-factor.

When you're talking about your larger institutional groups, your fund managers, anybody who is, again, using regular custodians or Fireblocks, it's just getting all of the infrastructure set up for that to be very seamless and easy for them. It's getting the exchanges on board. It's getting everything across the board to make it as easy as possible. But for the end user, it's going to be really easy. We've spent the last 6 years making this stuff seamless. It will be very, very trivial to deposit.

Bryan, I'm super excited for the launch. You heard it here: we're launching in September. It's going to be fun, and I really appreciate you spending the time to talk with me today. Hopefully, everyone found this to be a great conversation. I really enjoyed it myself, and I'm definitely going to be poking around LayerZero. Thank you, Bryan.

Appreciate it. Awesome. Thank you for having me. I'm looking forward to, in another short period here, being able to come and say, “All right, all those names I couldn't mention—here they all are.”

We're going to have to have you back on once you can talk more publicly about these things. One thing I do want to get into—we could probably do a whole episode on your discussions and learnings, and what you found interesting from all these people, because I think a lot of people just don't understand what's going on behind the scenes right now. So we'll have to have you back on once you can talk about it.

Deep in the weeds. Looking forward to it.

Yeah, Bryan. It's just crazy. The world has changed so much since we first got into crypto, and so has LayerZero.

The difference between our early days of BD and doing BD with SushiSwap to now spending 100% of my time with Fidelity, with BlackRock, with [?], with all these different groups is just night and day different. It's a totally different world. It's been really fun and enjoyable, but it's been a great transition.

I mean, it's a sign that we've made it.

For sure.