# We Are Only 1% Done: Evan Cheng on Sui, AI, and the Next Trillion-Dollar Web3 Era | EP 163 Frictionless · 2026-04-16 · 58 min · https://www.youtube.com/watch?v=CJ4RbnmJag8 ## Transcript Evan Cheng I don’t think the industry has done a good job of making it easy because of fragmentation, lots of different chains, a lot of specialization, and a lot of noise about what you should be using and in what way. It creates a lot of barriers to entry for a lot of developers. Internally, we are building a bunch of things in the AI space and even in the financial space, and there’s an enormous amount of work to do. All we are focusing on right now is raising the abstraction. Logan Jastremski Would you say the verticalization is because of platforms? Just like you don’t build toward a microkernel on top of the macOS or iOS kernel, you build against a framework. Evan Cheng I think the questions around scalability will come back again. We are at the flat part of the adoption curve, which means in 10 years, you’re going to look back at this stage as basically flat. Even though if you look at the last few years, you would say, “Okay, maybe this is the beginning of the curve,” but no, we’re not there. Logan Jastremski Well, Evan, thank you so much for coming on the podcast. As I was saying before, it’s been over 3 years since we did our first podcast, and I think that was even prior to Sui’s mainnet. I think it was testnet. Evan Cheng No, no, it’s 3 years. It’s mainnet. Logan Jastremski Okay. Evan Cheng Yeah, the world has changed. Logan Jastremski The world has changed. I remember doing back-to-back podcasts with the entire team, and I think in hour 4, my brain was mush trying to absorb everything that you guys said. But it’s kind of cool looking back and seeing those early podcasts because a lot has changed. Evan Cheng A lot has indeed changed. ### The Scalability Maze: Why L2s Failed to Deliver Logan Jastremski I think when we first talked, a lot of the industry was still centered around Ethereum, the EVM, and Layer 2s. In large part, I think that has failed in terms of really enabling scale. Now the industry is trying to find its footing, so to speak. It’s less about throughput and has centered more around trading, different revenues, and trying to find different ways for blockchains to make money. I don’t know if you would agree or disagree. Evan Cheng I think the question is wrong. Scalability will come back again. We are at the flat part of the adoption curve, which means in 10 years, you’re going to look back at this stage as basically flat. Even though if you look at the last few years, you would say, “Okay, maybe this is the beginning of the curve,” but no, we’re not there. Clearly not. We’re only at the point where there is proven adoption in stablecoins. Even that’s early. People are quoting all kinds of numbers, but look at the absolute number: it’s tiny. RWA and tokenization in general are still very, very early. They create a lot of assets and a lot of experimentation, but mostly these are not productive assets. The industry is shifting in a lot of ways. We gravitate toward areas where there are some signs of product-market fit and some early interest from institutions. You see that, but all theses remain unchanged from the beginning. It doesn’t really matter where adoption first takes place— which sector, which industry, or what particular use case. Ultimately, the thesis is that the world is going through automation. The current model of a human in the loop who becomes the verifier and the sort of service producer of trust is a guaranteed model that will break. Whether we prepare to solve that problem or it’s just going to be chaos for some time until we solve the problem doesn’t matter. Ultimately, it will happen. It’s even clearer than ever that it’s happening. The world has gotten to the point where the scarce resource is not intelligence. It’s verifying the output of LLMs and other AI models that will come out. If you don’t provide verification, the iterative process and the feedback will drive the output away from correctness, truth, and doing right by humans. Then we have that mismatch. Ultimately, it will have to happen, and blockchain will play a role. How big a role? We’ll see how well we do it. At that point, all the questions about scalability and everything else will come back again, at a completely different scale. Logan Jastremski I think the interesting thing has been that we kind of explored the idea maze of scalability. We started with the low-throughput blockchains like Ethereum and Bitcoin. Obviously, Ethereum made it Turing-complete, and then the ideas around Layer 2s largely failed, in my opinion. Then we did sharding and modular blockchains. In my mind, you guys were obviously very early in the direction of enabling high-throughput blockchains. The one thing I would have loved to see across the high-throughput blockchains is, to your point, taking more advantage of the throughput. I’ve seen high-level numbers showing blockchains still only doing megabytes of throughput versus gigabytes and beyond. I’ve always been a little confused about why engineers or builders never really tried to push things on the throughput side. Evan Cheng There’s not enough demand at this point. If you look at what we have been working on in terms of scaling, we’re dealing with cases where you have a period of persistent, relatively high throughput or load. We found other bottlenecks, such as the storage layer, so we actually built our own database instead of adopting an open-source database. We’re finding that scaling use cases are all about building to the needs of products today and their usage, and making sure we can handle those cases so we have a smooth, seamless experience. I think people perhaps forget that most of the very big systems today, in general, are built from a highly successful product going top-down: Facebook, Instagram, Google, YouTube, TikTok, and all that. All these companies had a super-successful product. Then they could build the infrastructure according to the demand they anticipated and scale the system that way. Blockchain companies were the same. We would have loved to start building these amazing products to serve human customers, but we didn’t have the blockchain infrastructure. Because blockchain infrastructure was meant to be built as public infrastructure, we had to build that first. But it’s a backwards way of doing things. You’re going to have this interesting situation where you initially overbuild, then learn where the demand is coming from and what the performance characteristics actually look like. You find the bottlenecks, fix them, and grow from there. We’re still doing this without control of the demand. That’s the next phase. We knew we were doing things backwards—bottom-up rather than top-down—because that’s how platforms are built. ### Custom Databases: Scaling Reads and State Access Logan Jastremski Can we talk about the custom database for a second? I’ve been very interested in this idea that blockchains were scaling the writes for a long time. They had relatively low throughput on the writes, but now, as you continue to scale the writes, I feel like the reads, or the custom databases, are going to be more and more of a challenge. You may have, for example, one Instagram write, but you may have 1,000 reads from that. That state access, and even the propagation of that information, is something I’ve been thinking more about. I’m curious if you could touch on more of that. Evan Cheng I probably don’t know enough to talk about the details. I think it's called Tide something, I forgot. Basically, it’s an observation of the traffic and why we end up having a bottleneck when you have a long period of sustained load. You analyze it, recognize a bottleneck, and then understand that Sui is very different and built very differently from, for example, EVM chains. They have an inherent bottleneck at— Logan Jastremski On the account axis. Evan Cheng The account axis, and a whole bunch of things where you cannot have one account generally have parallel execution across lots of transactions. That’s just a fundamental bottleneck. Good luck ever solving that without tearing up the whole thing. We don’t have that kind of restriction, so all patterns of disk access are going to look very, very different from one another. That’s just one of many examples of how we learn based on our characteristics and figure out how to solve the bottleneck. The interesting thing is that this is the pattern you see with a company. I saw this inside Apple and Facebook. You see this: “Here’s a load, and the load is increasing. We anticipate the load will be this much, and that pattern will look like this.” At the same time, we’re trying to steer users from maybe photo- or text-based content to more video, and that load will look completely different. You have this multiyear project to improve your underlying infrastructure. That’s the kind of scale we’re talking about. We’re fortunate that we have that kind of experience. Logan Jastremski Yeah. Evan Cheng We can handle these things. If you’re just taking a generic blockchain, forking it, or trying to rebuild it, but you have this inherent bottleneck built in, good luck with that. Logan Jastremski Yeah. Evan Cheng You can engineer your way out of a lot of the technical debt to a certain degree, but there are some fundamental issues that will force the system to be rebuilt or rearchitected. ### Walrus & Deepbook: Verticalizing the Web3 Stack Logan Jastremski One of the other very cool things that I've seen Mysten and the Sui team do is really, I would say, put a lot of energy into DeepBook and Walrus. I've been increasingly interested in the verticalization of some of these products. If, at least for now, product-market fit on the blockchain side is trading or trading derivatives, DeepBook is very interesting, and then kind of pulling apart the read layer and even storage and what you can do with gated information, so to speak. Walrus is also very interesting, so maybe, at a high level, touch upon why you guys decided to do those. Evan Cheng Yeah, because this is the right way to build a platform. Imagine you're building for an iOS app. You have to build one program that focuses on compute and moving some things around, and you have to write a completely different program to deal with disk access and storage and everything else, and maybe another one to deal with this and that. I mean, that's basically the blockchain model for everything else out there. You don't have a central, single-entry-point control plane—a way to program the whole thing with all the different components working together. That's insanity. That will never work. Nobody will ever program that way, which is what you have today. If you want to use Solana for your smart contract and compute, and you need to store something persistent—for example, we just released an AI persistent memory—then you say, "Okay, I can build this either in a centralized solution or maybe Arweave or Filecoin, which is a completely different blockchain." Logan Jastremski Yeah. Evan Cheng And you somehow have to figure out the interaction between them. You have to program 2 different sets of things. They are completely disconnected. That is, I'm sorry, one of those architectural decisions you make where, at the beginning, you think this is perfectly fine. Only later on do you say, "Well, this is about as dumb as it gets. Let's shoot ourselves in the head and not do that again." So, a platform has to be one where you write one program and have a unified interface. In that program, you deal with different components and different kinds of policies, and you can't ask developers to program lots of different things. Imagine this is actually being adopted in a real, complex program that does everything. You have storage. You have a control plane. You have encryption and decryption of secrets. You have all sorts of other things. Yeah, it's not going to work. This has always been built with the way we designed it to be built. You have 3 layers at the base: a smart contract layer, a coordination layer, and a layer where assets get settled. Then you have worlds built on top. The Walrus storage and Seal are tokenized. They represent themselves as objects on Sui. You may manipulate them in smart contracts using the same interface. Seal and everything else built on top is just building out the vertical stack. I don't know why people are surprised, but that's the way to build an actual software stack. Now, however, this is other magical thinking that's happening with all the other blockchain ecosystems. Logan Jastremski I very much appreciate the platform approach that you guys have taken because I think now—I mean, Ethereum kind of just threw spaghetti at the wall and experimented, in my opinion. And now what we're at least starting to see—I mean, Solana was pretty explicitly focused on trading. Aptos published the global trading engine. I talked with Keone and did a podcast with him earlier this week, and he was saying that really, decentralization and just being a credibly neutral platform was their main goal. And so, what I find fascinating, just from my investor lens, is the different paths that everybody's taking and how those will play out over time. ### Beyond Digital Ledgers: Sui's Object Model Explained Evan Cheng Yeah, I mean, it's a completely different approach. I like to think of this digital ledger technology, which is all the other blockchains. They focus on one problem: You have a ledger that's tracking the movement of assets. These assets mostly don't carry a lot of state. They're not very stateful, or they have minimal state, but they're uniform. Fungible or not, you write logic to deal with what changes to the asset, but in a very, very limited way. Whereas on Sui, we don't believe that distributed ledger technology will extend beyond the early use cases we're seeing, which is just tracking movement. For stablecoins, that's completely fine, but anything that's more complex—where you can say 2 different instances of the exact same class of asset will carry completely different states, and you have to change them independently, and you can combine them, and they may go through different kinds of life cycles—that digital ledger technology model mental model break down immediately. And you see all the hacks to try to work around them, but ultimately, they're going to be very limited. So, whether that's right or wrong, one can argue we're overbuilding for the near-term use cases. One can make that argument. That does make our job a little bit harder. We knew at the beginning there would be a lot of complaints: Because your concept is not just a balance of coins, it makes a lot of existing, well-known DeFi-type products or payment-type products a little bit complicated. But we want to build something that's ultimately much more flexible, then come back and add a layer on top to solve this problem, and that's what we're doing right now. You're going to have something that looks like a coin with a balance, where you can easily manipulate the balance rather than dealing with each individual coin separately. But we still retain the benefit of being able to completely parallelize transactions. You can manipulate them independently at scale. So, what that means is, it did take us longer to develop some of the things in terms of making it easy for certain types of applications to be built. But the upside is, as more problems get more complicated and products get more complicated, especially the ones outside your pure crypto use cases, you're going to see they're just going to be much, much easier to build on Sui, on the Sui stack. And now we're talking about the stack as a whole, not just Sui by itself. Nothing else can claim that. Logan Jastremski Yeah, it's cool because, again, I increasingly see everybody choose their lane. And the fact that you guys—I wouldn't say are rejecting the trading thesis by any means, because you guys have DeepBook, and the object model transactions are very easily parallelizable—but it's more so continuing to build on the platform. I do see now, especially with Hyperliquid and people running discounted cash flow models, they're like, "Hey, why are these assets worth X, Y, and Z?" And then on the exchange side and trading, it's very easy to say, "Here's the revenue. This is why things should be valued high," with some justification, so to speak. And so, I'm just very interested broadly in all the different paths, because I have seen more people gravitate toward the trading side, and then you and Monad, so to speak, focusing more on building the holistic platform. I don't know. They're different approaches, and I'm excited to see how they work out. Evan Cheng Yeah, and you saw Aftermath. They're purely on-chain perps, right? That cannot be built anywhere else. Logan Jastremski Those guys are crushing it. Evan Cheng And they've increasingly started getting attention and getting more usage. It's now one of the more dominant applications on Sui, and that took them a long time. They're an example of taking a long road, having conviction in the underlying technology, and building the product. Ultimately, it's very, very difficult to build elsewhere, if possible at all. I know some of the others claim to have that, but there's always some off-chain elements doing the heavy lifting and matching and all that. So, I think, as far as I know, they're the only real, pure on-chain perps. But, you know, we don't have to say, "Well, we built a blockchain purely for that." Logan Jastremski And one would argue, why wouldn't you have gone down the Hyperliquid model? Evan Cheng Well, that's more of a top-down model, as I say, which is how most of the platforms have been built. You have a hyper-successful product, a very successful product, and there's demand to build on top of your product—you know, reach the audience you've already captured. Then you open up as a platform. They've been very successful; kudos to them. They came from the right background, they have the know-how, and they built a great product. We'll see in the long run whether that becomes a successful platform or not, and whether our approach will ultimately be even more successful. What we want is to have a Hyperliquid, or multiple Hyperliquids, on Sui, while we have everything else that Hyperliquid and all the other blockchains cannot build. Logan Jastremski Yeah. And Hyperliquid specifically is interesting just in the sense that they have built a good matching engine, but if you look at HyperEVM, I think, going back to the limitations of the EVM, it's doing 15 TPS. And so, it's hard to build a scalable platform on 15 TPS. Evan Cheng Yes. Logan Jastremski Yeah. So, I guess, broadly, it seems like, again, you have an appreciation for trading, but it's much more, I would say, in my mind, similar to the original ethos of what we wanted blockchain to be and aspired to, which was Web3, which had a much more grandiose vision than just peer-to-peer trading and digital ledgers, so to speak. ### AI, Automation, and Providing Trust for LLMs I'm curious what other products you think are going to be useful. Maybe I know AI and Claude and all these things are now actually becoming extremely useful, but what other product lines would you love to see get built? Evan Cheng We’re actually internally building a bunch of stuff in the AI space, and even in the financial space, there’s an enormous amount of work to do. All we’re focusing on right now is raising the abstraction—continuing, if you would say, that verticalization. You don’t build directly on top of a microkernel, or on top of the macOS or iOS kernel; you build against a framework. When people are looking for solutions to a certain problem, maybe they say, “I want to solve this payments-of-multiple-parties or micropayment problem.” They want a solution. You don’t give them a very low-level primitive and say, “Go solve it based on that.” So we’re building lots and lots of solutions, some of them in the financial space and the payment space. Some could be trading, and maybe even some DeFi niche products. There’s also a lot of AI—or, rather, automation and trust. Providing trust is kind of the problem. I’ll just use one example. We talk about what is perhaps a little cryptic: when we proposed our memory layer for AI, we talked about how AI currently has some kind of local memory. Every chat window may have some memory. Sometimes it extends beyond that single chat window; it remembers some aspect of you, but not all of you. It’s kind of limited, and the window is relatively small. What if you have persistent memory that’s actually controlled by a single entity, human or AI? That’s what we produce. Over time, that’s going to move toward other kinds of solutions that provide context to the underlying AI in a way that can be even more efficient. That’s one example. There’s a whole bunch of other examples. We’re working with others on everything around how you make sure an AI agent can be trained on very sensitive data that may want to remain encrypted all the time. How do you do that sort of thing? How do you have data sovereignty, have individual control, and maybe allow data to be used in certain ways? There’s a lot more. The current wave of agent frameworks has been used to provide services to humans. It tends to be very one-agent-centric: you have to open up lots of channels with lots of services, and they’re not talking to each other. How do you make that into one hyper-interoperable system for this future world to be built correctly? This is really exciting because so much of the fundamental technology, especially around AI, is moving incredibly fast. Then there are a lot of hacks on top of it. Why are they hacks? The web—the internet—is really built with humans in mind. Everything, from simple examples to more complex ones, is going to present all kinds of friction for an automated world because it’s built for humans. All these problems come down to a lot of distributed systems and a lot of cryptography, as well as the programming languages and constructs underneath them, to make sure everything is properly coordinated and everything is automated correctly. The amount of opportunity is basically unbounded. That’s the thing: we’re really looking to build a lot of stuff that may take some time to come out, but we’re excited about all that possibility. We never consider ourselves as, “Well, you want to be a blockchain for this or a blockchain for that.” We just see the fundamental problem with all of that. It’s all centered around the automation of trust. Logan Jastremski Yeah. You mentioned increased abstraction, which resonates because, over time, it should be easier to build these applications. You shouldn’t have to nerd out on all the specific details to automate the things you want to automate or move money or assets around. It should be fairly easy. Evan Cheng I don’t think the industry has done a good job of making it easy because of fragmentation, lots of different chains, a lot of specialization, and a lot of noise about what you should be using and in what way. It creates a lot of barriers to entry for a lot of developers. Developers today have to think in terms of, “I want to build for Sui. I want to build for Solana. I want to build for an EVM chain,” as completely separate platforms, rather than saying, “I want to build a product that has a payment element, and I want to use a stablecoin. I want to build a product that has this secret-sharing challenge that I need to solve.” That’s ultimately where we need to go. We’re going a long way toward making things more complex than they should be, but that will all be sorted out over time. Logan Jastremski I think even from the pure investing lens and just the market dynamics, the market has very much punished that increased fragmentation. When we initially chatted, it was, “Oh, there are going to be millions of L2s and millions of blockchains, and they’re all going to be interoperable. Everybody’s going to have their own L2.” In reality, a lot of the L2s have really failed to gain any momentum. It’s increased fragmentation. Even the app-chain thesis more broadly—outside of maybe Hyperliquid or a verticalized product—hasn’t really taken off. To your point, it’s introduced way too much friction into that entire stack, and it’s made it hard for both users and engineers to actually build useful things. Evan Cheng This is how every big change in technology has always looked: super messy for a period of time, until the world standardizes on a few things. It’s very similar to how the infrastructure for companies went from intranets and the internet, kind of this gatekeeping way, to everything being built on the cloud for most everybody else—except for the very few that can and should have their own data centers and internal networks. People forget that even before that, entry into the internet was through America Online, CompuServe, and all of that. Only then did you have the web, which standardized everything around that model for the vast majority of people and the vast majority of tasks. Tasks go through a few browsers, but that model is the same for everyone. This will be the same. Blockchains are unique only because there’s immediate value. That’s the only part that’s really different. Otherwise, it’s just like any other technological evolution. It will get messy until people figure out how best to use it and there’s one standard way of using it—the canonical way of using it. That will come sooner than people think. That was the case before. ### Lessons from Libra: Permissionless vs. Corporate Chains Logan Jastremski Since you and the team were really the primary drivers of Libra and Diem, and all that early work that you did at and after Facebook, does it ever frustrate you now, seeing Crypto Twitter, so to speak, stablecoin chains, and what’s currently happening on that side of the world? Evan Cheng It’s not really—I mean, it’s a little bit, in a way. I’ve been very transparent with my thinking. I always want to build something more general. Libra was a really great learning experience, and it’s good to see people take advantage of that experience. Not only did the people who left Libra start a bunch of companies, but others building Tempo and whatnot have also learned a lot from Libra. So it’s not really that frustrating. What’s probably frustrating is that only now are we saying, “Well, stablecoins are a great thing.” In hindsight, it’s very obvious. This is an obviously great thing for the world. It’s always frustrating when you’re in the middle of something: Why did it take so long? Logan Jastremski It’s interesting to me that with Tempo or some of these almost corporate chains— Evan Cheng Consortium chain. Logan Jastremski I think that’s a consortium chain. But it’s almost rerunning the playbook, in terms of an intranet, so to speak. If you can make it permissionless, why would a Mastercard or a competitor to Stripe use the Tempo chain? To me, it’s been interesting that we’re trying to run back that playbook when we already have these decentralized, permissionless, credibly neutral blockchains. Evan Cheng If you think about it from the company’s perspective, the go-to-market strategy is to build something that you need, build it quickly, and make use of it. You can always throw the technology away and use something else down the road, so I don’t think that’s necessarily the wrong thing. Even though, back then, when I was at Libra, I was frustrated and asked, “Why don’t we build something more general?” it’s absolutely the right thing for a big company. They want to solve a specific problem, and they have an advantage that they can leverage to be successful using that technology. They should absolutely go that way. But again, as you say, can it be used for other things? Who knows? We’ll see. We still believe that ultimately taking the longer path to build a much more general platform for lots of different things is going to be the winning strategy. As you say, it could be too grandiose and could take too long. There’s always a risk of being too early for anybody who’s thinking that way, but I think it’s worth taking that risk. Try to take big swings. I can’t imagine anything else. I really wouldn’t be able to be that excited. It’s one thing to be motivated by being successful in the marketplace, competing against others. That’s definitely one of the very strong driving forces behind us. Another driving force is that we have a mission that we want to pursue. We want to deliver the kind of product and impact on the world that we envision. We can't really achieve that if we don't build what we believe in. Again, one can argue we could have built something much more focused and immediate and then grown from there, but that gets a little bit tricky because a lot of the time you get stuck. You're stuck with legacy technology. You cannot get out of that. So, I think overall, we sort of find the balance. Right, kind of balancing. It's a general-purpose blockchain with a lot of attributes that's great, and we can grow this into a stack that solves significantly more kinds of problems than anything else out there. And now we are at a point where we are building the higher abstraction layer, providing solutions for others and for ourselves to build products to deliver to the world. So, we're happy with where we are on that front. ### Sustainable Economics: Business Models for Blockchains Logan Jastremski Maybe this is a super broad question, but do you feel like blockchains should almost have business models? I feel like, in general, at least currently, the market is more focused on trading just because there is a business model—selling block space, priority auctions, or taking a certain amount of fees. Do you think blockchains should be businesses, in that sense? Evan Cheng Yeah. Pretty much everything should have a business model, should be a business, and run like a business. Otherwise, you have uncontrolled costs. I don't want to get into a debate on that sensitive topic, but you can see, for example, government spending is out of control pretty much everywhere in the world. You absolutely need to be able to have a business model. Do I believe that this has its own business model? What people think about when they start a new crypto project—probably a lot of the time, they are not, but I think the market is also correcting that. A lot of the last cycle, or we can say the end of this current cycle, you saw that a lot of chains were just making up literally things. Whether it's a Move or an EVM chain, for example, it turns out there's not really a business model for it other than selling tokens. Yeah, those things are going away. And now I think a lot of people are thinking, “Okay, how do we save money? How do we be sustainable?” Not just selling tokens. And we absolutely are seeing that, and we have a very concrete business model. We are actually executing on our business model. We're pretty happy with that. Logan Jastremski And would you kind of put that for Mysten or Sui within the different product lines, like DeepBook, Walrus, Sui itself, or are they all one? How are you broadly thinking about it? Evan Cheng Yeah, that's a good question. Mysten is ultimately the for-profit company. So when you ask the question, it's 2 different things. One is: How do Sui and Walrus become self-sustainable? They have foundations managing a lot of these, but they may not be in there to be profitable. They just have to be sustained because ultimately they serve the community and the ecosystem. They're thinking more about how long-lived a chain can be and how a chain becomes sustainable. If the fees are going to be low or remain low, or a lot of things are going to be free, how do you be sustainable? How does the validator be able to make money to survive and all that? For Mysten Labs, it's a little different. We have, obviously, a lot of experience on the technology front and the product front for these chains. We'd love for others to join as well, but right now we're driving a lot of stuff, so we care a lot about that. We also care about the company making money and being able to be sustainable. We owe it to our investors; we owe it to our employees to address those issues. So it's different. It's a very complex business. Yeah, but don't ask me how I got into this complex business. It is a very complex business. Logan Jastremski Yeah, makes sense. You mentioned cycles. Crypto has been very cyclical, with high highs and low lows. I guess just a broad question is: You've now been through a couple of these cycles. Has anything ever surprised you, or are there any major learning lessons from each of them? Evan Cheng Sui has only been through 1 cycle, more or less. I've been in the space now 3-plus cycles, 3 or 4 cycles. The thing that surprised us is—you never... there's always a surprise, even though you learn a lot. Logan Jastremski Yeah. Evan Cheng Every time, you learn something and think, “There's no way we'll see anything that crazy again,” and something crazy happens again in the next cycle. Or somebody gets away with it, or you miss certain signals and get confused about where the market is focused or where it should be going. You forget, right? Maybe not confused—you forget. And sometimes you take your eyes off the goal a little bit; your priorities could be off, and you have to catch it and address it. There's no end to learning. The good and bad thing about being in this business is that the learning comes hard. Although we're seeing the AI space as kind of very similar, things are moving at a speed that's catching everybody by surprise. A lot of current winners may not be winners of tomorrow, and vice versa. So, yeah, it's fascinating. The real-time feedback is something you really have to be in it to understand. ### Unlocking Stagnant Capital: Bringing Yield to Bitcoin Logan Jastremski It's interesting—we're here in New York at DAS, or Digital Asset Summit, which is happening this week. The difference, I would say, obviously, between broad market sentiment and what we're now seeing—the more open-minded regulatory side and the increased adoption from larger asset managers or institutional clients looking at blockchains—feels like it's almost at an all-time high. And so it's a weird juxtaposition: The market is what it is. Crypto traders may be in their own little bubble, and maybe that's not ultimately the end customer base, but Main Street, so to speak, is now adopting this technology. So it's a strange time. Evan Cheng Yes, it's a strange time. This is another surprise for the cycle. Everybody misread it. Institutional interest is at an all-time high, but the price is nowhere near the all-time high. In fact, Bitcoin has been a somewhat stagnant asset if you owned it for years or something. Basically, it hasn't done anything. We're adjusting. This is actually a good segue, right? Why do we produce Hashi? It's a recognition of this problem. You have an asset that basically has been stagnant. For anybody on the institutional side, they would be saying, “A trillion-dollar asset that's generally basically very little yield is insane.” So, we're trying to unlock it. Like I say, we adjust. We learn from the mistakes. We build things to address it. Logan Jastremski And there—I mean, yeah, on the Bitcoin point, I think you have broad supporters and broad critics on each side. And just a multitrillion-dollar asset, to your point, that earns no yield. I don't know. I've never personally had a strong opinion on Bitcoin. I've always just liked products. But Bitcoin seems to be the special snowflake that doesn't have to produce cash flows or revenues, and it looks like everything else does. So, interesting. Evan Cheng Well, I mean, people are asking the question. It's getting to the point where it's a mainstream asset. It's not just crypto believers that hold Bitcoin. A lot of institutions, a lot of other individuals, hold Bitcoin. So they ask a completely different question: How do I make the most of my capital? I have maybe, say, 20% of my assets in Bitcoin. It's not doing anything for me. I'd like to make sure it generates some yield while I'm waiting for the price to rise and all that. It's okay, but I can't have a stale asset, a nonproductive asset. So maybe from the point of view of real crypto believers, the OGs, they're used to this. They're comfortable with it. They're comfortable holding it for a very long time because ultimately it will keep on going up in the long-term horizon. But for others, it might not be acceptable. So it's an example of a real problem that needs to be addressed. And the good thing, I'll just say, is once you address this problem, think about what you can address for all the other kinds of tokenized assets. Logan Jastremski Yeah. It's been interesting, even just the attempts, I would say, at making Bitcoin L2s, or even people like Nick Carter, I think, have been pushing quantum-resistant Bitcoin. But I don't know. It's definitely an interesting asset. We'll see how it continues to evolve. How do you feel? Maybe again, just a broad question: I think there was a lot of focus early—in the early days of smart-contract platforms—on community and adoption. Some projects have created fake communities, and in large part those have petered out. But I think it's always really been product-focused, and I appreciate, again, that you guys have been very product-focused. How do you feel like the Sui ecosystem, the products, and the builders within the ecosystem are continuing to evolve over the past couple of years? Evan Cheng It's all maturing. I think, like you say, there's a lot of people who are part of the community for the wrong reason. Those mostly go away when the market is not so good, when they're not getting free stuff anymore. That never was real. Logan Jastremski I never understood the airdrop farmers and why we're giving people free things for doing nothing. Evan Cheng Yeah, all the meta is that only meme coins represent community. I think we have a true community—people who believe in the long-term potential, who are looking at the long-term view and are excited about what they're seeing. We can debate the real numbers. Nobody really knows. What we're seeing is that the quality of the builders and the experiences is improving, and they're getting more mature. In a down market like this, you don't necessarily have a huge influx of people who are trying to catch the meta and make a quick buck, but there's a steady inflow of people building real products. It’s really exciting. Again, using Aftermath as an example, they’ve been with the Sui community for a long time, and some people complain about them. They complain about us all the time, but this is an example of somebody who’s been building for many years and is getting to the point where they’re basically the first purely on-chain perp, and it’s doing well. That’s great. We’re going to have more of these. We have some real builders who’ve been with us from the very beginning, or sometime even before that, in the Internet’s early days, and only now they’re finishing the product and are excitedly trying to bring it to market. I think we’re very healthy on that front. We’re seeing the right thing. People keep forgetting: it’s never the quantity that matters. Even today, you can look at the established products on Ethereum and Solana. You can count on one or, at most, two hands how many products have been successful over long periods of time. It’s not just about how many people on Twitter, X, or Reddit are talking about it. They will always come and go, and people will always have opinions. We listen to them all the time, but we’ve always had a lot of people who believe in us and build with us. In fact, every time we have an event—not a party, but an event—I’m always surprised by how many people show up. I haven’t been to one recently, so I don’t know if there are still lines, but the last time I went to one last year, there were still lines around the block waiting to get in. I don’t know. My co-founders love it. All 5 co-founders are still here, working really hard every day. That’s saying something, too. Logan Jastremski I’ve been to both of the Basecamp events: the first one in Paris and the last one in Dubai. I love the energy. The Paris one was special because people were excited. It reminded me of the early Ethereum and Solana days. Then in Dubai, I was just surprised by the number of people. I was like, “Wow.” It’s cool because you feel the energy in the room. Evan Cheng Yeah. That’s one of the many things that drive us forward. Feeling the excitement and knowing that there are people who believe in us is a big thing. It’s a big driver. When you think about the mission we established on day 1 and continue to have today, we’re progressing toward it. We’re building real, cutting-edge stuff. I don’t think there’s any dispute about that, regardless of who you believe in. We build super-cool technology—the best technology. That’s something to be proud of, and that’s why we keep working really hard. People ask me sometimes, “What’s driving you forward?” I say, “What’s not driving me forward?” All the stuff I set out to do—I can’t say that it’s already been done and is over with. We haven’t even done 1%, by my estimate. I keep giving my team heart attacks and saying, “Yeah, we’re about 1% done.” They’re like, “Keep going? When is this ever going to end?” I mean, technology improvements and moving forward—the mission is so big that it’s probably never going to end. But it’s great. We love it. Logan Jastremski That is. It’s hard with 1 co-founder, let alone 5. Evan Cheng Yeah. Logan Jastremski Very impressive, though. I continuously admire what you guys have built. To your point on quality, I’m excited because, generally, what happens in bear markets is that people go back to the fundamentals. People focus on getting from 0 to 1 on the product side. It’s all the things that Silicon Valley has known and loved for many years: focusing on product, focusing on distribution, focusing on revenue. There’s always that refocus when things take a bit of a downturn. I’m excited to see that teams within the Sui ecosystem are continuing to focus on getting from 0 to 1. Evan Cheng Yeah. ### The 2026 Roadmap: EVE Online, Privacy Guardrails, and Product Infra Logan Jastremski Absolutely. Maybe as we wrap up, is there anything in particular that you’re looking forward to in 2026, either on the product side with Sui, upcoming products or technical improvements, or on the regulatory side, such as increased assets coming on-chain? Evan Cheng Sure. So much—so many things. I indicated that we’re building a lot of products. I think more in terms of product infrastructure. This is maybe meta terminology, but it’s constantly shared among the big tech companies. You have your core infrastructure, your low-level infrastructure. What we really need to build is the product infrastructure for products to build on top of. It’s solutions—a bunch of solutions—not your nitty-gritty, low-level stuff. Not everything needs to write its own smart contract. Many of these things, such as Hashi, are in that bucket: product infrastructure that people integrate into their products or build on top of. We have several of those we’re building, and we’re super excited about them. Everything from stablecoin orchestration to more DeFi-type products and solutions, to the continuing improvement of DeepBook to be a real powerhouse in the central liquidity layer, to some of the things we’re building with our partners. Our work with the Eve Online team is huge. Not only is the game huge—I’m sure this one will be really, truly that kind of forever game that will be huge—but it’s also a developer platform. There are thousands of developers writing Smart Assemblies and building stuff on top of it. It’s the perfect kind of game to weave AI into, and I’m super excited about that part. You’re going to see a game with AI elements developed and delivered by another big company. We will provide the other side of the equation: the smart contract, the trust layer, asset management, the settlement layer, payment, and all that sort of stuff coming together. This is probably one of the first real large-scale consumer products you’re going to see. Logan Jastremski I’ve always been super excited about gaming. I figured that Ready Player One-style gaming would eventually come, where you get the virtual-reality headsets. I used to play RuneScape and World of Warcraft a lot, and it always made sense that you could play one of those games and put it on blockchain rails for different auction houses, different items, and game items. Evan Cheng Yeah, I think when you start focusing on, “I’m going to make a Web3 game,” that’s when it will happen. The reality is that more games will say, “I’ll use blockchain or cryptocurrency to solve this kind of problem,” which just happens, rather than forcing your mental model that you’re going to start with the blockchain— Logan Jastremski —and build a game. Right, right. Evan Cheng This is the example. It just happened to be almost the perfect type of game for simulation, empire building, and resource management. There are lots of things to automate and lots of guardrails that need to be laid down by smart contracts. It’s a perfect example. We’re so excited that it’s coming very soon. On the infrastructure side, I touched on something that a lot of people complain about. It’s much easier to have something that looks like an address for your coins, while retaining all the benefits of Sui—parallelization and everything else. We’ll continue to get faster and more scalable. That’s continuous. On privacy, we’ll have private payments done properly—not hacky solutions that route things through multiple centralized exchanges and come back, or other hacky approaches. We’ll have a proper regulatory approach that’s done right, so you can have the experience of buying them from an exchange or minting them and using them in different ways. The way we build privacy is as a suite of solutions. We already have Seal, which provides data privacy: managing your data in a decentralized way and managing your keys and secrets that way. We have private payments, and we have something that looks like private computation off to the side, allowing us to construct those kinds of private interactions between parties off-chain while still retaining the verifiability and benefits of a public chain. That part will probably extend beyond this year. There’s just so much to build, but we’re now really building these kinds of solutions for products. You see this different level of abstraction. We’re building a different abstraction layer, and that’s super exciting. There’s also all the stuff we haven’t talked about on the AI front. All of this is basically unchanged from what we’ve been saying for the last couple of years. People are working really hard. Logan Jastremski Amazing. I appreciate the update, and I appreciate you coming on the podcast and sharing everything that’s going on with Mysten and Sui. I’ve always admired what you guys have built and appreciate the product focus. Even now, I like the differentiated approach. Everybody’s going one way, and I think you guys are continuously focusing on the platform and doubling down on the abstraction layer. I’m very excited to see how that continues to progress. I appreciate you coming on again. Evan Cheng Thank you for having me.