The DeFi Model Institutions Actually Want

Recorded at EthCC, Cannes, day three.

Hodl Up: Welcome back to EthCC. This is day three, and we're here today with David from Lotus Labs and Lotus Protocol. He's going to talk about one of the most exciting upcoming launches in crypto this year, and we can't wait to hear more about it. But before we get into Lotus, I want to hear how you got into crypto. What was the journey like, and what led you to build what you've been building?

David: Great to be here with you guys. Let me take it way back first, and then we'll speed up. I was interested in baseball cards, to read the statistics. I didn't care much about the game itself. That turned into, and I'm so old now, calling my dad's stockbroker to place trades. This was back in the dot-com days. So I developed a love for finance, and for trading in particular. Then I took some different paths in my career. I was in audio engineering for a while. At one point I worked for an Imperial Russian art dealer. I tried a few different things, and at some point I thought, since I love trading and finance so much, I should actually get my education in finance. So I went back to school, and just as I was finishing, around 2018, the big bear market happened. One of my professors was really excited about blockchain, and he brought me along; he tried to get me into Ripple. So we learned and we progressed. Shortly after that I was working in aviation, doing a lot of statistics, analyzing when parts would break down on certain planes on certain days, so I was working a lot with systems and programmatic stuff. Then I got a call from a recruiter at Exodus Wallet. I'd never heard of Exodus. I don't think I even had a wallet at the time; I was just using an exchange account. This was right when COVID was going on, the planes weren't flying, and the business was looking pretty sus. I didn't know if it would work out, so crypto felt like the better risk-adjusted bet at the time. I'm so glad I made that move, because it's been a ton of fun since then.

I joined Exodus, and we launched what was, arguably, the first digitally represented Reg A security, on Algorand back in 2021. I came in helping with a lot of that, and then I started doing the same kind of statistics work I'd done in aviation, on Exodus' exchange and asset integrations, really understanding their performance. That led me into a product career. One thing led to another, and I spent several years as head of product at Exodus as we did a number of expansions and launched some really cool products, including payments products that my friend Kevin at Exodus is still running and doing really well with. After Exodus, I was really interested in building more in DeFi. I'd worked with Everstake and built a partial ETH staking product that's done really well, and I wanted to do more of that. Around then I met my co-founder, Bill, who was leading Index Coop. Index Coop started with a lot of thematic index products, but what found the most product-market fit was structured leverage tokens that give you 2x or 3x exposure to Bitcoin or ETH. That sounded like a great time, so I joined up with Bill. But we realized pretty quickly that we couldn't expand that product suite the way we wanted, because we were limited by the underlying credit markets we could build those structured products on. That's what led to the beginning of Lotus.

Hodl Up: You did a little bit of everything and then decided, "I'm just going to do crypto, because that's where everything is."

David: It was too much fun. I couldn't imagine working in any other industry. It moves so fast. It's great.

Hodl Up: But I think that's a blessing and a curse, because it moves so fast that sometimes you need a break, and no one's going to give you one.

David: That's very true. That's why you work with good people who can pick you up when you need a breather.

Hodl Up: Now I want to talk about Lotus, which you co-founded. Why did you decide to start a company like Lotus in DeFi right now? What pushed you to that point?

David: Going back to what I mentioned about structured products: if you look at the two dominant models in on-chain credit today, you have the monolithic pooled model, with Aave as the most familiar example. Aave is probably the best place to build these structured products right now, because the liquidity is so deep, and you really need that liquidity to scale these products. To keep a 2x or 3x leverage token within its target leverage range, you constantly need to rebalance, drawing on more liquidity or releasing it. So Aave is great for that. But they also have very conservative, governance-set risk parameters. Chaos Labs, a fantastic risk manager for them, basically says: think of the liquidation loan-to-value as your axis for risk, and set the maximum liquidation LTV at the point where your risk as a lender of incurring bad debt goes from immaterial to material. They just chop it off there. That lets you build a 3x, maybe a 4x leverage token, depending on how safe the collateral is. We wanted to build something that could really compete with perps. You have much lower funding costs in these spot markets, and we thought there was a great opportunity somewhere between the really expensive funding rates of perps and something a swing trader could hold for a period of time. The other model is the isolated model, and Morpho is the best example. We love the curation model they've championed, and we think it's been a big unlock. But in their markets, each market gets an immutable risk parameter that never changes. Collateral quality changes all the time, though, whether because of the market regime or secondary liquidity. So the appropriate leverage or liquidation level in a particular market, BTC/USD or ETH/USDC, changes over time, and it changes more the further out you go on the tail of collateral. All these new collaterals coming online de-risk over time as they build up liquidity.

So what Lotus does is this: imagine you had the liquidity of Aave with the more precise, tighter spreads and rates of the isolated models. We do that with a tranched design. In the most senior zones of our market, you have very predictable rates for borrowers and lenders. Then imagine there are eight different levels in, say, the BTC/USDC market. The further out you go on the curve, the more your risk increases, but so does your return as a lender. Why would a borrower want to go out and take one of the riskier loans? Or the opposite: you might want a lower liquidation LTV. Maybe you're taking a mortgage on a home, and the most important thing is not getting liquidated, so you're happy to pay a slightly higher interest rate for that extra downside protection. Or you're running some kind of trade, say long spot, short perp, because you're hedged, and there's no reason not to be maximally capital-efficient. So that's great. But let's say you have a million dollars and you want to lend it at a 95% liquidation LTV. That's pretty far out on our credit curve, and depending on where the market is on a given day, you might have almost full utilization locally at that tranche, but most of the time maybe only $400,000 or $500,000 of it actually gets used. To compensate you as a lender for the bad-debt risk you take by going further out on the curve, we make a key assumption: that you'd be happy for any of your unutilized capital to be used by the next more senior tranche, and so on, until your utilization is fully optimized. So you're never lending to anything with higher exposure to bad debt; you're only lending to something safer. That continues all the way to the most senior tranche of our markets. And let's say there isn't much utilization in on-chain credit generally on a given day. We have Lotus USD, which is an ERC-4626 vault. Inside the vault are tokenized money market funds: one is the WisdomTree fund, and a Coinbase Prime fund is another we've onboarded. As a user of our vaults or a participant in our markets, you wouldn't interact with those two assets directly. But they create a base rate for all our markets, so you're always earning the money market rate, or something very close to it. Even if you hold everything else equal, the other approaches have a kink rate, usually around 90% utilization, where the rate shoots up much faster, because they want to preserve some liquidity so people can exit the market. With us, Lotus USD is the productive debt, the actual internal debt asset in our markets, so a lender always earns at least that, no matter what the utilization is.

Hodl Up: From what I understand, you're going live in Q2.

David: We're looking to go live in early June, give or take a couple of weeks.

Hodl Up: Have you had a testnet? I want to know about user feedback.

David: We're actually launching the testnet next week; we have it running internally now. We're really curious too, because we want to get the UI/UX just right, and I'm glad you mentioned it, because there's more complexity in what I'm describing. You might think, "Oh gosh, how am I as a lender going to manage all these different levels? How would I know the most appropriate level to borrow or lend at?" On the lending side, you don't need to worry about that as much. If you just have USDC and want to earn yield, you'll have a conservative vault, a balanced vault and a growth vault, and the vault managers allocate appropriately within that mandate. But how do you, as a borrower, come in without it being super complex? You enter what you want to borrow and what your collateral is, and as you type, it shows you three levels to choose from. It'll tell you, "This is the safest level for you as a borrower: higher interest rate, but lower liquidation price," and so on. So you can make an informed decision without looking at eight different tranches at the same time.

Hodl Up: You mentioned the UI/UX. With a protocol as complex as this, in DeFi specifically, how do you approach that? I've been in this industry for six years, and I still find myself googling and researching to navigate DeFi protocols, because they seem complex to me at first glance. What's your approach?

David: That's a great question. Having worked as head of product at Exodus, where UI/UX and great design are such a core part of how they operate, we wanted to bring that here too. Our target user isn't really the retail user. Of course we'd love to have them contribute to our vaults, but it's more institutional. We still wanted a really good UI/UX, because I think it matters whether you're putting in $10 or borrowing $100 million. So there are a few key abstraction layers, and we use progressive disclosure throughout the experience: the information you need is always accessible, but it isn't pushed in front of you at every step. That's the progressive disclosure part. Now the abstraction piece. Very early on, we had the same question you're hinting at: this sounds really cool, but will people actually be able to use it? The hardest thing to understand might be your actual risk level. We just saw it with USR from Resolv, and with Stream Finance and xUSD, where collateral failed for totally different reasons. But there's still an incentive for risk curators, because they don't have different risk and rate profiles to choose from on high-quality prime collateral, so they go further out on the collateral-quality curve. We're saying: if you want to earn a higher yield as a lender or vault manager, you can do it against high-quality collateral by taking credit risk instead. That removes some of the need to go really deep on tail collateral, because imagine you have 20 different collaterals and one of them is a USR. How do you go deep enough to really understand what that vault is allocating to? You're trusting that the risk manager has done that, and done it really well. On our side, the credit quality piece could also be pretty complex for people to understand, so very early on we partnered with Credora Network. You're probably familiar with their risk ratings; they rate collaterals and do ratings for Spark. We wanted a credit rating for each tranche in each of our markets, and then a vault portfolio rating that gives you the overall aggregate, including diversification and so on. Again, with progressive disclosure: maybe I just see the rating, that's all I need, I trust Credora and I go forward. If you need to learn more, especially as an institutional player in our markets, you can also see the probability of significant loss: the probability that I'll lose more than 1% of my position if I lend at this tranche or that one. That's the kind of thing institutions need to incorporate into their risk frameworks. So we see it both as a key abstraction layer for the user experience and as something really important for onboarding more serious capital.

Hodl Up: The issue I have with using only one risk assessment company is that I think we need more of them, for your vaults for example. If you use one and they fail, that also damages your brand and company. Have you thought about using three or four similar companies or partners?

David: That's a really astute observation. Right now we're really happy with Credora. They're a great partner, and we think they have the right approach. But you're also seeing Moody's and S&P starting to come into the space and do ratings. I tend to think our on-chain-native folks will get it a bit more right, certainly in the beginning, but we welcome more ratings. There are also some really cool other approaches, like ratings that travel with an asset across chains; Particula is working on some of that. What the ratings do provide is a transversal check and balance on what our risk managers are doing. For example, Block Analitica runs our conservative vault. They manage for Spark, so they're very incentivized not to make mistakes, and we think they're really the best for that conservative profile. But say we had a different risk manager on our balanced vault, which is supposed to stay within a certain risk mandate with its credit strategies. If they deviate from that and move too far out on risk, beyond their defined parameters, you have Credora ratings that are third-party and independent of that risk manager's own practice. We think it's really important to have that check on the risk managers themselves and to provide both to our users.

Hodl Up: Believe me, if we'd recorded this episode a year ago, I would never have thought to ask this: how AI-friendly, or bot-friendly, will the platform be? We're living in an age where AI agents control funds and are autonomous enough to use different DeFi tools and protocols. I know you said you're targeting more institutional clients, but even institutions today use DeFAI on a daily basis.

David: That's a great question, and I'm really glad you asked about AI. Even the risk managers are using agents. You have to, if you want to keep up and stay at the top of your game, because everybody else is going to start using these tools too. Even if it's not yet an execution environment where agents sign transactions or run smart accounts executing strategies, I think that will come in time, and we're starting to see more of that approach. We'll see it used across the board, in risk management and in distribution. On distribution: one of our partners, Superform, is a great yield optimizer with a great suite of yield products. They plan to use hooks to go directly into our markets to allocate, rebalance and add that to certain yield strategies. Agents do many of the same things; look at Giza, for example. It's really cool watching how fast those agents can move around and rebalance. So we see that as a real distribution angle. To make sure you can do that, the test is: can I point Claude at our docs and have it one-shot an integration with our product? That's really the bar I think we have to meet, and we're definitely thinking about it as a viable distribution path.

Hodl Up: There's a current trend in the United States where banks keep exploring ways to incorporate crypto, because they want to outcompete DeFi. A few years ago it made total sense not to leave your money in the bank, because the return didn't even cover inflation. Now they're exploring tokenized solutions, or even DeFi, but not proper DeFi, and US regulation favors that stance as well. How do you navigate these evolving regulations, especially in the US, and make sure DeFi remains DeFi and doesn't just become banks, a wolf in sheep's clothing?

David: That's such a great question. On regulation, you're absolutely right, and we still need to see exactly where the CLARITY Act lands and the fine details. If I recall correctly, I just saw something about Circle wanting to make sure they have a permissionless, yield-bearing dollar coin that can offer the yield you may not be able to get in the future on idle stablecoin deposits. That's what the banks are complaining about: "We feel that's unfair." Oh, boohoo for the banks. We're done with that. Let the people have their yield. But the mechanism by which that yield is generated is where compliance comes in, and from my understanding, active utility in DeFi, whether an AMM pool or a credit market, is fair game. So we feel really good about that. There are other big pieces with banks too: they can now put digital assets on their balance sheets or use them as collateral, which can bring a lot more capital into the space. But then, of course, how do you preserve what DeFi is supposed to be? How does it not just get co-opted by the same institutions we built it to get away from? I think a lot of that comes down to decisions you make in the product, thinking past the immediate opportunity. We've seen the Ethereum Foundation come out with its latest roadmap clarifying its priorities, and without weighing in too much on the specifics, they were cypherpunk priorities: privacy, permissionlessness. That's one reason we're still building on Ethereum and not looking at some of the new corporate chains coming out. We're long-term believers in Ethereum's vision. As a protocol, yes, we absolutely want these institutions to come in and drive success in our business, and more liquidity actually leads to better outcomes for all the other participants. If bad debt needs to be liquidated, you have a really deep secondary market where that can happen, and better outcomes for lenders, whether it's me, you or Tommy Joe from the bank. But there are a lot of little decisions where you have to think past an immediate opportunity that might be very exciting and great for your business, because long term it could be the death of your business, or the reason people don't want to use you in the future. As we've onboarded these tokenized money market funds and other things, we really have to think about whether people will be able to get liquidity, and what happens if an API gets shut down where some centralized node has functionality in the protocol. We make sure there are fail-safes so you're not dead in the water.

Hodl Up: As you mentioned, institutions must come on-chain if we want mass adoption. That's how it's going to be. In my eyes, there will be a merging of the digital economy and the real economy into digital finance. But do you think we can do that while staying true to the ethos of crypto? Like you said, the Ethereum Foundation just released its newest roadmap and claimed it wants to stay true to that ethos, with privacy, decentralization and so on. But you can see what's happening across the industry, not just Ethereum: VC money is going to companies that are more centralized and have more control over their communities and the technology. Do you really believe we can maintain the original ethos, Satoshi's ethos, let's say?

David: First, the asset I hold the most of is Bitcoin, and at Exodus most of our customers held Bitcoin, so personally I'm certainly in that camp. But it would be naive to say it won't change. It will, and it won't change uniformly for every team, project and application. Some of the things we held to be part of the permissionless model, the DAO model for example: we now have about four years of data on it, and frankly it's kind of a disaster. People exit projects over time, as will always happen, and you end up with structures that are really difficult to run like a real business, with clear leadership. Some things worked before crypto existed for a reason. It's common sense: any organization needs to run efficiently, and sometimes that means some centralization. Certain things have been tried, and maybe versions of them will be tried again and be really successful, but it's certainly not one-size-fits-all. The space is still young enough that we'll keep seeing that evolution, and we'll have to see how much we're able to preserve. We'll certainly see assets that maintain it, I think, and then we'll see where the rest of the space goes. But it's a big change for sure, with the institutions coming in. It's a whole different world today than it was three years ago, right?

Hodl Up: Yeah. Suits and ties everywhere, so it's a bit different. Have you heard of the futarchy governance model?

David: I have. My financial engineer, Anthony, is a big fan. I'm not as familiar with it.

Hodl Up: Since you're not familiar with it, I won't go into it, but it's a really cool concept of how DAO governance can work, and it should probably be tested at scale. But I want to get into your company: what three things make your protocol better than the rest when it comes to institutions? How would you convince them to use you instead of someone else?

David: The first is the range of borrowing rates. If you're an institutional borrower, you may be putting on different kinds of trades, where, as we discussed, more junior tranches are ideal for specific kinds of trades. For market-neutral, delta-neutral basis trades, you can find an ROE-optimized point on our curve as a borrower, and you can refinance along that curve. So the point that was best for the trade you were running yesterday might change, and you can refinance some of it to a different part of the curve, which optimizes your returns. That's just not possible anywhere else today. At the other end of the spectrum, we're seeing a big push right now for fixed-rate borrowing in DeFi. Why fixed rate? Fixed rate will always be a bit more expensive, because you pay a premium for the lock-in, or the lender expects a premium for it. It's interesting, because when I worked a lot on the exchange business and integrated different RFQ options in an aggregator, we always had a floating-rate or fixed-rate option, and we never went with fixed, because with a fixed rate you pay extra for the stability. Now, with on-chain credit markets, especially while they're still emerging and liquidity can be fragmented, you see a lot of rate volatility. Going back to your comment on agents: the optimizers and agents get hold of capital and move it around very quickly, because that's their job. They're out there to optimize yield. But with the way some of these interest rate models work, you get tremendous rate volatility, so volatile that a borrower, an institution that needs to stay within certain deviations in the risk model for its strategies, can't take on that risk, or has to dial down other strategies to take it on. Because we've segmented things along a precise risk curve, when you get to the most senior parts of our markets, where you'd borrow at a more conservative level, what you get is basically the funds rate plus a credit spread. The more senior you are, the smaller that spread, and it's the spread that's volatile, while money market funds and treasuries move much more slowly. That gives you floating-rate borrowing that's very stable, without paying the fixed-rate premium. So we're excited about that. The third thing I'm really excited about is at the vault level: the idea that I can have differentiated vaults that go out to fewer markets, lending against collateral that I know and understand and would already hold in my wallet. I don't want exposure to all these other collaterals. Being able to generate real, varied credit returns against those blue chips is going to be really special for our vaults. From my time on the consumer side of crypto: people just want yield opportunities they can understand. They don't want all the technical jargon, and they don't want to have to dig too far past the opportunity in front of them. So for us, number one is removing the incentive to have things in our markets that fail and lose people money, and number two is giving people tools to understand the risk without overwhelming them.

Hodl Up: How long ago did you start working on Lotus?

David: I think we started ideating around it 10 or 11 months ago.

Hodl Up: Because my question was going to be: why now? I know the launch date isn't locked, and it may not be Q2 at all, for example. But what needs to happen in 2026 for Lotus Protocol to be a success? And where do you see Lotus by the next EthCC?

David: That's a great question. We should launch in June, so we're trying to hit Q2 for you. I think a really successful outcome at launch and through the rest of the year is that we prove this model: our curves really come to life, and people can see that they produce the endogenous returns in our credit markets that I'm describing, and that you can get these higher rates. Once we see that, and once borrowers see the other benefits I've described, I think we'll scale up pretty fast, especially with some of our institutional partners, like FalconX and Dialectic. We have some big names that can move money around quickly for us. By the next EthCC, depending on the month, maybe three weeks later than this year...

Hodl Up: That would be just perfect.

David: We see this scaling to billions in deposits over the next 12 months. If we're at about a billion and a half in 12 months, I'd feel we're on the growth path I'd expect. From there, do we expand into different collaterals, or what else? We don't plan to expand out on the tail of collateral for a while. We want to go deep on the big markets, Bitcoin and ETH, which is where 80% of all activity still happens today, even with all the RWAs. We will have a fixed-rate option. And then we'll start doing some really interesting collateral transformation. Imagine you have three or four different wrapped versions of Bitcoin, or custodied Bitcoin, in one market. The redemption path becomes the extra risk layer: one might take a day to fully settle, while another has a liquidity constraint where you can only get a certain amount back. We can order those on a curve as well, a kind of redemption-path curve, or do it by oracle, so you could have a market ordered by a market-price oracle or a NAV oracle. There are different situations where one or the other is most appropriate. Maybe somebody is very comfortable with a market oracle and willing to take on what the market sees as a bit more risk.

Hodl Up: And a side event by the beach by next year.

David: A side event, yeah. Let's have a side event by the beach. You guys will come back.

Hodl Up: We should be invited, right? You have to invite us privately.

David: You guys are right.

Hodl Up: So here's to a successful launch. We're really happy we got to have this conversation, and glad to host you. Good luck with the launch, and thank you for joining us.

David: Delighted. Great to be here.