How to Build Sustainable DeFi

Transcript

Hello, hello, nice to meet you. Today's session is about sustainable DeFi. I'm Guzus, lead DeFi engineer at Axis. Axis is a global liquidity engine for tokenized assets. We do arbitrage and liquidity market making to tackle pricing inefficiencies and fragmented liquidity. So super glad to be with the best builders in this space. Can you introduce yourself and briefly explain what your team is focusing on right now?

Starting from Derek. Hi everyone, Derek Walkush, I'm a group manager at Uniswap Labs. Uniswap Labs is the main team behind Uniswap Protocol, the largest decentralized exchange. Recently, we've been very focused on a lot of, like, asset issuers, asset deployments in Uniswap, AI integrations, and then finally institutional partnerships. Hi everybody, my name is Chris Donovan, I am the COO at Compound. Compound is one of the oldest DeFi lending protocols in crypto, and right now we're focused on building a new version of Compound, Compound V4, for what we are calling the institutional credit network, specifically focusing on institutions and institutional use cases in

DeFi. Hi, my name is Mike and I'm the founder and CEO of EtherFi, and we're building a crypto neobank. Our goal is to ultimately build the next generation of the Revoluts, the Nubanks, the Chimes of the world, but built on a DeFi foundation. Hi everyone, my name is Heejin, I'm from Radius. I work on business development and strategy for the Radius team. Radius builds infrastructure for DeFi execution and block building design.

We originally started from block building and MEV infrastructure, and these days we are expanding our business to solving and arbitrage as well. I'm really happy to be here and I'm looking forward to today's discussion. Thank you. Thank you, guys. So today's session will be about some topics I prepared. So the first topic will be incentives, then fees, then execution, and how should we respond when things break, and then finally institutions.

So let's start with incentives first. So this question is for Christopher first. So Compound is quite an OG in DeFi, right? It helped kick off DeFi Summer back in 2020 with liquidity mining. Looking back, what did the era teach you, and what's Compound's next chapter? Yeah sure. So first of all, indeed, Compound is very old. It was actually originally launched in 2013, which by crypto standards is really old. Lots of gray hair, I guess.

And indeed, in 2020, it was a big part of DeFi Summer, and it helped to, I think, create the category of DeFi and show what's possible, particularly for decentralized borrowing and lending at scale. And incentives back in 2020, in 2021, especially DeFi Summer, were a big part of fueling that for you. But I think the flip side is also evident. We're talking about sustainability and sustainable practices, which I think is not a word that is always associated, at least instinctively, with crypto. I think we get a lot of feedback from folks outside the industry that say how we operate in quite short terms.

And I think incentives are an aspect of that, where they often form an important part of product launches, and I think they will continue to do so. But I think if incentives are the only reason why users are engaging with your products for a period, then when the incentives run out, you may have an issue with retention. So I think the key thing, this is a lesson to be learned, is if incentives are the main reason that's driving growth, that often you will see a tail-off afterwards. Number one, and number two, you have to give people, your users, a reason to continue wanting to use this product as well, beyond incentives.

And that's what we're trying to do right now. You know, V4, as I mentioned, is in the process of being built, with a bunch of interesting features and use cases that I think will be very unique. And one of the reasons why Compound is a very interesting protocol, it has gone through a period of realignment, I'd say, and there's been some challenges in the ecosystem, and part of that was because the protocol was not iterated or developed over the last few years, and that means you lose your competitive edge and you lose that reason to stay within your protocol. Thank you for the answer.

Another question about incentives for Mike. So we all know that EtherFi was a really big, one of the largest liquid restaking protocols, and today you're expanding to the crypto neobank. I am actually using EtherFi really well as well. How has the way you attract and keep users changed from points to cashback? Yeah, it's kind of the end of the era. We recently announced that we turned off free staking on EtherFi.

That's just regular, boring staking. The future of the protocol really is about this broader platform, neobank, the... I don't know, I suppose, Yeah, I mean, I guess the difference in terms of the platform that we've now evolved into is that there's an actual revenue stream, like an external revenue stream, from a variety of the revenue lines that we've got. For example, on the credit card business, merchants pay about 3% every time there's someone that goes out to buy something through the EtherFi product.

And a portion of that goes to EtherFi, and some of that is then used to pay users cashback incentives. When it comes to the DeFi strategy vaults, there's a fee associated with those. Those, the lending products, again, there's a fee associated with that. So there's multiple lines of revenue that we use to grow the business. And a portion of that is paid out as incentives, and we basically think of it as an acquisition cost.

So we think, OK, what does it cost to acquire a customer? Let's say we're willing to spend, for example, $100 to acquire a customer. All right, well, that's the incentive budget. So it is, this is, you know, to speak to the theme, I guess, of this panel, this is the magic of building a product that actually has a good use beyond just the gambling mechanism, is you have revenue. You have a revenue stream that you can use to pay out the incentives. And EtherFi is now, now we've launched the token, we're in a steady state mode, where it's a closed loop, where the amount of revenue generated is more than

enough to fund all the incentives that we pay out, so there's no more of this effectively going into debt by promising a token, promising all kinds of stuff in the future. Now, the system just works. I hope more of DeFi finds a way to pay off that debt and looks like a normal business. Yeah. Great answer, thank you. So, EtherFi has an earnings feature, right? So, people, like protocols like Compound, can build on top of EtherFi.

So, the next question for Christopher will be: apps like EtherFi are building on top of lending protocols as apps, increasingly owning the customer relationship. Where should value accrue between the app and the underlying lending protocol? That's a great question. I think there's a lot of value in that stack and lots of other stacks. So I think there's often a tendency for, let's say, the infrastructure and the apps that we can buy, whether they are competitive or they can be symbiotic, and I'm quite, I definitely, in the latter camp.

We think that the direction that Compound will go in the future is to become more neutral infrastructure that can support and plug into a bunch of different user experiences, including EtherFi and others, and vice versa. And I think when it comes to what protocols and infrastructure can do well, it's handling things at scale, particularly from an execution and risk perspective in particular, that individual apps either would not want to have to take on and deal with, or they might want to do that later.

But as I said, I think there's generally speaking a pretty good balance that you can strike. Thank you. So let's move on to the next topic, about fees. So this question is for Derek. So Uniswap turned on protocol fees on V2 and V3 in December last year, and V4 in July. What has changed since then for users, LPs and Uniswap? I want to know the idea of that.

Definitely, walking through each of them, perspective of users, so traders, LPs, and then the Uniswap community as a whole. Traders has definitely been the most straightforward. There's effectively no change in economics in the protocol layer for them. And in fact, as part of UNIfication, which was the broader process of turning on fees, interface fees were turned off, along with one of our highest-grossing products, the API. That product was made completely self-service and free. We removed use of it as well. So we actually saw some of our market share grow, particularly on the API.

We were winning a lot more, like, routing from some of our integrators, because we were charging a fee on top of that before, and we saw more users return to our interface as well. So, trader impact, that one was the simplest when this process started off, and there was effectively no impact there. LPs, candidly, were the one that we before and after fees were turned on, because effectively, the way Uniswap works in the protocol layer is a certain portion of fees that normally go to our LPs are going to a TokenJar contract, which is like the UNI being burned.

We were very happy with the results, where 99% of liquidity, dollar-adjusted, remained at Uniswap after fees were turned on. And why is that? A lot of what we've been up to at Labs on the LP side has been market structure improvements, and the volume our pools do is just strictly, materially higher than what you can get at other AMMs. So the impact on LPs was actually quite ideal. There was a little bit of liquidity that flaked away, but like I said, it was at 1% dollar-adjusted in all Uniswap liquidity. And I think the most exciting sort of shift in general is what's going on for Uniswap as a whole.

I mean, I remember when I first joined crypto five years ago, the Uniswap fee switch has just been a debate that's been going on for years. And I think having one of the largest protocols, having a more sustainable long-term path, generating some kind of token burn and having some sort of value-accrual mechanism, it's just net good for the space, kind of pushing forward some of these token model businesses. Because ultimately they're kind of like a new business structure that a lot of people are figuring out. I think having one example of something, which, like I said, from a trader retention and LP retention perspective, is doing relatively well, is just net good for the entire industry.

So, follow-up for Heejin, so I know that Radius is trading actively on Uniswap, so from the arbitrage side, how do DEX fee changes show up, and how do you trade? From the arbitrage perspective, changes in fees have a much bigger impact than simply the... First, they change which opportunities are actually profitable and how much we want to trade.

When fees go up, smaller price discrepancies are no longer worth capturing, and optimal trade size also becomes smaller. Second, fees affect route selection as well. This is especially important for multi-hop or cyclic arbitrage, because fees accumulate across multiple pools. Even relatively small changes in fees can significantly change which route is profitable. More recently, these fees are becoming dynamic, and this is also a really big, important change for us as well.

In V2 and V3, fees were mostly fixed, so we could take the cost and factor in the fee and calculate whether an opportunity is profitable. But with V4 hooks and other dynamic pools, the fees can change depending on the market state or blocks, so now we also have to keep recalculating the fees as they change. And this actually affects how we design our code architecture as well. So since most of our logic uses dynamic fee quotes, we often have to recalculate quotes instead of reusing the previous one.

In some cases, even if the liquidity in the pool hasn't changed, the fee can change from one block to the next as well. Which means we may need to calculate the route again. Ultimately, we need to simulate more routes more frequently. At this point, CPU capacity and latency start to determine how much of the search space we can realistically explore. For us, the fee doesn't just act as a cost; the fee factor affects opportunity size and route selection.

As fees become more dynamic, they also affect how we design the search and execution infrastructure. Thank you for the answer. Another question for Heejin. So Radius was doing block building, is what I heard, and is now doing solving and arbitrage. So what's the problem in execution in DeFi for users, and what would Radius fix? Yeah, we started to do solving on CoW Protocol and UniswapX as well these days.

So we started this solver business, like And one of the biggest things we learned from building it ourselves is that the hardest part is not necessarily finding a good path or good route. It's managing uncertainty between the moment you quote and the moment you actually execute. So gas can be one of the simplest examples. For example, a solver may look at an order and decide, I can fill at this price and still make money.

But a few seconds later, when the transaction is actually settled, gas can spike. So an order that was profitable a few seconds ago suddenly becomes loss- making. At that point, the solver has a few bad options. First, take the loss and fill the quote. Second, fail to fill it, or price the risk into the quote by adding a wider margin. Some solvers may even stop taking small orders as well.

So ultimately users may get worse prices or less reliable fills. And there's a similar problem in cross-chain execution as well. A solver may observe a deposit on the source chain and use its own inventory to fill it, and fill the user on the destination chain. But if the solver assumes finality too early and the source chain reorgs, the solver may have already paid the user on the destination, while the state it expected on the source chain disappears.

So the question is not simply how many blocks we should wait. Solvers need different confirmation policies depending on the chain, block time, finality model, reorg risk, order size and potential risk as well. So there is always a trade-off. The faster you fill, the better the user experience and the more competitive you are, but solvers have to take more risk as well. But if you wait longer, you are safer, but execution quality gets worse.

So I think the key is to make execution more predictable. Solvers need better ways to manage cost, finality liquidity capital and risk. Because good execution is just finding the best route, it's actually delivering what you promised to the user. Thank you. Speaking of the Uniswap V4 hooks from Heejin earlier, can you explain the StablePair hook in Uniswap, which was released recently?

How does it manage between and a... Yes. So StablePair is a new release for stable-stable liquidity. I think an important backdrop to releasing it was, earlier this year, our market share for EVM stable-stable swaps was about 30%. Now it's about 65%. And a major driver of that has been these new releases and market structure that just make it better to do stable-stable on Uniswap. And StablePair was one of these, released about a month ago. StablePair is a dynamic fee hook that effectively, when the pool drifts off peg, increases fees.

So when an arbitrageur comes in to trade through the pool and bring it back to peg, LPs are able to earn more fees at that time. And effectively what it's been able to do is increase LP returns. It's been up for about a month, so the data is still very early, but by about 20-40% relative to the benchmarks, which is very powerful. So you can effectively think of it as a hook that takes some of the earnings of an arbitrageur or a solver and brings some of it to retail LPs through the pool. So long term, we think this is very exciting.

Over the past week, I think, one of these pools did about a billion in volume. They continue to grow and count very well. And like I said, it's very exciting, kind of putting it against the backdrop of growing market share in stable-stable for us. Thank you. So the next topic would be when things break. So earlier on Aave, there was a Kelp DAO exploit. It was quite a stress test for the whole stack and protocols, and EtherFi also stepped in.

When things break, who should absorb the loss, and how should we intervene, is my question. This is for every panelist. So really a joint conversation. Yeah, I don't know if everyone knows exactly what happened, so just to give folks a quick primer. I guess it was a few months ago now, there was a big hack, reportedly the North Koreans decided to go after Kelp, specifically there was a bridge hack, I think they stole about $300 million.

And the dangerous part about this particular hack was that the hackers took $300 million, but they encumbered about $1.5 billion of staked ETH, because the asset became basically insolvent. And so the entire $1.5 billion pool was basically locked up. No redemptions, no transfers. And then there was another, it was sort of a cascade.

$300 million locked up $1.5 billion, which in turn locked up $30 billion in Aave. Because as soon as the collateral asset became insolvent, well, then the entire Aave market got locked up because people pulled their liquidity. So it was just this kind of cascade that meant $30 billion dollars, like, that's enough of a chunk of DeFi, having the entire thing... And so thankfully, of course, Aave, EtherFi and a number of others stepped in, and just out of an obligation and the desire, I guess, for sort of self- preservation,

stepped in and contributed capital, contributed effort in order to ensure that DeFi didn't blow up, because this could have been, I mean, as bad as or even worse than an FTX or Terra Luna type blow-up, because, you know, the scale is quite large here. So yeah, I don't know, with EtherFi we just felt like we had to do something. There wasn't enough momentum, and the default path was that the entire system would have just got locked up in years of bankruptcy litigation, so we, we had an obligation to, you know, to step in and do something. I'll add, I mean, from Compound's perspective, I mean, we also

contributed to the DeFi effort and I think it was a great testament to how much we achieve when we work together as an industry, instead of so fiercely competing against one another with a lot of zero-sum thinking, often in crypto, and I think that's just generally doesn't provide great outcomes for us as a whole. We still have so much of global financial markets still out there to capture, and I think generally we'd be better suited working together. I think also taking a step back, when we think about, you know, absorbing losses and whatnot, in an ideal world, losses like the rsETH situation wouldn't happen, right? But of course, in today's current environment,

especially when actors are all AI-enabled and much more sophisticated than we were. Even six months ago, it is a reality that we all have to live with. So a few ways that we're thinking about it: number one, you can approach product design and development with a very much risk-management mindset. You can try and build some of these protections and principles into your products at the product level. And as I said, in an ideal world, if they do arise, then they're limited or they're isolated and they don't create the kind of contagion that Mike was just mentioning, that can become very kind of systemic.

And so, you know, some ways you could do that with things like isolated markets or very careful curation around the types of collateral, for example, within a platform. That's definitely a big part of how we've been thinking about it, and how institutions can get more comfortable with onchain exposure when they're able to audit and understand the collateral that's involved. And I think in general, moving forward, whilst the attackers have access to all this fantastic technology, that means they can be more sophisticated, more aggressive.

We as builders, we also have access to that same technology too. And I think there are ways in which you can leverage that technology, both from a building and an auditing perspective, but also threat response as well. So that's another thing that we're continuing to migrate to. I have too much to add, I just echo what both of them have said, especially the point that a hack, even on a competitor, is just negative for the space. Just thinking of the markets that we're attacking, having any extra scrutiny on the security side of the industry is just negative for everyone.

So I think it's been great to see, you know, moments of, like, hardship, like Kelp and some of the recent ones, people in the industry coming out there that otherwise would probably not be in the same room. I feel that, like I said, we're just not going to put the other pieces as well. I think you guys all said very important things as well. So I'm not an expert in this area, but my basic view is that parties that are in the best position to control a particular risk should also take more responsibility for the risk.

So that doesn't necessarily mean they should absorb every dollar of loss, but responsibility should follow control. So if you're in the best position to prevent or mitigate risk, I think you should also have stronger responsibility when something goes wrong. This is my small opinion on this. Thank you, everyone. So, because we are running out of time, this is the final question for today. What should builders in this room build?

You can answer with one sentence. Starting from Derek. Starting from Heejin. I personally think that hooks like StablePair are one of the very interesting things. Of course, we can bring the existing 20 -25 product launching as well, but I think a StablePair-like hook contract is one of the new innovations with which we can really structure and design the financial market as well, so I think this is the... we can be the first to try it in the

setting. Uniswap shilling. I'll flip the question and say what you shouldn't build is another casino. Stop building casinos. Everything in crypto is just a casino with extra steps. So build apps that actually generate value and external revenue streams. I think that's the way to really make the space grow. I would say everything is a casino, casinos that's very true.

And unfortunately, I don't think the casino space casino-like stuff. I agree with what both of you said already, and I reckon that Mike's going to have a hard time going last. But I would say build anything that enables a use case that you can do onchain, that you can't do offchain. That's how we differentiate ourselves, that's how we build products with lasting value that people want to continue using regardless of what the incentives are. Very much agree with Mike and Chris. More tactically, based on what we're seeing, is tokenized stock volume on Uniswap has been a massive bright spot over the past three months for us.

I think we're at 20 billion on Robinhood Chain, 2 billion on X Layer, and a lot more in there as well. And one product category that I think we're very eager to work with people on is just better interfaces, especially, like, different geographies for accessing some of the tokenized stocks on Uniswap Wallet. And I think that's a great opportunity. Thank you, everyone. So give them a big applause.