《让 DeFi 再伟大一次》｜Adrian Cachinero Vasiljevic & Luca Prosperi

Everybody loves stablecoins because they’re safer than a bank, faster than a bank, easier to transact with, cheaper, and hopefully they're gonna get some yield or you get some meal from a back door. On this, we are so much better, and there’s so much stuff to do because now we’re revolutionizing the banks. The banks hate us because now we’re eating their lunch. There are things where crypto is so much better.

Maybe we’re not going to be better in so many other things, where there are other primitives that are better than crypto. I think this idea of crypto-maxing the world, where everything is going to go on decentralized rails, probably isn’t true. Crypto is a great distribution channel because everything is permissionless and global from day one. You have a great product, and suddenly you put it in the market and everybody can buy it. But that doesn’t necessarily mean that we need to have every product in the world on crypto rails.

Nothing said on empire is a recommendation to buy or sell any investments or products. This podcast is forformational purposes only and the views expressed by anyone on the show are solely their opinions, not financial advice or necessarily the views of Block Works. Our hosts, guests, and the Blockworks team may hold positions in the company's funds or projects discussed.

These gentlemen need no introduction, but I’ll do it anyway for folks who have been hiding behind a rock and not following the discussion on Twitter. Luca from M^0 and Adrian from Steakhouse. Guys, spend a minute giving some context before we jump straight into it.

I’ll start. Great spending time together, guys. I’m one of the co-founders and CEO of a company called M^0. We are one of the most active stablecoin infrastructure companies. We build on-chain infrastructure for stablecoins that want to be issued and used on-chain, with a big on-chain skew.

I’ve been researching and writing about DeFi for a long time, way before my time at M^0, on a Substack called Dirt Roads, where I nerded out every now and then. Since my Maker days—and Adrian and I go a long way in crypto years, since the time at Maker when we were trying to connect Maker’s balance sheet with the real world of lending—it’s been great chatting.

I’m Adrian. I’m a co-founder at Steakhouse. Like Luca mentioned, we started as contributors to Maker: first as the Strategic Finance Core Unit, then the Real-World Finance Core Unit, then the Strategic Finance Core Unit, and then the Real-World Assets or Finance Unit. There were a number of initiatives that had to do with the real world and with finance in some regard.

Since then, we spun out an advisory business that has contributed to protocols like Lido, ENS, and Arbitrum. Since at least the launch of Morpho in January 2024, we’ve established ourselves as the leading Morpho curator, operating about $2 billion—slightly under $2 billion—in non-custodial deposits across different vaults, mostly on Morpho, but including Camino and Solana.

We also run one of the Sky ecosystem’s sub-balance sheets, otherwise known as Stars, called Grove, with a mandate to build out credit infrastructure for the purpose of asset-liability management on behalf of a stablecoin. I know both of you, so it’s nice to see you again.

1. Why Are Rates in DeFi So Low?

Great, thanks, guys. Look, we want to have this conversation because the backdrop for this discussion is that rates in DeFi are very low. Not all vaults are created equal, but vaults are all the rage. Morpho obviously plays a huge part in that. Adrian’s been key to that.

A big value proposition of DeFi is bringing more and more assets on-chain. But when you’re getting paid close to or below the risk-free rate—meaning Treasuries—it really begs the question: What are we doing here? I think we can all agree that there is way more risk in DeFi, some of which we may not even fully appreciate, with AI progressing quite a bit.

I think we all saw the announcement from Anthropic about its ability to find bugs in the codebases of well-capitalized and far more capitalized companies. So there’s still a lot of risk in DeFi. Nonetheless, we’re all here because we believe in the promise of DeFi. There are some interesting properties that make it unique.

I want to unpack first why rates are so low in DeFi. Let’s start there before we go into how we fix it and what some interesting strategies are. Why are rates 2% to 4% in DeFi?

I can kick off, and I’m interested to hear Luca’s input on this. I think DeFi has matured to a point where it’s worth drilling down into exactly what we’re referring to. I agree that, generally, rates are low across the board, regardless of the risk of the collateral type or the type of instrument.

But if you anchor on something like overcollateralized Bitcoin lending, or loans against overcollateralized ETH, that’s where you see rates at, near, or slightly under SOFR—basically trending around the traditional-finance risk-free rate. I think we would generally expect this, first, and second, see this as an example of DeFi’s maturation. So we’re not necessarily super stressed about it.

The reason for this is worth explaining, so I’ll hog the mic for a second and give you the Steakhouse vision. Our view of vaults is that they work. It’s not necessarily just about randomly bringing capital on-chain; it’s about recognizing the unique value propositions of crypto and DeFi and maximizing that surface area.

We make a distinction between what we call crypto guarantees and social guarantees. Crypto guarantees would be examples of something simple, like a rule or a very simple set of parameters, where execution is enforced by immutable code or smart contracts in a very minimalistic way: liquidation thresholds, NAV accounting, and things like this. Basically, anything that can’t be overridden by anyone and is sort of the purpose of crypto anyway.

You need these guarantees because, although Ethereum has, I don’t know, half an hour to get finality, there is de facto instant finality. When a transaction executes, it’s basically gone. Nobody is trying to reverse the blockchain, saying, “Quick, we have to do it before finality.” Once it happens, it’s out there, and people assume it’s just going to finalize.

That’s the reason you need these very strong crypto guarantees. We view vaults as a way of expanding the remit of all the things that can be cryptographically guaranteed. To come back to your point about why rates are low, a vault that only lends to overcollateralized ETH loans comes very close to maximizing the surface of crypto guarantees that are possible on that specific instrument. You would really expect it to be very near SOFR.

But in everything else that we call social guarantees, it’s obviously far more complex. It may be the more interesting area to dig into, because then you get into operational security, smart-contract risks, different types of economic primitives and protocols, trust in counterparties, and so on. The levers for risk multiply ad infinitum. There’s a legitimate question of what rate it’s at and at what rate it’s actually worth participating in these.

My TL;DR answer for why rates are so low is that I have an uncorroborated suspicion that, because all of DeFi lending has to work on supply-and-demand incentive equilibrium, the cost of capital for DeFi to go on-chain has actually spiked in recent days. There are more risks, more hacks, more macro uncertainty, and so forth.

Ironically, this tilts the incentive back down toward borrowers. It has the effect of making rates lower because people are using it less, lending less, and borrowing less because there’s more risk. Depending on the context and the protocol, this may have some perverse effects.

I’ll go next. Let’s start with the headline: Why are rates so low in DeFi? Ultimately, I think the reason is that DeFi’s capital markets are disconnected from traditional ones. You have money in DeFi, and you want to keep it in DeFi for whatever reason—convenience, because DeFi is optimized for trading or leveraged trading, taxes, KYC, or whatever.

A lot has been written about this convenience. It’s easier for you to plug in and out, put it in a Morpho vault, and have it be instant and atomic, instead of just buying a Treasury on-chain or off-chain. I think there’s a disconnection between the markets, and that’s what’s making the yield so low.

The question I wanted to unpack when I started writing about this is that I have no horse in this race. I’m a very good friend of, and indirectly an investor in, the Morpho product. I think these guys have been amazing, including the Steakhouse guys, who are partnering with me on many other initiatives. But I like to look at things, try to understand what’s going on, and have a framework.

In my opinion, I would split this into 2 parts. The first one is: What are lending markets actually doing? Jason, you were saying that vaults are all the rage, and we see, “Oh my God, the whole capital markets are coming on-chain, and vaults are going to intermediate credit around the world,” et cetera, et cetera.

Is this true? Absolutely not. Do vaults have all the required infrastructure to actually avoid rug pulls or adverse selection of credit on-chain? Absolutely not. Is it going to happen in the future? Hopefully, hopefully, but this is not where we are right now.

DeFi is great at one thing, which is overcollateralized lending on crypto collateral, as Adrian was talking about, or looping. These are the 2 great things DeFi is doing. If you look at Morpho TVL, this is not 100% of the TVL. I think the question here is, first of all, let’s look at things for what they are so that we can build better things. The Steakhouse guys are absolutely in the same camp, I’m sure. The second one is trying to understand where it makes sense to actually participate in the current market, right?

Given that this is the best we can do, I think it depends, right? If you’re a whale, if you’re a large DeFi protocol and you have money stuck in DeFi, this is the best alternative—it’s the safest alternative you have. It’s still not risk-free because there are a lot of unknown unknowns. We’ve seen liquidation cascades not go well in the past.

The problem becomes when you’re connecting this to CeFi retail front ends, and de facto this is becoming a very cheap and easy-to-use, crypto-native deposit account. Then there’s a risk of misleading the user and actually attracting users or exposing users to risks that they don’t really understand. That’s why I wanted to point the finger, because it’s good to have smart conversations and understand exactly what’s going on.

There’s a reason why yields are so low. I think I’m in the same camp: it’s a supply-and-demand thing. There’s a lot more demand than supply.

Is that so? I agree with that. I think that’s my diagnosis too, and I think it’s been like that since rates went up in 2021. You’ve just seen a lot of participants—well, actually, I don’t know what you would attribute that to. My assessment is that a lot of the strategies just became less interesting when you could capture a pretty nice rate off-chain, so there’s a whole set of participants that, over the course of the last 4 years, have decided to exit the market for a variety of reasons. You’ve had blowups like Terra/Luna, and you’ve had the off-chain world just be more interesting.

When you guys see this—trying to decompose the supply and demand—where is the demand coming from? What is that? Is that mostly retail? I’d be curious. Maybe I’ll go with you, Adrian: where are you seeing most of the demand on-chain come from today?

Until we partnered with Coinbase to power their DeFi Lend integration, we had a very small number of depositors and a very large TVL. That was suggestive of the focus we had on working with integration partners: supplying, let’s say, a custodian of stablecoins or an exchange and so forth access to these prime vaults, or large users and funds that were parking capital for a period of time between strategies.

Then there’s a long tail of retail users, whatever the distribution of wealth is.

Sorry if I interrupt. And protocols—I think Maker, before the Coinbase integration, was probably 30–40% of Morpho liquidity.

Yeah, that’s true. Maker had that through the Spark curator. Now, with Grove and Spark, they kind of dip in and out depending on the rates. It’s true that, among stablecoin reserves, staked USDC was a reserve asset for Angle Protocol. I don’t know if you remember this amazing stablecoin that went down, unfortunately, a few months ago.

So, yeah, there’s also some DeFi-native B2B at play, and I think the Morpho construct lends itself well. You were making a point earlier, Luca, about how the infrastructure is not there yet, and I generally agree. I do think there are some bright spots, though, and the Morpho vault—the Morpho philosophy to the vault—is, I think, the right direction.

Vaults, essentially—and forgive any vault enjoyers—fundamentally have to track the NAV and not get hacked. It’s mostly around downside avoidance, and the less they do, the better. That’s been Morpho’s philosophy since the outset: trying to find the smallest primitive, maximizing the surface of things that you can secure in a cryptographic way, and then kind of releasing the brakes. There are nuances around how you configure those, but I do think that’s the right approach.

With Coinbase DeFi Lend, that’s where we saw the first influx of tens of thousands of retail depositors. But Coinbase is actually quite a good example of a good actor and how these integrations are made. I don’t know if you’ve had the chance to use this application, but they don’t call it a deposit account. They don’t call it a savings account. They call it DeFi Lend.

There’s a very long FAQ. They explain bad debt, what can go wrong, and so forth. This is for a prime vault; they’re mostly lending to cbBTC borrowers on Coinbase as well, for that matter. There’s obviously a spectrum of how you can market these. We ourselves go out of our way to say this is the closest proxy to the DeFi risk-free rate, but it’s obviously not risk-free.

Luca, how do you see it? I mean, I hear you. A lot of the demand is coming from exchanges that, unlike the last cycle, purported that these algo stablecoins were just sort of dollars and the yield was really interesting. I think now we’re doing a better job of disclosures.

Luca, I read your TL;DR article, which I think everyone should go check out, and I think you just did a great analysis around this: the yields are just not there. They’re so-so. Maybe give a little bit of context, and then we’d love your take on explaining the difference between Morpho and prior instantiations of yield farming, like Yearn. What has made Morpho more interesting, and how do you see the risk of interacting with someone like Morpho or other protocols?

Yeah, first of all, I want to start with a couple of comments. The first one is that we are in a much better position compared to 4 years ago. To Adrian’s point, I think the work that the guys at Morpho have done is phenomenal because they’ve isolated simple primitives, and they’re building only one primitive on top of the other.

The only thing I’m saying is that there’s a long way to go before we start incorporating more sophisticated assets. The thing that works in crypto is still overcollateralized lending on crypto stuff, or looping, which could be a different strategy. The direction is absolutely there, and we are in a much better position. I don’t think we’re misleading retail at all in any way that’s similar to the Terra days or the CeFi days in general.

2. How To Mitigate Risk In DeFi

And then, Jason, a short comment on what you said: I think the lack of interesting, higher-yielding opportunities in DeFi in the last 4 years is true. I think there was one bright spot, which was Ethena. Ethena came out with a very good idea of tokenizing the basis trade in a way that was high-yielding enough to attract a lot of customers. It was a short-lived trade, but I think it was a very good idea, and the guys executed very well.

I wrote this article for my Dirt Roads newsletter on Luna Park last week because I wanted to try to decompose risk and understand exactly what we were underwriting. This is not my day job. I do this over weekends and nights. A lot of people came out criticizing the parameterization and criticizing the model, and I loved it because that was exactly what I wanted: to start a smart conversation about this stuff instead of just doing crypto Twitter wars between Morpho and stuff like that. It’s great that we’re here.

What I wanted to say is: what are we actually underwriting? What is the risk that borrowers have? I focused on market risk. There’s a lot more that I’m going to touch on in a second, but, to simplify things—to brute simplicity—for these vaults, the prime vaults Adrian was talking about, you’re putting money in a vault in the form of a stablecoin, and somebody else on the other side is depositing collateral like Bitcoin or ETH and taking out a loan that’s significantly overcollateralized. There’s way more Bitcoin than the loan you take out.

If the price of Bitcoin and ETH goes down below a certain threshold, which has a very healthy buffer, a liquidation cascade happens. The collateral is liquidated, and you get repaid. The reason why the yields are so low today is that, first of all, these systems are extremely battle-tested, and the liquidation algorithms have been tested. At least the latest vintage hasn’t been catastrophically affected in the past; past vintages, yes—we remember Maker. The algorithms work, and there’s much more supply than demand.

Everybody’s depositing. Not many people are taking money out because there isn’t a lot of demand for leverage on Bitcoin and ETH these days, right? I think that’s one of the reasons. The reason why the yields are so low is that the system works very well.

The way Morpho does it differently from even Aave is that—and I think that was absolutely a winner for the team—they inherited it spiritually from another team called Rari back in the day, where they had been isolating those markets. Instead of having one big pot where all the collaterals are there and there’s a DAO or somebody doing risk management, they’re isolating these markets. These markets are very simple; you understand them and can analyze them very well.

Then they’re simplifying it and creating another layer called the vaults, where there are asset managers like Steakhouse or Gauntlet doing the risk management across those markets. The composability is there. I haven’t written about the vaults yet; I have another piece coming on vaults as soon as I have some time, in the next few weeks. But I think the way the Morpho system works today, to Adrian’s point, is that it has super-simple markets—simple-to-understand markets compared to the past.

You don’t need to get lost in the DeFi labyrinth of interconnected things. That’s also why I think it’s fair to say that the risk is relatively low. Now, in my opinion, it shouldn’t be so low, and I think it’s being pushed artificially low because there’s way too much supply compared with demand. There’s some inherent risk embedded in this: if the market moves abruptly, ultimately you’re underwriting some volatility in the crypto assets, which is not risk-free.

There are other risks included, like hack risk, smart-contract risk, and catastrophic risks that can happen, which we’re not parameterizing. So this stuff is not risk-free. Interestingly, the reason why we do not have a true risk-free rate on-chain is that the regulator is avoiding it. We cannot stream yield from stablecoins.

If USDC, or whatever stablecoin partner M^0, could stream yield continuously for retail investors, this would be pure risk-free, but you cannot do it. Interestingly enough, what is happening is that Circle is making money and Coinbase is making money. Retail investors and borrowers are underwriting the risk. That’s what is happening.

We could do much better, but in no way are we in a position comparable to where we were before. This only focuses on overcollateralized, prime, super-safe vaults. When you start doing something more exotic, it’s different. When you have loops, like when Morpho spiked in TVL when Ethena and Maker were looping USDe yield, that’s different, right? Then you’re putting leverage on leverage.

Even real-world-asset looping is different. There’s no free lunch. When yields are 20% or 30%, it’s a different story. But to your point, I’ve seen your comment on one of my research pieces: Yes, I know that these are all the dynamics, but if you’re a sophisticated investor, are you getting paid for the risk you’re underwriting?

3. ZKsync Ad

I wouldn’t put my money in personally because I don’t think it’s enough. I would just park it and probably do a barbell with Bitcoin, but this is my own personal money. I think everybody should ask whether, based on the market conditions, the money, and the emerging yields, it’s enough for the risk they’re underwriting. Everybody has a personal view on this, of course, but they should have a view.

ZKYNC is the bank stack of Ethereum. ZKSync's Pervidium is the only Ethereum secured platform that is purpose-built for institutions that demand privacy, compliance, and full control of their data. ZKSYNC enables financial institutions to thrive in the digital assets economy by launching and monetizing their own chains, tokenize real world assets, build programmable markets, and settle crossborder in real time. Visit zksync.io to learn more about the bank stack of Ethereum and book a demo. As always, investments in blockchain technology involve risk terms and conditions apply.

4. The Biggest Risk In DeFi Today

So, just to summarize before we go any further, it sounds like a lot of the demand is retail. You can attribute most of the compression and low yields to a lack of demand. As I was hearing you say that, I thought about short sellers. If you own a stock and there’s not a lot of short interest, you’re not going to get paid much for lending out your stock.

I think that’s sort of the rut where crypto is. You have a lot of these long-term HODLers that have Bitcoin and Ethereum and are saying, “Okay, I’m going to deposit into a lending protocol, so I might as well clip some yield, and I don’t necessarily care. It’s just extra.”

I think Luca, you and I agree—or maybe Adrian, I don’t know if you would—that there’s still some tail risk in the protocol that you’re interacting with. As battle-tested as it is and as good as the liquidations are, there’s still a noninsignificant probability of impairment and total principal loss, smart-contract hacks, OBSC issues, and other problems.

If you go to Aave V4, there’s just more surface area. When you factor that into whatever sophisticated math or simple math you want to do, it’s a very tricky situation. It reminds me a lot of staking yield, right? Adrian, you mentioned that, right? That could be the risk-free rate; you should be staking your assets.

I’ve heard so many institutions say, “We actually don’t want to stake because there’s a nonzero probability of our assets being locked,” or what have you. I’m not holding this asset for the incremental 2%, 3%, 4%, or 6%. I’m holding this asset because it has a venture-like return outcome. So, anyway, there’s a lot there, but what is the solution to this problem?

I think the solution—so, I agree: in DeFi, this is kind of what I love about DeFi. We can tell ourselves, de facto, that the Treasury rate is essentially risk-free, right? Within the framework of the U.S. monetary system, it may as well be because of the ability to create new dollars. Within that fixed frame of reference, every fiat currency is risk-free—the dollar more than other currencies, to be sure.

What I personally like about DeFi intellectually is that it brings all of these issues to the surface. They get complicated when you conflate them and when it becomes difficult to sort out one thing from the other. To your example, if I put my Bitcoin in Aave to earn 2 basis points, what are the risks that I’m implicitly underwriting there?

I agree that it’s more complicated to parse that because it’s a more complicated protocol. It’s bigger and has LTV, Umbrella, and a bunch of other features, but those are all remediations to the central fact that it’s just a design that’s a bit more aggregated and a bit more difficult to parse out.

Not to be the eternal Morpho apologist, but I think the advantage of doing things with simpler primitives is that it makes it easier to isolate what the risks are. I think this may well be true for high-yield repo lending, but for Bitcoin and ETH overcollateralized lending, I think it overstates it because I do think we’re nearing a point of close to efficiency in Bitcoin and ETH overcollateralized lending.

For sure, there is still risk. A neat example is WETH-USDC overcollateralized loans, because the collateral is about as cypherpunk as it can get. The WETH contract was deployed by—was it Nikolai Mushegian?—and is signature Nikolai: unchangeable, immutable, and completely bulletproof.

The only real risk there is, I suppose, if USDC faced some kind of impairment for whatever reason. So it’s not immaterial, but it’s not specific to lending. Or if the oracle for the ETH-to-USD price suffered a loss. ETH would have to gap 60% in a block, in a single block—we’re talking about very volatile price gaps, which are not impossible but are very unlikely in our view.

Alternatively, the oracle network would have to fail, and Chainlink would have to shut down or have some kind of malicious behavior. Again, it’s not risk-free, right? Our solution is this unbundling to rebundle, which I think is very helpful.

In high-yield repo lending, you have more of the effects of what you’re describing. I also don’t necessarily agree with the characterization of a short sale or a put, particularly for the prime vaults, because there is so much credit enhancement built into the structure of the market and the protocol. It does start to take on more of those characteristics with higher-risk collateral.

What we see in the vault space, in particular, if you’re restricted to underwriting repo markets against different types of collateral, is the adverse incentive: as a curator, I want to print the highest APY, so I’m going to add more and more stuff randomly. The lender is not going to be commensurately rewarded for that risk, and the risk starts to take on more tail-risk-like characteristics. That’s a bad outcome in our view.

We take the view that lending should be different. The degrees of risk in lending should be a function of liquidity and market size. If a stablecoin investor wants to take on more risk, they should just take on more risk. In other words, they should allocate into vaults that expose them to leverage looping, for example, because that’s the fair price for the risk they’re taking.

In this framing, the borrower—it may sound like Steakhouse has an adversarial relationship with the borrower, and it does to a degree, because our main priority is the lenders in the vault. The borrower is actively taking risk, so if they get liquidated, it's kind of their fault. That's the risk they knew they were taking, and they were pursuing a reward greater than what they could have achieved by lending into the same vault.

So, in brief, that's more on bundling and unpacking what the different levels of risk are, and relying more on crypto guarantees than on social guarantees to secure those. But once you get to certain levels of risk—the bulk of our volume is on lending, but we have a small program to try and launch these leverage-looping vaults, term vaults, and things like this—they're risky, and we're not hiding that. That's what we're trying to do: say, for different tiers of risk exposure, what is the fair rate that investors will demand to get in? If you work on that market mechanism, and if the industry trends toward that market mechanism over time, I think we'll get to a point where those risks are adequately compensated.

Yeah. I wanted to touch on a few of those points. One is actually against the argument I made at the beginning. The first one is: selling a put or not—is this a repo or is it an option? I don't think it's really relevant. The reason why, for example, I brought up—and I don't want to nerd out, I promise—the Merton model is that everything is volatility and return. The only way to model a credit instrument that is not a continuous-payout instrument is to model it as risk-free plus an option. It's just a way to model it; it's a mathematical framework. It's not that it's a very deep-in-the-money option, so it doesn't look like an option or smell like an option. The option is there, and there's more volatility in it.

Again, is it enough? Are you paid enough or not paid enough? It's always in the eyes of the lender. Ultimately, you do your own underwriting. You decide what the probability is, do your own modeling, and then act upon it. We don't have enough primitives. We don't have insurance primitives in crypto for tail risk. Unfortunately, it's difficult to price, but we're getting there.

I think we're at least perfecting the base case. We've been at this for about 5 years—less than 10 years, which is nothing. The truth is, we've perfected the base case. We are far away from perfecting the most sophisticated stuff, especially when social proofs are important. There is so much talk about lending and private credit in crypto. We've been here before many times, and we know how this is going to play out. There are large asset managers that love it. Of course, they have a great pipeline to connect their products to unsophisticated investors. They shouldn't do that.

The last point that I want to make, which I think is important, is against what I was saying before. I was reading somebody writing about this on Twitter today, and it's that these protocols are so good for the lender. They're super safe for the lender, but they're brutal for the borrower. If you're sleeping and you have a position—you're borrowing money collateralized by ETH or BTC—your bots stop working, whatever happens, you get liquidated, you take a 5% liquidation cost, and crystallize your losses because the protocols are protecting themselves and the lenders.

The reason why the yield is so low to lend is because the only way you can attract borrowers is for the price these guys need to pay to be very low. If you're trying to get a BTC-backed loan from an OTC desk, it's way safer. You're not going to get liquidated overnight. These guys are going to call you and restructure, but they will ask you for much more money.

So maybe the other question is: should we build new protocols that are actually more balanced between borrower and lender, so we can have a better market? I know that Morpho is trying to do this with intent-based Morpho V2, where they are matching. But let's see if there's enough Lindy effect to actually attract liquidity.

One of the reasons why the rates are so low is that these instruments and mechanisms are brutal for the borrower, so the only way to attract borrowers is that they don't need to pay.

That's 100%. I would argue that maybe the missing piece there is the borrower. The DeFi borrowers that are using it need to be—I mean, we think about borrowers a lot, but obviously the priorities are on the lenders from the curator perspective. Everything is kind of adversarial against the borrower.

But again, if you flip the framing, it's like, look, the borrower is taking risk. You get what you pay for. It's a really good point, actually, that the low borrow cost may be the difference that compensates for the experience of borrowing on variable-rate protocols. You can do it with a bank or an OTC desk, and you get a phone call. They invite you to events and stuff. But on Morpho, you just get destroyed.

It's not only that they invite you to events. They call you and say, "Let's restructure this position," instead of just wiping out your position overnight.

Yeah, exactly. That's a huge advantage for somebody who has a long-term position. But it's also kind of the only way that you can make variable-rate repo work on-chain in 12-second increments. I do think the priority should be on the lender in this instance.

But I agree that the fixed-rate model that Morpho is experimenting with is going to be an interesting marketplace to figure out whether we can build a term structure in DeFi to compensate for this risk. We're indicatively getting some feelers from very cheeky people who want to capture the variable rate at a fixed term with all of the protections.

I do think it'll require a little bit more than the variable rate. So I expect the fixed-rate term structure at equilibrium to settle above the variable rate in that instance, if the market is efficient. The way that this would collapse is if a curator underwrites too much illiquidity for too little yield, but then they would see an outflow from their vaults, so the market will self-correct.

I want to unpack that a little bit. First of all, the point you're making is that there's an asymmetry: the borrower is taking way more risk because it's 24/7 finance. Someone's calling you for a margin call. But give me the bear case for the lender, because I think, Luca, you're saying, "Look, you're making it seem like the lender is taking very little risk." There's still some risk, right? Smart contract risk, especially when they're upgradable contracts.

So maybe paint the picture. I don't want to make this a doom-and-gloom podcast, but I do think it's important to set what the tail risk is, because a lot of this conversation gets interesting when you go to the tails and say, okay, what is actually the worst scenario for a borrower and the worst scenario for a lender? We've talked about the borrower, but not so much the lender.

Yeah, I can start. As I mentioned, going back to the primitive, these positions are overcollateralized, deeply in the money, and backed by super-liquid crypto collateral. This is what we're talking about on Prime, right? Again, if we talk about higher risk, that's a completely different business, but let's put it aside.

Most of this stuff is Prime on Morpho, or anywhere currently in crypto. So, significantly overcollateralized positions on super-safe and super-liquid crypto-native collateral.

Now, what is the risk? The risk is that the BTC and ETH price moves. If it moves against you, it's actually reducing in value. At some point, you're going to hit a liquidation. This stuff is actively managed, so the risk of hitting a liquidation is lower because there are people like the Steakhouse guys continuously rebalancing it, putting up more collateral or taking some money out.

There is active rebalancing, but still, money is not infinite. If prices start going down, down, down, down, the people who took the position, if it goes very quickly, don't really have perfect, infinite money and perfect timing to rebalance.

If you look at the super-safe scenario that I modeled personally on market risk, forgetting all the other risks that I'm going to touch on in a second, my models were roughly 40–45 basis points above risk-free. What we're seeing in Morpho is a bit less than that—20, 25, or 30 basis points. So what is real? Morpho is pricing a perfect scenario. Are we in a perfect scenario? I don't believe in perfection.

So probably there is some friction that hasn’t been priced in. The debate here between Adrian and me is whether this imperfection is 20 basis points or 200 or 500. There is definitely something there.

This could come from imperfect bots, operations not working, non-infinite capital, or liquidation systems not working very well because prices move super fast, gas prices spike, and everything gets congested. You cannot really move. There can be some friction in the liquidation process where the buffer you have is not enough, and if that is not enough, then you’re going to have some sort of loss-given-default case.

My model was: The LGD is 5%. If there is a default or a liquidation, you’re going to get hit by a 5% loss. Adrian’s point was that this is too harsh—it’s probably closer to zero, a few basis points, maybe 20 or 30 basis points. As usual, the truth is probably in the middle.

You need to do the calculations yourself. The good thing about doing what we think is good-quality analysis is that I put a model out, someone else can take my model and change the parameters, and you can do it yourself. You can put in the LGD that you think makes sense, and it’s going to give you the yield that you should have, whether it’s a 400-basis-point spread, a 300-basis-point spread, a 20-basis-point spread, or zero spread.

Now, this is only market risk. There are other risks, like Santi was mentioning: hacks and operational-security risk, somebody hacking something, or the multisigs not working. This stuff is not fully permissionless or fully immutable. The base level is immutable, but there are still other risks.

How do you price that? Very difficult, because this stuff is not statistical. You probably need to do some sort of probability scenario. What do you think the risk of hacking is? Five years ago, it was probably 30%; now it’s probably 1% or 0.5%. You need to do your own math.

The point here is that I think my analysis was interesting because it put at least a quantifiable floor on the risk, which I quantified at 40 or 45 basis points above risk-free, assuming everything is perfect. If everything is perfect, you should get paid SOFR plus 40 basis points. Maybe I’m wrong by a few bips, but let’s say that’s right, and then you can do your own modeling. You can change the parameters and understand exactly what you think the spread should be.

In my opinion, the spread should probably be a couple of hundred basis points if you consider everything together: market risk, hack risk, custody risk, and so on. But again, you do your own math. What I think we should do is provide good frameworks to the market so people can do their own calculations. I have Luca’s position, Adrian has his own perspective, and I’m sure these guys are using it.

The more you move up the risk profile, the more interesting these models become, because then you really need to understand what you’re underwriting. Nobody knows.

I love the framework because it goes through that unbundling approach: Let’s take the simplest possible case—the spherical cow—and model it a little better, step by step.

On prime, I think, Santi, to your question, our view is that the risk on BTC and ETH overcollateralized lending is not zero. Historically, the market risk we found is 0.94%, so it’s very, very low. That doesn’t say anything about future events.

The realistic loss risks on BTC and ETH are a failure at the collateral issuer, a failure in the oracle, or Bitcoin and Ether prices gapping by more than 20% within 12 seconds. We rate each of these risks for prime markets as relatively low—probably lower than Luca’s 100 basis points—but I would say we’re in the same order of magnitude.

Where I think this framework is illustrative is when you move out on the risk curve.

Do you think it’s worth, when you paint DeFi with a broad brush, to say that it’s all risky? You can do the same with traditional finance. You can say all private credit is risky, and therefore I’m not going to lend to a bank because it lends to mortgages.

Sure, but it is worth drilling into the details and seeing what exactly is happening. I think the discussion is not as controversial in the very narrow use case of depositing Bitcoin or Ether into a very liquid, overcollateralized market with a very good liquidation system and robust price oracles, and then having a borrower who is very adversarial.

Where it gets more amorphous is the operational-security risk that you’re taking, and that’s hard to price because it’s kind of binary in that sense. Every year, we get 1% to 3% losses in aggregate TVL, which is an unfair characterization because obviously you have to drill down to the protocol level you’re interacting with and consider the counterparty risk.

To your point, I’m more in Luca’s camp: You don’t have to approximate the risk at the basis-point level, but I think we can definitely agree that it’s probably worth a couple of hundred basis points above SOFR, no matter what you’re interacting with if you’re doing crypto.

But I disagree for prime vaults because it depends on what protocol you’re using. You do need to get specific. I don’t want to be the eternal shill for Steakhouse products, but in our vaults, if we stopped reallocating because we all got caught on a plane and the plane crashed into the ocean, the vault would perform slightly less well over time, and users would just exit. There’s no dependency on us.

You should have your private plane guys fly.

We take separate planes, just in case. But all the same, it wouldn’t matter. That’s because we research all of the possible surfaces that could introduce that risk, and Morpho is very good because the number of those surfaces is very small.

Realistically, in Morpho, the attack surface on the curator is quite minimal. It does exist, but it’s quite minimal and can be very strongly mitigated. The way we mitigate it is by making the users of the vault a check on our behavior as curators, and we use a trustless Aragon DAO to do this.

If we get compromised, for example, and onboard a fake token onto a prime vault, all of the users have a 7-day timelock, and they can stop it from happening. If there weren’t enough liquidity to simply withdraw, they could at least wrap their positions in the Aragon DAO and vote to kill the onboarding.

That gap exists in some vaults, for sure. Even two prime vaults aren’t exactly the same, and this is what makes DeFi very hard to parse. DeFi is dense, jargonistic, and full of intricate details, but for the prime-vault category in particular, I do feel that the operational-security risk is relatively constrained and concentrates mostly on the collateral issuer and the oracle network.

There’s less of it on the market side. You would have to assume that BitGo or Coinbase could suffer an operational-security exploit that would allow them to mint infinite Bitcoin, or that Chainlink could suffer an operational-security exploit that would lead it to gap the price by half. Those are certainly real possibilities, but you’d have to evaluate their likelihood.

Yeah, I wanted to say a couple of things here. The first is that risk-free is an illusion. It doesn’t exist. Risk-free is a proxy for the lowest possible risk we can have. The U.S. government can still wake up tomorrow and say, “I’m not going to pay out.” Many interesting things are happening in the U.S. these days, so this is probably not zero risk.

I’m curious, and I’m going to end with an interesting question for you guys. Before that, I think we shouldn’t paint DeFi with too broad a brush because DeFi is not great at lending. DeFi is great at doing exactly what we’re talking about: overcollateralized margin lending that is atomic and super optimized for trading and arbitrage. This is great.

This is not lending to corporations. This is not lending to mortgages. This is not private credit or lending private credit to private companies. DeFi absolutely sucks at those things in so many ways. I could start now, and we’d have five hours. I’m going to be finished by then.

But DeFi is optimized for a specific use case, so we need to talk about what we’re actually doing. I think DeFi is amazing for overcollateralized, no-KYC, super-quick, algorithmic lending that is great for trading and arbitrage. This is amazing, but it’s a small subset.

The question is: How big is the target market, the addressable market, for DeFi lending? That remains to be proven. If it’s as big as credit, we are very, very far away.

I know that the guys who take stables are optimizing for single-basis-point security in order to make this exact use case as safe as possible. But if that wasn’t your use case, what would you do? Would you buy a tokenized money market fund that gives you 3.5%, sleep on it, and forget about it, or would you put money into a prime vault?

I would buy a tokenized money market fund, of course.

I know what it is. I'm not going to get rug-pulled by prices. I don't worry about it. That's fine. Maybe they break the buck and come back.

So if you want to have risk-free lending on-chain, there are better ways to do it. If you want to do trading, leveraged trading, or arbitrage with your bots, then you cannot do it with a tokenized money-market fund because these guys do not operate in the same time frame.

5. What’s Next For DeFi In 2026 & Beyond?

But we should actually define better what we're optimizing for, because if we're thinking about DeFi lending, is the addressable market credit? Currently, it's not. Well, let's go there, because I wanted to carve out a piece of the discussion around interesting products that I think you mentioned, Adrian.

There have been, since the beginning of time, investments in creating fixed rates, swaps, and insurance. You've had Nexus Mutual and a couple of other instantiations, none of which, in my opinion, have really gotten much traction. I don't know if you want to respond to what Luca said, but I'd actually be more curious, looking forward: What are some of the things you're excited about that you think could bring more security to DeFi or attract more demand? What do you see on the horizon that is really interesting?

So I think it relates to Luca's point because, to date, DeFi has a term structure of 12 seconds, and that makes it nearly impossible to have much of a monetary economy beyond statistical trading and whatever speculative activity, right? Some amount of on-chain B2B, whatever.

The proportion fluctuates, but I agree that DeFi still has to prove itself to win credit. I do think that it can win it. We're just of the perspective that it's not going to be next year. I do see a world where traditional finance, origination, and what have you continue growing and being very big, with DeFi just behind it, catching up and maybe at a higher growth rate, sort of dependent on the degree to which newer generations of users and consumers adopt stablecoins as a natural means of payment.

It's not a super interesting answer, but I feel like we go back to this crypto-versus-social-guarantee construction. Whenever somebody talks about insurance or swaps or whatever, it's always possible to build Rube Goldberg machines and gadgets in crypto. I think crypto is really good for this space because it's the perfect intersection—the Venn diagram that intersects perfectly between mathematicians and scammers.

Both of them love very complicated Rube Goldberg machines that have bells and whistles and complicated models. But the models that are going to be the most successful are the ones that lean on the crypto guarantees to make the product better, more transparent, and safer, and lean on the guarantees of operating on public blockchains as a value proposition.

When you look at that today, it's true that the addressable space is not that large. It gets larger by the day. I do think there's a world where private credit could originate on-chain, but for that to be true, the borrower, the originator, and the securitizer would all have to transact in stablecoins.

Ideally, they would do it using a protocol or smart-contract layer that was as hardened as possible against all of the elements that introduce things like operational-security risk. To answer your question, yes, insurance would be great. It still remains to be seen how we price all of these different variables, and I think that's probably the biggest headwind against the sector really taking off in DeFi because nobody knows how to actually price these risks.

You just let the market price it. I mean, that's the whole point of having a liquid market behind a CDS. You have all these actuaries in the real world trying to price catastrophe bonds and all this stuff, but the market just has a real-time price to it, and ultimately the market prevails. That's why, when we were back in the day farming with size at an institutional level with LPs, we felt pretty comfortable farming because we could go buy cover on Nexus Mutual.

That was grossly underpriced, to the point where it was a phenomenal risk-adjusted trade because you could go buy cover on Nexus and no one was buying it. There are all these efficiencies that are still in DeFi today.

I've been a long advocate—and maybe I'm overthinking it—of risk-management instruments that could ultimately give you more comfort, because a lot of these things are theoretical. The problem with crypto is that when something breaks, there's very little recourse.

If you have cryptographic guarantees—that you buy insurance and your payout is guaranteed because a smart contract governs it—to me, that sounds like insurance is the natural next evolution of something that could really instill confidence in a world where AI is finding a whole lot of bugs in systems that you thought were super battle-tested.

For me, I have no interest in playing in DeFi right now because the Lindy is sort of reset across the board in a system that is algorithmic and automated. I think you want to have the natural brakes that you have, and circuit breakers and time locks that exist in the real world. You want to interact with a human.

For sure.

For now, AI could also make these protocols super safe—much, much safer. It's just a gap that we're in today, and I think it's sounding the alarm around every protocol.

I sort of posted yesterday about this initiative: How can we create and price risk in response to Luca's paper? I said, okay, could you create a simple metric of how much a protocol is spending per year or per quarter on security versus the TVL that it's insuring?

It's by no means a perfect measure. It's very crude, but at least more transparency could allow the market to make that distinction so that it doesn't use a broad stroke right across the board.

One of the things that we're going to try to do as curators is select issuers that have been through the Steakhouse wringer. They often emerge from it with a more hardened protocol—not always, but typically. We tend to be quite demanding of the issuers that we underwrite, and our approach to improving this space is being harder on them and introducing operational-security audits as a requirement to be listed in the vaults, at a minimum.

I think that raising the bar at that level helps. The other thing is that having a term structure in DeFi would help the market be a bit more efficient, because insurance is a long-term instrument. But DeFi still operates on 12-second blocks, so it's very hard to match the two. I feel like that's maybe another reason why insurance has struggled to take off, despite the potential inefficiencies you described.

Yeah. What you're saying really resonates with me. I don't think it's too interesting to discuss what the exact parameter we should use is, because the way it works in DeFi—in TradFi—is that every desk has its own models. You do your own pricing, you have your own expectations, and then you express your views through the market.

There is a market price, and then you think, okay, this is attractive for me or not attractive for me. My views, by definition, are different from Adrian's and different from yours, because I have a different set of views of the world.

What is important is that we have good primitives to express those different views. I think Morpho did this very well for short-term overcollateralized borrowing, differently from Aave, because it's easier to express views and isolate them. The problem is that we don't have other tools to express our views.

A few reasons—I think the lack of talent in DeFi is one of those reasons. I operate in the stablecoin space, and we're very, very on-chain native. But now most of the fun in stablecoins is connecting with bank virtual accounts and Visa, et cetera. Great, but intellectually, it's not what we're talking about. This is different. This is plumbing with payments, right? It's not.

Are there not enough mathematicians or not enough scammers, or both?

Yeah, both, I think. Probably both are in AI these days. So I think we need more talent.

The insurance point is that insurance is a long-dated balance sheet. It's the best balance sheet to express views that take a long time to materialize. That's why insurance is the best vehicle for the Berkshire Hathaways of this world.

We need to have insurance balance sheets. There are some people who are trying it, saying, okay, I price my balance sheet on Bitcoin, for example, and I'm going to take positional views on Bitcoin over the very, very long term. I don't care what happens in the meantime.

We need those primitives to express views that are very different from the block-by-block movement of a price, or a liquidation, or a call that's going to break or not. Hopefully, we start putting them in place and start having them. I think Morpho V2 is an interesting development, but we need much more to do interesting stuff.

6. The Physics of Onchain Lending

So, you guys are very talented. You're still here. Why? Why not? I've had this idea, and you guys tell me how dumb it is. My issue with insurance is the way Nexus and Anchor Protocol have tried to do it. I think the more interesting one was Andre like allowing unwrapping of Nexus cover, because that gave you more pricing. With the KYC on Nexus, it's kind of antithetical to the fluidity of DeFi, but it was interesting because the pricing on the cover all of a sudden reflected something more real-time. You could see the jump in price and say, “Okay, actually, the market's telling me that X protocol is way riskier,” and that gave you a really good oracle of risk in and of itself.

My issue with crypto-native insurance protocols is correlation. In insurance, you want to be uncorrelated. You want to have property and casualty, and you want to be diversified. You want to build resilience and antifragility into your balance sheet. The problem is that crypto is very correlated, and composability, if you're interacting with a lot of protocols, just tends to correlate really tightly. Maybe that's no longer as much the case, but my view is: go buy traditional insurance that has exposure very far removed from crypto, and then slowly start issuing policies and serving the DeFi ecosystem. The problem is, again, we go back to square 1: yields are low, and when yields are low, you can't really offer that.

Composability spreads contagion, but is isolation and simpler primitives how it's prevented?

That's like—internally, we call these biweekly 10-sigma events. We seem to have a 10-sigma catastrophe in DeFi every 2 weeks lately, at least since October, if not longer. In each of these events, we saw high-yield repo lending rates spike through the roof as a kind of lagging indicator of risk in these pools, but the Bitcoin rates stayed completely unperturbed in Morpho because they're isolated.

And so I do think that you can have some degree of uncoupling of risk contagion if you build on sufficiently isolated primitives. What that looks like for crypto-native insurance is not super clear to me. But, yeah, you would need to have a balance sheet like Luca describes—far longer-term, completely untouched, unrelated investments in cut bonds whatever something else—for it to really work.

Yeah. I think I would separate crypto-native insurance, meaning that you're building the mechanisms on-chain, from actually giving long-dated balance sheets the possibility to take crypto assets on board. I think, for example, with Bitcoin, there's so much to do that we haven't done. Imagine even if you start allowing Bitcoin to be reserves of life insurers. Suddenly, you have a very, very long-term-oriented buyer of Bitcoin, and the volatility of Bitcoin's price goes down massively.

I think that even if those assets permeate—because now all the holders of, I mean, Bitcoin is probably an exception, but the holders of all those assets are all traders—the correlation of the prices of those assets is because everybody is moving with liquidity. If you're bringing more buyers into this industry that have a different time horizon, then suddenly the behavior of these prices changes.

I think even just starting by having long-term buyers—because, you know, I'm buying Bitcoin, insuring Bitcoin for a life insurance policy that I know I don't need to pay out on average in the next 20 or 25 years—my perspective on prices and price volatility is very, very different. I mean, that's how Warren Buffett buys companies, at least for a while. So I think this stuff needs to permeate, to stop being only a pure trading arb asset and just be in the hands of other people. Suddenly, we start seeing behavior differently.

What we can put directly on-chain remains to be seen, but to your point, Jason, if you have somebody who is giving you insurance, it is not on-chain, but it's an OTC desk that is really able to take a position against you, you will just do it. You will just negotiate with Goldman. You don't buy it on-chain. It's fine.

Luca opened this discussion with how regulation was actually forcing stablecoins to be narrow banks, and therefore pushing all of this risk on-chain. It would have been interesting to have a world with more Tethers, because Tether is essentially the only issuer that can actually take long-term balance-sheet risk. In the real world, this is typically one source of long-dated balance sheets that could serve that purpose or originate these products.

So we're forced to contend with doing it ourselves. But I actually love that, because I'm a passionate believer—I may be the last believer—in free banking.

But isn't the issue with Tether, though, that, yes, they're less regulated than a lot of the others, like Circle, but they're buying and doing some attestation, though not to the level you'd want? They're not fully collateralized. They have some other instruments in there, and they're not passing through the yield to the holder. So, as we think about where the risk sits squarely, that's with the holder of USDT.

Yeah. I think Adrian's point is—I've done some research on Tether's balance sheet as well, for fun, and published on it—but I don't think Tether is going to do it because, as you were saying, Luca, they have the best trade in the world. They take long-dated risk and don't pay anything out to anybody. Amazing.

But what I guess Edan was saying is that Tether is actually one of the few institutions that is in a position to issue long-term paper because they have duration on their asset side. So they could actually issue loans at 5%, 6%, or 7%.

They are trading funds, not crypto. It's tiny, but they're doing it with commodity traders. Tether is in a different universe than mortals. This might be the best business in history. It's a shame that regulation has obligated Tether not to have any competition, which is crazy.

It would have been interesting to see a world where more stablecoin issuers, in a free-banking-type environment, could have competed on similar terms and offered these products. It would have brought on different types of risk, for sure. But instead, we have to contend with building this ourselves. Again, as a free-banking enjoyer, I'm all for it. I love the idea that the regulator is forcing us to have free banking. It's great, because now we have to come up with all of these solutions ourselves.

I mean, the only institution that is trying to do it is MicroStrategy, right? They're trying to create a term structure on Bitcoin.

11.5%.

There's also, Meanwhile, this very interesting company. I'm sure you guys are aware of it. It's building a life insurance product on Bitcoin. Its balance sheet has enormous duration and nowhere to put it.

7. Fallout From Drift’s $280M Exploit

Yeah. I want to go back to something that I found interesting and don't want to gloss over. You said that, after 10/10, there's been a 10-sigma event every other week. You said—

I was calling it jokingly a biweekly 10-sigma event.

Why is that? Because they seem to be the type—like Drift Protocol blowing up for $250 million is quite a big event—and these events have a binary characteristic. Whenever the actual exploit is relatively minor, the finality of blockchain settlement means that the loss is almost total. When they happen, they're traumatic for everybody, and they seem to happen all the time.

Both of you guys interact with real-world financial institutions and others, and I want to make this not about the exchanges and folks that stand to profit no matter what, right? You have a principal-agent problem that exists in traditional finance. But does an event like Drift update—force you to underwrite and really question Lindy strategies—to your point, Luca, what DeFi should be really good at and what it should maybe not do? Or, if you're going to do it, maybe do it in a different way? Because, to your point, Adrian, it is very binary and catastrophic.

So, maybe your reaction to the Drift hack with AI, like Anthropic's release, and how you're seeing the market react to that, especially people who have been curious about DeFi that you're interfacing with? Do you think there's a material setback because of that?

I think—I don't know what Luca thinks. I'd be keen to hear what Luca thinks—but my experience has been that this type of event dramatically increases the cost of capital for operating on-chain. They're very public and transparent. They're very unique to crypto in that regard.

Although I would like to lower the tone of self-flagellation a little. We didn't have the luxury of monitoring positions in insolvent hedge funds before they went under, so we can also take some cold comfort in that. It's not like these events never happen in traditional finance. you get updates on Khalis you know four times a year.

I think the reaction from crypto Twitter has been encouraging. Frankly, there's been a lot more discussion about the risks and a lot more appreciation of operational security as a vector. It seems like there's at least a genuine intent to take it more seriously, in particular with AI attackers—AI-powered attackers.

You know, part of it may well be that the increase in frequency of biweekly 10-sigma events is just people with AI spotting exploits where nobody caught them before. So it stresses the importance of having stronger operational security, for sure, and maybe keeping things more minimal. Instead of building protocols that do more, harden the layer that has to rely on cryptographic guarantees to make it stronger.

I don't have the statistics on hand, but my impression from reading different reports was that the smart contract audit side is gradually hardening and we're getting better at it over time. That's exposing the remaining surface, which is operational security in these protocols, and that will hopefully harden as a result and as a reaction.

Luca, before I go to you, one thought is back to CDS, which I find really interesting. In Nexus, it would almost sell out, and it's almost like—I think Cozy, Sherlock, Opium—there have been a lot of iterations of this, but none that, in my opinion, have really taken off. Your point around credit default swaps is a really good one, but I think the market was telling you that there was something there if you looked at just CDS, because I remember there were whispers in financial markets that the problem with crypto is that it just happens, it's catastrophic, and everything is automatic, so it doesn't allow you to respond.

Is there an opportunity for prediction markets to play a role here, pricing in market sentiment? If you spin up a world where you have all these prediction markets on, “Is there going to be a hack on DeFi?” and you see a spike, that's revealing in itself. Before I cede the ground to Luca, the nearest parallel I can think of to something like a crypto hack in traditional finance was the attempted heist on Bangladesh Bank by North Korea.

8. Prediction Markets

They were probably the most talented cyber operators in the world, frankly, and they were able to social-engineer access to the accounts of Bangladesh Bank and time the time zones to ensure that nobody was watching. I think they were successful in exfiltrating a couple hundred million dollars, but they were thwarted at the last second. That's much rarer because these transactions can be reverted and they're not final, to your point; the finality of blockchain settlement makes what's at stake much higher in crypto.

Yeah. And I think there is an incentive, exactly because of the way crypto works, to keep stuff super, super, super confidential until the second when everything happens. So I think the leakage of information is way less, but I think your point, Jason, on prediction markets is actually great.

Again, I don't shill my business, but sometimes I shill my writing. I've been writing about prediction markets for a long time. I think the prediction market is probably one of the best ways to express uncorrelated risk out there. It's a very good primitive.

Unfortunately, prediction markets are very fancy today, but they're not doing that; they're just doing sports betting, right? So it's a different use. The valuations are pricing retail access to gamble, but I think prediction markets are a very interesting primitive to actually express uncorrelated risks and cover.

I also think that what we are seeing now in crypto—I agree, we see it in TradFi—the difference is TradFi has a significant capital buffer. Because there are unknown unknowns, you need to know that there are unknown unknowns, accept it, and somehow price it and buffer it with a cushion. We don't have it in crypto; we don't have that luxury. If you do have insurance, you make a lot of money.

You have some insurance funds, but you could argue they're the hardest hit.

But it's minuscule, right? So I think if you had more money, that's why whales in crypto make a lot of money: They have their own insurance funds personally, at the fund level, and they can take positions. There is no algorithm that is super optimized; you have your own balance sheet. That's why smart venture investors have been good at this game, because they know how to do portfolio management.

There is a lot of stuff to do. Maybe we're not going to be better in so many other things where there are other primitives that are better than crypto. I think this idea of crypto-maxing the world and everything going on decentralized rails probably is not true, right? We are better at some things.

The reason why we're talking about it is because crypto is a great distribution channel. Everything is permissionless and global from day 1. You have a great product, and suddenly you put it in the market and everybody can buy it. But that doesn't necessarily mean that we need to have every product in the world on crypto rails, including private credit.

I used to be a private credit investor. I love the asset class. Now, do we need crypto rails for private credit investors? Not really. If you're a hedge fund, you just buy it yourself and you know how to price it. But there are things where we are much, much, much better. I think prediction markets are one of those surfaces.

I just hope that we go back to building very smart things on prediction markets, because right now we're not. We're just transforming prediction markets into partially unregulated—

Gambling stuff. So my thinking around that, just to extend it a bit further, is that you have to sell insurance at the point of sale. If you're going to deposit in a vault, you should then simultaneously buy a contract in a prediction market that hedges your position in case there's a catastrophic event, because otherwise people forget about it.

It's like flood insurance. There's a reason why, when you buy a car, you can't step out of the dealer unless you have insurance. When you buy a home, or you're a tenant, you have to furnish proof of insurance. I think there could be an interesting mechanism here where, by default—not all strategies, at least some—provide some protection for the user automatically embedded.

I think it hasn't been adopted, or it won't be, because it's very cyclical: Rates are low.

Rates are low. Exactly.

It is. Again, it goes—it's difficult to do this. I don't want to see the world where you have a prediction market that incentivizes people to hack protocols beyond just—

But that's the protocol—

9. What's Exciting In Crypto Today?

True, I guess you could say Immunefi has been pretty good. I think there's a pretty good stat that protocols that have bug bounties, and healthy ones, inherently get hacked less, right? But yeah, it's a bit of a quasi-assassination market.

This has been a fascinating discussion. I get the feeling we could go on and on and on and on. Maybe we should do another series here, because both of you individually could come on the pod and talk so much about your respective fields.

We want to make it practical for users, but also introduce some more excitement. This is not meant to be a “DeFi is dead” discussion. If you picked up on the nuance, it's not. What are you guys doing in your respective companies, or how do you see the market? I think you've hinted at this a bit in the pod, but what are you really excited about that you think is probably not talked about as much in the market, but will probably grow and be very successful?

I will start. I've been in the stablecoin space for a very, very long time, before it was fancy. We started M^0 in January 2023 with a very strong stablecoin infrastructure thesis, and most people outside of the core investors we had had no clue what stablecoins were or why we were so hyped about them.

But there are 2 elements that are still very important that I'm fighting for. One is super popular these days, and the other one not so much. The one that is super popular is that we need stablecoins to pass risk-free yield on-chain.

If we allow stablecoins to stream risk-free yield on-chain, we are repricing the whole curve on DeFi. I think then you can really start doing smarter stuff because the market is more efficient, and banks are fighting it. Everybody's fighting it; everybody has a horse in the race. But I think it's absolutely in the interest of the whole space if we do that, and I'm going to keep fighting for it to do it in the safest and most direct way.

At M^0, we have systems that allow the streaming of the yield of the risk-free rate on-chain on a continuous basis. We can do it offshore; we cannot do it onshore, sadly, which is crazy, because it creates these very weird situations where retail is taking a lot of risk and they're forced to do that.

The second one that we don't talk about too much is stablecoins, in my field. Stablecoins are all the hype today, but we spend a lot of time just bundling stablecoins with fiat, and most of the so-called stablecoin infrastructure companies—I’m not going to name names, but very successful and profitable companies—they're just patching tokens with accounts and cards and FX and making spreads. This is okay, but I think in our thesis at M^0, stablecoins are on-chain infrastructure. So you need to build infrastructure primitives on-chain to do liquidity, to do FX, to do yield programmability, create accounts, move stuff, and create applications around it.

I think we should spend much more time on that. We feel we are alone in this field, which probably means there's enough capital to build it and be a couple of years ahead of everyone. But I think now most of the effort in stablecoins is just doing on- and off-ramping. It's like we're building scanners and printers. We're not writing Photoshop. We should spend more time writing Photoshop.

On my end, we're really focusing on the nerdy parts around plumbing. We view crypto and DeFi as plumbing, and what we're excited about is either replacing or building adjacent plumbing that will allow financial markets to flourish on DeFi.

Coming back to the point about the long-term balance sheets, as a team that's been entrusted with a part of the balance sheet of Sky through the Grove Star, this is not as big or as long-term-focused as Tether or as an insurer, but it's not that far off. I think that allows us to do very interesting experiments that can help increase the accessibility of things like the risk-free rate on-chain and take on more duration to try and lengthen the term structure of DeFi beyond the point where it's just a 12-second increment.

Nothing that I can discuss publicly now, but we have a few projects in the works that will be released in the next few weeks. To your point, for people thinking that this is DeFi doom and gloom, I hope I didn't give that impression. As DeFi burns to the ground, we will be the last people standing in the ashes defending it.

We're super committed. We're a family business, not venture-backed. We grew up in this space and will die with it, and we're very excited about DeFi going forward, for sure.

You talked a lot about the frameworks and the analysis you guys do. Is that open-source? Do you plan on publishing?

We publish all the time on our Substack, The Steakhouse Kitchen: kitchen.stakehouse.financial. We tweet about it all the time. Every time we onboard an asset issuer, we underwrite them and talk about it. We have our risk management framework updated on a daily basis on our docs page, so there's plenty of material.

Like you said, we could talk about this for hours. God forbid.

That's amazing, guys. We really appreciate you coming on, and we'll link to your blog post, Luca and Adrian. We'll definitely post to that dashboard. We'd love to have you guys on again. There's just so much that we didn't talk about, like rating agencies. Are there going to be rating instruments? Insurance? We can go on and on and on about a lot of the different things that we talked about in this pod.

But we'll have to leave it there for now. Really enjoyed the conversation, and we appreciate you guys coming on.

You should see the prediction market on what we will talk about next time.

Definitely. Awesome, guys. Thanks so much for listening. Have a great weekend, and we'll see you next time.