Rune Christensen and Guy Young on DeFi's maturity, AI agents, and the battle for user capture

AI: So Rune, let's start with how MakerDAO is trying to navigate the issue of scaling. You need to scale because you have new opportunities, but you don't need to scale because you serve the people that most platforms already serve better than most. So how are you thinking about the balance there?

Rune Christensen: First of all, I think what's really important to note about Maker is that we actually don't think in those terms first. So the very first question we ask ourselves is what sort of DeFi platform do we want to build? And that's guided by the DAO, our mission to create money, and we guide our principles in terms of how we configure the system.

And once we decide how we want to make the system, maybe it doesn't fit so well with the mainstream users. For example, we have 100% backing of Dai, we have a lot of risk management, we're diversifying our collateral. And that means the yields on the platform are lower. The leverage levels you can get are lower, but it makes the system safer. And that means we've kind of been pushed out of the mainstream DeFi user group because they just want more leverage, they want higher risk, they want more exotic stuff.

Over time, we've been evolving the protocol by just letting the governance process run. I think it's quite remarkable that MakerDAO is still here today and is the only protocol that has remained at the center of fundamentally always being on Ethereum, Layer 1, and remaining the base protocol like we were at the beginning. Many of the others have had to evolve, completely pivot, or they've had the opportunity to move elsewhere — which they've been prevented from doing because it's not what the users want.

AI: Compared to 2019, when Dai was being used to interact with Compound, and MakerDAO had a much more central role, the big shift since those days has been the expansion of stablecoins from competitors. So how do you think about your role in DeFi? Are you fine with accepting a more marginal value in DeFi as a more conservative project? Or if you want to still be the leader, how does that happen?

Rune: So I think what's important to note is that Maker and Dai have already gone through this once. Back in 2016, DeFi was pretty much growing on top of Maker. For example, I spoke to Leshner, and he was telling me how initially the concept of Compound was something like a platform that would enable people to lend Dai. And then the big switch was, basically USDC came, and then it completely changed into being about crypto speculation.

So, I think it's very important to not confuse what happened with Maker being similar to what's happening to Curve. Curve is a system that integrated centralized stablecoins. And actually right now, Curve is dealing with the fact that all of those centralized coins are more or less jumping off Curve. So Curve does have this problem of, well, what do they do now? Because they based their value proposition on intermediation. And then when you build the system like that, as soon as the centralized protocol can have their own liquidity elsewhere, you're now completely relying on that centralized protocol permitting you to be involved.

That's not the situation Dai is in. Dai is its own money. It's a decentralized stablecoin that is built from the bottom up. Maker can potentially be a competitor. It can be viewed as a competitor to the centralized stablecoins. We're a system that's fulfilling a very important role in the ecosystem. It's just a different role.

Guy Young: Maker's the prime bank, right? It's actually competing with central banks. And in terms of price stability, it's still early enough, so yeah, centralized stablecoins are competing with us. But it's a bit like the evolution from the commodity money to our current modern banking system. The idea is that Dai becomes the ultimate reserve currency, not just instead of the dollar or as banks evolve, but in my mind, MakerDAO becomes the central bank.

AI: I want to talk about AI agents as well. A lot of the large language models include cutoff points where they don't know about the world after, say, 2021. So they don't know about DeFi Summer, they don't know about a lot of the more important developments in crypto. You've been working on open source models?

Rune: I've been working on LLMs for a while... I have to clarify that I'm not working on anything that's specific to MakerDAO in this context. It's not a MakerDAO project. I 100% agree with you that it will be with AI that DeFi will reach mass adoption in the future.

LLMs are becoming so powerful, so quickly. It's so obvious that eventually you'll have an LLM that has more knowledge about finance than humans generally have. And in that situation, if an LLM can specialize in finance—in DeFi—it will be able to help users, normal untechnical users, manage their finances better than they could themselves.

Every interface today is either too simplistic to include all the functionality you need, or it's like power-user interfaces that are too hard to understand. Whereas with LLMs... because LLMs are gaining knowledge so quickly, it's just a matter of time before you can go to an LLM that can go on-chain, that can go off-chain, that can look up the right information, and make the right decisions.

Guy: One of the theoretical computer scientists who designed FoundationDB, Dave Rosenthal, is working on AI agents internally within the US government now.

Rune: Our stack itself at MakerDAO is challenging, which is why we need to fund more R&D in this area. When you're talking about stablecoins, something that's purely a currency, it requires much more complicated risk management. And you don't see as much development. It's very challenging territory.

AI: And that's the argument for why Maker has to go multi-chain as well to remain competitive in DeFi, right?

Rune: It's a bigger conversation about path dependency in history. It absolutely is a serious problem that we have this fragmented ecosystem where you end up with each chain having its own walled garden of stablecoins and DEXes. I do think the solution has to be to integrate more and more chains into Ethereum. I do think all chains need to be able to talk to each other. I think our endgame architecture will be that basically the stablecoins are fundamental, and it makes sense they should probably be on Ethereum. This will change once we get much more scalable Layer-2s and technological innovations on top of Ethereum.

Guy: If you have a hub and spoke model, people have to pay the hub toll. So they say, let's build a new network. One of the reasons we're seeing so much innovation and competition in DeFi is that everyone is trying to solve it slightly differently with different technical architecture.

Anyway, that's my rant on multi-chain.

AI: Let's switch topics a bit. So most of you guys know each other from the early days of DeFi. How many of you guys are moving to Singapore? Are all of you moving to Singapore?

Rune: I've been in Singapore for like three years.

AI: That was prescient! So is the US just a lost cause for DeFi?

Guy: It's jurisdiction shopping. And a lot of people are going to London because London has the right regulatory framework. I think a couple of my lawyer friends have been doing licensing work in Hong Kong because Hong Kong is trying to become a hub again. Especially if the Hong Kong dollar keeps its peg to the US dollar, it's going to be a nice trading venue.

Rune: I would say Singapore is in a great position. I think it's a big opportunity for them. The financial system is extremely stable here. And the regulatory environment is conservative, but in a way that it's at least predictable.

Guy: There's also a lot more early-stage funding, so VCs all end up going there too.