Hello everyone. I'm Vlad, the founder and CEO of Lighter. It's great to be back in New to be back in New York. I started my career at TradFi many years ago, so it's great to be back in town at ETHDenver. It's great to see, you know, I think there's a lot of talk online about the merger of TradFi and DeFi.
And I think what we create at Lighter is part of that story. But it's great to see in person, at this event, a lot of conversations between people from TradFi and DeFi, as well as from the policy side. So, the future has finally arrived. But today I'm going to talk about how Lighter works, dig a little under the hood, and see how we operate at scale on Ethereum. So basically Ethereum.
So basically , you know, I think some of you are familiar with Lighter from a product perspective. It's used mainly for criminal trading, but also spot assets, hmm, and I've been working on it for a year and a half . And I'll go into some metrics later, but what I really want to focus on today is the infrastructure side infrastructure side , you know, how lighter actually works? Because there are some really interesting technical problems that our team has solved, and also we have built on solutions from across the Ethereum ecosystem, and so I think building on Ethereum was a decision that we made at the very beginning, you know, 3 and a half years ago, and I think it really paid off. So we'll talk about that too.
So, um, yeah. So, the key concept that follows is that every implementation can be verified by proof. So, you know, finance is a very broad concept, but in finance there are only a certain number of different primitives. I mean, the basic primitive that we start with we start with is the order book. You know, orders can come in, limit orders are matched against market orders, you know, orders are canceled.
Hmm, there may be liquidations. There's financing, you know, there are other forms of financial primitives, but the number of financial primitives is relatively small compared to everything that a general- purpose computer can do. So, you know, Ethereum is obviously a base layer that can verify anything, but it's a little slower. We want to run processes at full high frequency, you know, so they run at the throughput level of something like Nasdaq, right? And yet, make sure everything is verified.
And this is the complex problem that we solved in Lyra. Uh, we're essentially building a dedicated L2 for finance on top of Ethereum, meaning for something like finance, you obviously need high throughput, but you also really need security and verifiability. I mean, we've seen this from the fact that most of the TVL in DeFi is on Ethereum and Ethereum L2. So, you know, we essentially have trading activity happening on Lyra. You know, every order, every cancellation is signed.
And then we have a sequencer that, uh, works. At the moment, it is a centralized sequencer. We, you know, on our roadmap we are also decentralizing the sequencer. But the really nice thing about really nice thing about , uh, you know, proving this activity with zero-disclosure proofs is that it actually matters less in terms of trust assumptions whether the sequencer is centralized or decentralized, because everything the sequencer can do has to conform to the rules that exist in ZK schemes. So, now it's implemented so that there's a cluster of evidence.
Actually, uh evidence. Actually, uh , you know, over 10,000-15,000 machines that generate evidence. You know, right now, as we speak, evidence is being generated, uh evidence is being generated, uh , by these machines. So, essentially, trading activity occurs in real time. Thousands of orders are matched every second , and then a proof cluster runs behind that, but not too far behind, and generates proofs for all of that, and then a single proof, you know, all the proofs are combined into a single proof of the transaction batch, which is published on Ethereum L1.
And the way that these proofs are supposed to be done is that they are public rules, right? So these rules include things like: when a warrant comes in, you know, first-in , first- out, right? So it's impossible for one level to have something where, you know, one market maker has different rules than another, right? Or that, you know, one market participant can review the order book before someone else, or even when it comes to liquidations, those rules are public as well , right? So you won't have a situation where "Hey, I don't know why I was eliminated," which sounds theoretical, right?
But, when markets move fast, and, you know, if they do, sometimes there can be fluctuations of a few percent within a few seconds. And at this point, it matters a lot for traders, does n't it? Whether you are an institutional or retail trader, how exactly the rules are followed is up to you. So in Lyra, all of these rules are public and compiled into a concise narrative. So, at any narrative.
So, at any given time, for example, there is a batch of 500,000 transactions. If even one of them doesn't match correctly match correctly , the code on top of Ethereum rejects the entire batch. And this ensures that everything that happens in Lyra happens fairly for all participants. So, a little bit of statistics on this, right? So, last week was actually the most active week most active week we've seen.
Every day we had a 24- hour period of 811 million transactions million transactions that were approved by lighter, you know, we're getting close to the billion mark. Um, you know, TPS, well, you know, in about five cases around 10,000 on average, you know, we ran into some individual parties where the TPS was closer to 20,000, right? So there are thousands of proofs running now, some using Mac mini infrastructure, some using Amazon. And again , the interesting thing about the way proof works on Ethereum is that we can't experiment with different proof machines. It doesn't really matter who generates the proof, does it?
Because the proof must comply with these public rules. So, you know, if one of these machines works, the other machine will generate the proof. And in the future, when proof-of-stake is decentralized, you know, a lot of, you know, so some nodes might use one architecture choice, some might use another. And so, ultimately, the proofs are aggregated and published on Ethereum from there. So, as I said, as many as 800,000 different actions can be compressed into one proof.
And another key thing about Lighter is that it's not storable, right? So other high-performance systems, especially in TradFi, can certainly operate at those TPS rates those TPS rates , but, you know, you're holding your assets and you're trusting a centralized party. In the case of Lyra, the assets remain in someone's wallet. And actually, there's also something called an escape hatch, which means that even in a world where Lyra stops working for whatever reason, you have the ability to get your assets back through Ethereum, which is a really important security guarantee. Obviously, this is not the case we want to see, but having such a guarantee is really important, especially for institutions and regulators, which is worth thinking about.
Er, because it ensures that even in the worst case scenario, there will still be plenty of protection. And this is the kind of protection that doesn't depend on whether Lyra does anything centrally. And so, in addition to that, you know, the rules that exist for trading logic can also be extended to asset administration. As you can imagine, where in the future some assets may be traded under different sets of rules. Um, you know, some assets might be more permissive in terms of, say, on-chain KYC or something like that.
Um, while most of the assets that exist now can continue to exist as they exist today. So these rules can be customized in some way as we add more derivatives markets and also more spot markets that may have...For example, as you move to tokenized stocks, the rules for things like dividends and splits will also be implemented through these ZK schemes. So through these ZK schemes. So , again, the really important takeaway here is that by building on Ethereum, we can achieve both high performance and this fairness that is proven by block-by- block mathematics.
So, every block, every order, every cancellation, every liquidation must comply with these rules. So, unlike a centralized exchange, which is how, you know, TradFi and many digital assets operate, there is a certain element of trust in the centralized player. There is a permanent record of Ethereum on Lyra. In fact, L2 Beat recently independently verified Lyra's full story. In fact, all the rules were followed during Lyra's year and a half of operation.
And the fact that third parties, obviously, not every trader has the technical ability to directly go in and check the hash of every transaction. But the fact that third parties can and do verify the entire history assures participants, especially at the institutional level, that the system works, and that breaking the rule is inherently impossible . So, a few more indicators. So, the number of daily active users of Lighter is now around 14-15,000. The average TPS is over 6,000.
As I said, on a day of very high volatility, there were spikes of over 10,000, which seemed to go into 20,000 seemed to go into 20,000 , and the latency, you know, was maintained below 200 milliseconds, although there were several other studies that showed that Lighter has the lowest latency of all decentralized exchanges. And again, this is what Vitalik was talking about earlier, which is that the most effective L2 level requires the safest L1 level underneath it. And I think we saw this in practice when we created the lighter. From an economic perspective, the number of orders translates into dollar volume, which as you know, I have done 140 billion transactions on lighter in the last 3 months. Uh, 7.
5 million of revenue came from that, and the scaling worked really well, where the cost of proof, I mean, you know, if you weren't using these special ZK schemes, the cost of proof would be much greater than the revenue generated, whereas instead, using these specialized schemes specialized schemes , specifically for finance, the cost of proof is actually, you know, very high margin here, where the cost of proof is only 20%, and actually, that's an average over the quarter. At the moment it is even lower. This is closer to 10% lower. This is closer to 10% of the income received. And the revenue generated, I should point out, actually, you know, our cost structure is lower than any lower than any other competitive product.
So, this is already a very efficient cost structure for participants, but the costs of proof are much lower than even that. Uh, you know, evidence, it's kind of magic, right? The magic lies in using mathematics to prove that everything that happens in financial transactions is done honestly, correctly, securely, and all this is possible thanks to the use of zero- disclosure technology and development on the basis of Ethereum development on the basis of Ethereum . So we're, um, very, very happy with the technical decisions technical decisions that we've made, very happy to be part of the Ethereum community that not only creates this efficiency, but also allows Lyra to be composable with the rest of what we're doing by creating, um, the Lyra VM, as by creating, um, the Lyra VM, as , um, mentioned in Eason's previous talk. We are working on this with the Axiom team on this with the Axiom team .
So, developers, be sure to reach out to us about Lyra VM. Traders, pay attention to the lighter, and we look forward to your response, and, um, thanks to everyone who was here and listened.
Automatic transcript — names and jargon may be misspelled.