FTX Chooses Solana for Serum: A High-Speed, Non-Custodial Decentralized Derivatives Exchange
FTX and Alameda Research created the Serum Foundation and announced Serum, a new high-speed, non-custodial DEX that’s built on Solana. You can learn more about Serum in the white paper and in this comprehensive write-up on The Block.
Serum is a complete, non-custodial spot and derivatives exchange running on an on-chain central limit order book (CLOB) on Solana’s mainnet. Yep, you read that right.
In order to maximize access to Serum, Serum will support cross-chain asset swaps, decentralized stablecoins, decentralized oracles, and non-custodial wrapped BTC, BCH, BSV, LTC, ZEC, ETH, and ERC-20s. Serum will decentralize the entire DeFi stack. Nothing will be left centralized.
By building on Solana, the Serum Foundation is enabling the best of both centralized and decentralized worlds: an exchange that is censorship-resistant and non-custodial and that is fast, inexpensive, and highly liquid. This is only possible because Solana enables Serum to run a CLOB that updates every 400 milliseconds.
Sam Bankman-Fried, the founder of Alameda Research and FTX, recently published five great Twitter threads on The State of DeFi (here, here, here, here, and here.). Building on his threads, Serum represents the first product for the next phase of DeFi.
DeFi 1.0 mainly focused on innovation in money markets (lending and borrowing). Defi 2.0 separates lending and trading and solves for decentralized exchange (high-speed trading and derivatives).
Despite clear product-market fit, DeFi still has many shortcomings. Most notably, it’s slow and expensive, which has limited the application design space. Gas costs on Ethereum are pushing trading costs above .05 ETH = $10 per trade.
This is precisely why centralized exchanges (CEXs) dominate trading today. But they won’t forever.
The Alameda + FTX Edge
Automated market makers (AMMs) have grown in popularity on Ethereum over the last 18 months primarily because they make it easy for yield and risk-insensitive asset owners to provide liquidity to the market. However, that doesn’t mean that AMMs are the optimal mechanism to provide liquidity. AMMs are obviously lacking in many dimensions (most notably, capital efficiency).
Why should liquidity be provided differently (AMM vs CLOB) just because assets are held non-custodially vs custodially? Why should the nature of custody change how liquidity is provided?
It shouldn’t. Liquidity and custody are unrelated concepts.
Serum has a profound edge that no one else has: Alameda and FTX.
Alameda is one of the largest liquidity providers in crypto across almost every major venue trading billions of dollars per day, and they are intimately involved with the Serum Foundation. Just as Alameda seeded FTX with liquidity to propel its exponential rise over the last 12 months, Alameda will imbue Serum with incredible liquidity from the outset.
FTX is a major destination for many organic takers in crypto. They prefer to trade on FTX because FTX offers the best trading products (UX, APIs, matching engine, liquidation engine, cross-collateralization, liquidity, etc). FTX will route taker flow to Serum.
No one in the existing DeFi ecosystem has the combined financial, operational, and technical resources of Alameda and FTX. Moreover, Alameda and FTX have deep knowledge moats about how to bootstrap liquidity in modern crypto markets that virtually no one else has. They are the most impressive growth story in crypto over the last 12 months. They came out of nowhere and established themselves as one of the dominant exchanges.
Latency And The Future Of Financial Markets
There is some time scale on which financial markets start meaningfully breaking. While there is no definitive answer to this question, I believe the right way to think about this is to think about the time scale at which new information propagates the world such that humans and computers can accurately process it. This timescale is probably 1–3 seconds.
The simple litmus test is: how long does it take for a qualified investor to read a headline about an asset, and estimate a general price movement with a conservative interval. Computers can do this in under 1 second, but they often miss the nuance. It generally takes humans a few seconds to accomplish this task.
Solana produces blocks every 400–600ms, which is more than 10x faster than Ethereum 1.9 and Ethereum 2.0, and 3–5x faster than the newer high-performance Layer 1s. Because of the constraint that is the speed of light, it’s impossible to reduce blocktimes meaningfully beyond 200ms in a globally distributed network of nodes. The Solana community is working to reduce blocktimes to 200–250ms in the coming years.
Solana operates on a timescale that’s sufficiently fast to support the rate of propagation of information around the world. This enables Solana to support a completely decentralized, permissionless, censorship-resistant financial system in which everyone can participate.
Layer 2s And Social Consensus
The Ethereum community has been discussing Layer 2 scaling solutions for years. Layer 2’s will be adopted in some niches, but even a preponderance of Layer 2’s does not solve the core problem of concurrent social global consensus.
The reason that Ethereum mainnet works are because there is a global social consensus that Ethereum mainnet is not the Ropsten testnet. This is so obvious that it sounds stupid, but it has deep implications for how to think about the growth of Layer 2s on Ethereum.
Each Layer 2 environment by definition has some separate social consensus as to the state of the system that is not captured by Layer 1 consensus. That is why Layer 2 scaling is insufficient.
There are going to be myriad Layer 2 environments:
- Starkware and Matter Labs envision a world with many application-specific zk roll-ups (each with custom zero-knowledge circuits).
- Skale envisions a world with thousands or millions of dynamically resharding Skale chains that exist as part of a logically centralized network.
- Optimism and Arbitrum envision a world of many optimistic roll up (ORU) chains.
The proliferation of Layer 2 scaling solutions is great for innovation. Each of these designs has unique trade-offs that will likely serve niche applications. However, by definition, the proliferation of Layer 2 solutions breaks global social consensus.
Consider the following:
If there is an Optimism node that powers the Synthetix in Layer 2, no one else in the world other than an Optimism node will recognize the state of all of those Synthetix trades until the users exit that Layer 2 instance. If that user wants to convert her Synths back into spot assets — for example by trading sUSD against a pool on Curve, Uniswap, 0x, Kyber, or any other venue — the user will have to exit the ORU.
As more and more liquidity venues move to Layer 2, liquidity aggregation breaks, making the UX meaningfully worse for most users.
This problem becomes exponentially more complex as the number of Layer 2 instances grows, and the number and types of applications grow. Everything will have to settle back through Layer 1 anyways, meaning that the fundamental scaling benefits only apply on a local basis, and not a global basis.
This is not to say that Layer 2’s won’t be used, or that they won’t power some very compelling applications. They certainly will! However, those applications will by definition be ones that only require some localized consensus, as opposed to global consensus.
The Road Ahead
The Serum team is moving at breakneck pace to launch, and we all expect to have more to announce in the coming weeks.
As always, make sure to follow us on our social channels for daily updates on the Solana ecosystem, including Serum.
Follow us on social media: