An Ultimate Vision for Serum
It’s easy to get lost in the weeds. Everything has strengths and weaknesses; what matters isn’t rattling them off or proving you know of at least 7. What matters is evaluating them, and weighing them against each other.
There’s a question of whether DeFi will ever become huge, and I can’t answer that question for you. Maybe none of DeFi will ever become a staple in society.
But maybe it will.
In the end, this post is addressing the following question: could Serum become a trillion dollar ecosystem with a billion users? It’s divided into two parts. First: why might DeFi become a trillion dollar industry. Second: if DeFi does get that large, why I think it’s most likely to be Serum.
For some this question might seem stupid: but for others DeFi seems stupid. This section is addressed to that second group.
I used to be part of it. A year ago, I was very skeptical of DeFi; if you’d asked me what I honestly thought, I would have said “slow, expensive, buggy, and underpowered”.
A few things changed my mind. Before getting to how DeFi might address its shortcomings let’s talk about why you would even bother with it in the first place.
For some, decentralization and censorship resistance might be enough. I don’t think it is, at least not if we want to get huge. Visa and the NYSE and Apple and Nissan might pay lip service to those values but in the end a ton of potential adoption hinges on showing the world that building on-chain can make their products better, not worse.
And I think the key to that is composability.
One group built Uniswap, a standard (AMM-based) decentralized exchange. Another built Aave, a standard decentralized borrow-lending protocol.
Now say you want to build a decentralized exchange with built-in margin trading. How do you do it? Do you build a risk engine, and liquidation engine, and dex, etc.?
Let’s say Bob has USDC and wants to get 2x long ETH. You build a GUI that, with a single click:
- Buys ETH with USDC on Uniswap
- Sends ETH to Aave, lends it out, and borrows USDC
- Sends USDC to Uniswap, buys ETH
- Sends ETH to Aave, lends it out, and borrows USDC
- Sends USDC to Uniswap, buys ETH
(If you want to get clever, you can use flash loans to cut out some steps.)
The key thing is: on-chain programs are open and permissionless and in the same language, so you can stitch them together to build something greater than the sum of their parts.
So you just build a GUI that does 1–5 above with a single button click, and now you’ve built a margin trading DEX.
And if you want to provide liquidity with 2.5x leverage and yield, you can glue together borrow/lending with a DEX, which is what Alpha Homora does.
You could try doing this with Binance borrow-lending and Coinbase orderbooks, but it wouldn’t work well: you’d run into ratelimits and borrow limits and withdrawal limits and withdrawal delays and liquidation risk in the middle and all manner of issues.
In DeFi, it just works.
Another example, from Serum. Let’s say that you want to launch a fast, cheap DEX on Solana.
You could learn Rust, and build out an on-chain matching engine, and test it, and find market makers, and build GUIs, and…
Or you could fork a GUI, set the fees to your address, and do any customizations you want, and 10 minutes later you have your own state of the art DEX.
So why would large products build on Serum?
Because if you’re Robinhood and you need liquidity for your users’ trading, you don’t have to reach out to proprietary trading firms, negotiate deals, compare, deal with settlement, negotiate tickers, etc. All you need to do is, well… host a DEX GUI, and 10 minutes later you’re ready.
Let’s say you want to create a token and get it listed. You could hire devs to make tokens, BD to interface with Binance and FTX and Coinbase about listings, etc. Or you could mint a token, add a DEX market or AMM pool, and put out some offers; and 10 minutes later your token is listed.
Composability means that any application building on DeFi can automatically integrate every other one as if it was native. Every project starts as if they’d already built out their own full service financial ecosystem.
Now this doesn’t necessarily mean that DeFi will get huge.
But there are real advantages to it if it’s done well — advantages that could potentially convince a large fraction of all companies to move on-chain.
You could imagine DeFi growing to a billion users. That, really, is the goal.
How valuable would that be?
It’s really hard to tell. The following isn’t meant to be a real estimate of what it would be worth, just what it could, theoretically, may be worth.
Let’s just do some really rough calculations.
1) What would NYSE + ARCA + CME + ICE + … make if they were decentralized and charged 0.25bps on each trade?
- Well, daily trading volume is around $10T. If 0.25bp fees reduced volume by 75% and otherwise 20% of the volume moved to a decentralized venue, the resulting volume would be about $500B/day.
- That would result in $3B of annual revenue and a valuation likely around $100B.
2) How about social media?
- Open access to APIs
- Freedom to build GUIs, clients, etc.
- Open search
- Cross-posting between communities
- Censorship resistance
Say that gas costs of ~$0.00002 could be maintained on e.g. Solana with the necessary throughput
- Daily cost of $10k
- Valuation of $30B
- Daily cost of $100k
- Valuation of $800B
Pictures are tough here; they would have to use IPFS or something. Maybe you could theoretically transfer 30% of social media, so it could be $300B value.
4) How about structured products?
- Most of this is really easy to put on-chain
- Blackrock alone has $100b valuation
- So this could hit $500b total
5) How about credit cards?
- VISA handles ~2k-50k TPS; scalable blockchains can handle this and has a $500b valuation
- So we could have $1T valuation of credit card companies on-chain
…and this is just the start.
Now these are super rough calculations — total bullshit. And this isn’t value to the blockchain, it’s the value of the companies and/or DAOs that could theoretically build on the chain.
But it’s pretty clear that there could theoretically be $10T worth of companies all on-chain, all composing with each other.
Which is not to say that it will happen, just that it could.
DeFi might flame out, especially if it comes to be known for its scams and failures to scale rather than its most powerful products.
But DeFi could succeed. And if it does, it could be huge.
The potential of DeFi could reach 1 billion users and $10T of on-chain value. What about the blockchain is necessary to get there?
1) Ability to process tens of thousands of trades per second and hundreds of thousands of orders per second
2) Ability to process 10 billion social media interactions per day, which is around 100k per second
3) Gas costs of less than $0.001
4) Scaling with the world
- As the world’s technology grows, the blockchain’s throughput has to grow
5) Ability to process all submissions on a human-reaction-time scale.
- If you click a button you expect it to take ~100ms or so to process on most products — a~1s lag would be really frustrating
6) Products composing
- Otherwise you lose a lot of the value-add for companies, meaning each industry will likely want to be roughly single-sharded
7) Ability to be decentralized and open
So basically, the blockchain needs to grow into:
1) One million TPS
200k per shard for a single industry
2) $0.001 gas
Much more makes sending orders, posting updates, etc. too expensive
3) Scaling with Moore’s law
Specifically, scaling total computational capability via parallelism
4) Decentralized and open
5) 100ms block times
Which blockchains that exist today could get there?
200k transactions per second in a single shard blows out every blockchain right now.
But even more worrying, it blows out almost every blockchain’s long-term vision. Most chains, even most faster chains, aren’t planning to focus on further increasing their throughput; rather they’re resting on the laurels of being able to handle the current 100k user DeFi ecosystem cheaply.
ETH can only handle ~15 TPS, but even ETH2.0 is not planning to handle nearly 200k TPS in a single shard with atomic composability.
Faster chains generally handle ~1k TPS; the fastest are around 50k.
So the only blockchains capable of really reaching the most ambitious goals of DeFi are ones which can handle tens of thousands of TPS right now, and have plans to grow that by an order of magnitude over time.
In other words it has to be one of the fastest chains, which will also continue to scale computation on increasingly parallel silicon.
Solana meets these criteria. It can handle ~50k TPS right now; scales with Moore’s law; and has a detailed set of plans that could get another factor of 10 over the next couple years in throughput by optimizing various parts of the system. It’s gas is cheap, too.
Solana’s blocks are about 400ms right now but can be cut to around 150ms with a bunch of work.
And the core thing is that Solana is built with the expectation that it will put huge resources into scaling over time. The core team and ecosystem participants are on board with this mission and collaborating to help scale.
It would be a surprising disappointment if Solana didn’t meet the necessary criteria.
Which other chains can that be said of?
A blockchain is everything that’s necessary for DeFi to get huge. But alone it’s also nothing: just a proposal for how a group could reach consensus if they were to follow it.
For DeFi to scale people have to choose to use that blockchain. Lots of people.
Once it grows to 10m users, the rest can be self-propelled: it might become the default, obvious answer for companies investigating on-chain functionality.
But how do you get there? You need to first achieve a very different sort of consensus: a Schelling point of sorts, that if you’re going to build out DeFi, it’s a place to look.
Ethereum has this. Currently, no other chains do.
But Serum presents a path towards adoption. Already, it has the world’s most powerful fully on-chain DEX, and the ecosystem is growing quickly. Behind the scenes, applications with combined hundreds of millions of users are working on it. The building has started.
And more than that it has backing: the promise that it will have support and products and liquidity and funds. The coalition of groups building on Serum is massive in scope.
This doesn’t mean that Serum will succeed.
But it does mean that it can.
And, bit by bit, it’s inching there. This week I’ve seen fully functional beta versions of social media applications, fantasy sports, token minting, oracles, AMMs, options, bridges, and more, courtesy of the Solana Wormhole hackathon.
Where did Serum come from?
The first ever proposal for Project Serum was a Google Doc describing a protocol for archiving website screenshots on Ethereum.
It evolved over time, bit by bit, until it hit a breaking point.
Every draft had the same fundamental properties:
- Serum would be completed within two months
- Serum would never be more valuable than a moderately cool dapp.
There was a fundamental fork in the road:
- Stay the course
- Build on another chain
With path (a), Serum would have been completed so that it could have participated fully in DeFi Season; it would’ve launched a fully functional token that could be sold; and it would have already reached its peak.
With path (b), Serum could potentially scale to billions of users. But that would take months for the first products, years for a huge ecosystem, and decades for its ultimate vision. If, that is, it succeeded in ever getting any traction in the first place.
Project Serum took a sharp turn off the beaten path; and so it was born from the same driving questions and visions as this essay: do what you have to do to be able — if everything works out — to make DeFi realize its most ambitious vision.
We’ve come a long way since Serum was founded oh so many years ago. (Or, wait, I guess about three months.) There’s a lot more distance left to go; two steps forward, one step back. But I guess that’s what it means to have an ambitious goal.