Uniswap founder Hayden Adams says automated market makers are starting to show a credible path toward becoming core market infrastructure for a much wider set of financial assets. In his view, tokenization is not just a faster, cheaper, 24/7 version of today’s markets. It makes markets programmable and changes which markets can exist, who can make those markets, and which assets can trade directly against one another.
The essay, written by Adams and translated by ChainCatcher, reflects on a question he says has stayed with him through nine years working on the front edge of DeFi: can AMMs really become the core engine of all financial markets? After years of iteration and growth, he argues that the answer is becoming easier to see.
From index funds to tokenized markets
Adams opens with a comparison to the launch of index funds in 1976. When Jack Bogle introduced the product, he had hoped to raise $150 million but ended up with only $11.3 million. Rivals called it “Bogle’s folly” and attacked index investing as something that was “un-American.” Their case was simple: a fund that made no active investment decisions could not beat highly paid professionals whose entire job was to make those decisions.
Today, most U.S. fund assets sit in passive investment vehicles. Adams says tokenization is moving out of the same phase of being treated as a fringe idea. He notes that the U.S. Securities and Exchange Commission has approved tokenized stock trading on Nasdaq and the New York Stock Exchange, while DTCC, which settles nearly all U.S. securities, ran a live-environment test of a tokenized transaction in July.
Those developments are often described in the same way, he writes: tokenization as an infrastructure upgrade. The markets stay the same, only faster, cheaper and always on. Adams says that description is not wrong, but he thinks it misses the larger shift. Once markets become programmable, the structure of market making itself starts to change.
Where AMMs won first
Adams revisits the creation of Uniswap in 2018. The protocol lets anyone deposit two assets into a shared pool and earn fees from each trade, while prices adjust automatically along a curve as users buy and sell. Since launch, he says, Uniswap has operated autonomously from day one, processed more than $4.6 trillion in cumulative volume, and helped decentralized exchanges raise their share of spot trading from less than 1% of centralized exchanges to more than 20%.
He argues that AMMs first found product-market fit in long-tail markets, where many assets simply could not attract professional market makers. On Uniswap, any user can create a market in a single transaction, and issuers or early backers can become the first liquidity providers.
The next category was stablecoin pairs. For markets such as USDC/USDT, Adams says a strong passive strategy can get close enough to optimal, and lower capital costs can make up for the efficiency gap. That, in his telling, is why professional trading firms now have little reason to make these stablecoin conversion markets: passive AMMs are pushing them out with lower costs and lower required returns.
The scale moat in traditional market making
Traditional financial markets are largely controlled by market-making firms, Adams writes. Those firms bundle capital, trading strategy, execution technology, settlement and distribution into one vertically integrated business. The model emerged for practical reasons: assets lived in disconnected systems, settlement was slow, and every function needed a specialized operator, so keeping the stack under one roof made sense.
Once that structure reaches scale, fixed costs can be spread across huge trading volumes. Adams points to Citadel Securities, which he says handles about 25% of U.S. stock trading volume. Last year, he writes, the firm generated a record $12.2 billion in net trading revenue using about $21 billion in trading capital.
Many readers would see those numbers as proof that the system works well. Adams says he sees something else: a market structure that has already been tightly captured.
How blockchains unbundle the old system
On blockchains, each layer becomes contestable. Execution is handled by code. Custody and settlement become shared services that anyone can access. Work that once depended on proprietary infrastructure can be delivered through open-source software.
For AMMs, he says, capital is the scarcest input, so the edge belongs to whoever can hold inventory at the lowest cost. A trading firm needs a higher return to cover its own overhead. A liquidity provider willing to accept a lower return can compete more cheaply.
He adds that most market makers hedge away nearly all price risk, and hedging is expensive. An investor who already wants to own those assets can absorb that exposure at no added cost. For issuers, the cost of capital can even be negative, because issuers often pay professional market makers to support liquidity for newly launched assets.
That lowers the barrier to market making in DeFi. According to Adams, new participants can enter with advantages that come from cheaper capital, a willingness to hold exposures professional firms would normally hedge out, or the fact that they are the asset issuers themselves. The remaining question is whether automated strategies are good enough for those structural advantages to matter in practice.
The case for correlation pairs
A central piece of Adams’ argument is what he calls correlation pairs. He says a recent call with one of the world’s largest financial institutions prompted the point. The institution asked what the most common quote assets are in DeFi. Adams answered that Ethereum ecosystem assets usually trade against ETH, Solana ecosystem assets usually trade against SOL, stablecoins trade against each other, and those liquidity clusters are linked by a small number of highly liquid pairs.
No one explicitly designed that structure, he says. It emerged naturally, in part because liquidity providers tend to perform better when the two assets in a pool move together. Correlation lowers inventory risk for LPs and supports deeper liquidity.
He argues that as more assets become tokenized, larger financial markets could reorganize the same way. The reason they do not look like this today is that, in traditional finance, almost everything ultimately has to settle in dollars. Assets live in separate systems, and fiat rails such as SWIFT and Fedwire serve as the glue between them.
Onchain, the setup is different. Once assets are tokenized and share the same settlement layer, any asset can trade directly against any other asset. Adams gives several examples: NVDA/USD could become NVDA/SPY, with SPY/USD serving as the bridge back to dollars. An oil company’s stock could trade against an oil ETF or tokenized crude. Private credit could trade against a tokenized U.S. Treasury fund. Tokenization could also create markets across asset classes that are extremely difficult, or impossible, to support in traditional infrastructure.
Why delta neutrality can be expensive
Traditional market makers usually try to remain delta neutral. In practice, Adams says, that means keeping the dollar as the base reference point and minimizing non-dollar exposure as much as possible. When they make markets in volatile assets, they spend money to reduce that non-dollar risk, often through options hedges. That is one of the more expensive parts of the old market-making model.
Grouping assets into lower-volatility correlation pairs, then connecting those clusters with a smaller number of higher-volatility bridge pairs, can unlock meaningful efficiency gains. The most important point, in his view, is that market making becomes cheaper and more efficient when the party providing liquidity already wants to hold the underlying assets.
The higher the correlation inside a pair, the smaller the efficiency gap between today’s passive AMM strategies and the most advanced active strategies. That makes it easier for passive liquidity to compete on the strength of lower inventory costs. Adams uses NVIDIA as an example. If someone already wants to hold NVIDIA over the long run, that investor is also likely to hold SPY. In that case, he says, the efficiency gap between passive AMMs and active strategies should be much smaller in NVDA/SPY than in NVDA/USD.
Bridge pairs and market structure
If stocks mainly trade against SPY, then any trade that starts or ends in dollars would route through one pair: SPY/USD. Adams says those bridge pairs would still require highly specialized market-making expertise, but there would be far fewer of them, and the trading flow through them would be large enough to justify professional attention.
He says DeFi has already shown that this architecture can work. ETH/USDC is one of the deepest onchain liquidity markets because trades between separate liquidity clusters route through it. Passive LPs supply liquidity to correlation pairs, while active LPs compete in bridge pairs.
Investors could still buy and sell everything in dollars because routing between pools happens automatically. Liquidity, meanwhile, would concentrate where risk is lowest rather than where legacy infrastructure says it has to sit. Adams says that dynamic should gradually pull the deepest markets toward correlation pairs, which is exactly where AMMs already look strongest.
Tokenized stocks are already trading this way
Adams says the first wave of onchain correlation liquidity came from crypto-native assets, but tokenized equities have now started to show the same pattern. He writes that 10 tokenized stocks currently trade directly against SPY through Uniswap pools on Robinhood Chain.
In their first 12 days after launch, those pools handled $33 million in volume and attracted more than 11,000 users. A large share of that activity took place while U.S. equity markets were closed. Some trades even moved directly from one stock into another without touching dollars at any point, he says.
He also notes that the market has started to produce pairings between meme coins and what he calls “correlated” stocks, including an Elon-related meme coin paired with Tesla stock and a hot dog meme coin paired with Costco stock. How correlated they really are in price terms remains unclear, he writes, though “vibes” may be a form of correlation too.
AMMs are still early, Adams says
Adams closes by arguing that correlation pairs are only one part of the picture. Another is AMM design and customizability. He says Uniswap v4 Hooks can support fully customized markets and materially improve LP returns. As one example, he points to the recently launched DualPool Hook, which can put idle capital in passive AMMs to work earning lending yield when that capital is not being used for trades.
Even with roughly $4.6 trillion in lifetime volume, Adams says AMMs remain in a very early stage, with many possible avenues still open to improve competitiveness. Uniswap Labs, its partners and the broader ecosystem are exploring more ways to raise LP returns, he writes.
He ends by returning to the original index-fund comparison. In 1976, critics argued that a fund making no decisions could never beat professionals paid to make them. Fifty years later, Adams says, that “decisionless” fund has beaten about 90% of professional investors. More than that, index funds broadened access to investing and improved ordinary lives. He believes passive liquidity can follow a similar path, with even larger consequences because it can cut the barrier to creating and participating in markets.