Hayden Adams broke a five-year silence on Monday. The Uniswap founder, who has not posted a blog since 2019, published a short essay arguing that automated market makers (AMMs) will eventually dominate the largest financial markets—namely, tokenized equities, ETFs, and index funds. Within 48 hours, a former XTX Markets trader fired back on X, calling AMMs' relevance in those markets "zero" and predicting their eventual extinction. The rapid exchange, heavy on assertion and light on data, is a classic signal: an industry wrestling with a paradigm shift, but lacking the quantitative rigor to settle it.
This is not a debate about technology. It is a debate about market microstructure—and the gap between theory and operational reality. As someone who has spent years dissecting protocol flaws from Tezos' formal verification to EigenLayer's slashing vectors, I recognize the pattern: grand claims from both sides, anchored in first principles, but detached from the messy constraints of execution. The proof is in the logic, not the promise. And the logic here is incomplete.
Context: The Tokenized Asset Thesis and Its Discontents
The debate sits at the intersection of two narratives: the ongoing tokenization of real-world assets (RWA) and the maturation of DeFi infrastructure. Hayden Adams posits that as securities like NVIDIA stock or SPY ETF become tokenized, the natural trading venue is not a traditional order book but an AMM pool. The rationale is elegant: in a world where assets are no longer quoted in dollars but in any other tokenized asset, AMMs provide the seamless interchangeability required. "Tokenized stocks will mainly be traded against index funds," he wrote, implying a cross-asset liquidity layer that order books cannot efficiently serve.
The former XTX trader, speaking from years of professional market-making, countered with brutal pragmatism. "Who would ever want to sell their NVIDIA for SPY?" he asked, highlighting the irrelevance of the cross-asset thesis. He argued that AMMs lack the price discovery, inventory management, and risk hedging capabilities that professional market makers provide. In high-volume, low-volatility assets like NVDA or SPY, the bid-ask spreads on traditional order books are razor-thin—a market structure AMMs cannot replicate due to their constant function mechanics.
Both positions are rooted in first principles. But neither provides the data needed to adjudicate. The debate is a symptom of a market in transition, where narratives drive expectations before empirical evidence exists.
Core: A Systematic Teardown of the AMM vs. Order Book Argument
Let me apply the same adversarial, worst-case modeling I used when analyzing Terra's algorithmic stablecoin or Yearn's slippage assumptions. The core question is not whether AMMs can technically handle tokenized stocks—they can, in the same way a bicycle can technically cross a highway. The question is whether they can do so at competitive efficiency, cost, and risk tolerance.
Point 1: The Mathematical Constraint of Constant Function Market Makers
Uniswap's AMM (v2 and v3) relies on a constant product formula: x * y = k. For high-liquidity, low-volatility assets like NVDA or SPY, the price impact for a given trade size is a function of pool depth. Professional market makers, by contrast, actively manage inventory, hedge delta, and adjust quotes based on real-time order flow and volatility. The AMM is passive; it reacts to trades after they occur. In a market where a single large sell order can move the price by several basis points, the passive model is structurally loss-making for liquidity providers. The result is wider spreads, higher slippage, and ultimately, inferior execution quality for end users.
Based on my audit experience, including the 2020 Yearn vault incident where I discovered their rebalancing algorithm assumed constant market depth, I know that such assumptions are dangerous. The assumption that AMMs can compete with professional market makers on price efficiency is analogous to assuming a fixed-rate bond can compete with a high-frequency trading desk. The underlying mathematics does not support it without additional mechanisms—such as concentrated liquidity or dynamic fee adjustments—which Uniswap v3 introduced but which still lag behind the sophistication of a Citadel or XTX.
Point 2: The Market Microstructure Gap
The former XTX trader's argument is not about technology but about market microstructure. In traditional finance, the role of a market maker is to provide two-sided quotes, absorb order flow, and manage risk through hedging. They use sophisticated models to predict short-term price movements and adjust their quotes accordingly. An AMM, even with concentrated liquidity, cannot replicate this because it does not hold inventory risk actively—it only provides liquidity at predetermined price ranges. The result is that AMMs are best suited for assets with high volatility and low liquidity, where the spread is wide enough to compensate LPs for the risk of adverse selection. Tokenized stocks and ETFs, however, are high-liquidity, low-volatility assets. The spreads are already tight. An AMM would need to offer even tighter spreads to attract traders, which is mathematically impossible given the passive LP model.
Point 3: The Regulatory Landmine
Hayden Adams' thesis conveniently ignores the compliance bottleneck. Tokenized securities are securities. Their trading, under U.S. law, likely requires registration as an exchange or alternative trading system (ATS). AMMs are permissionless, pseudonymous, and global. The SEC has not sanctioned any AMM for trading tokenized equities. The risk of enforcement is high. In 2021, I analyzed the Bored Ape Yacht Club metadata storage and found that 30% of top NFT collections had similar centralization vulnerabilities. The lesson: what is marketed as "decentralized" often has hidden centralization. Similarly, the compliance path for AMM trading tokenized securities is not just unclear—it is likely blocked without a fundamental redesign of the protocol. The former XTX trader's skepticism may be rooted in this reality: professional market makers already operate within regulated frameworks, and they know the cost of compliance. An AMM cannot undercut that without breaking the law.
Point 4: The Unspoken Advantage of the Order Book
Let me introduce a concept from my 2022 Terra/Luna collapse analysis: the requirement for infinite growth. In Terra's case, the seigniorage model required continuous expansion to maintain peg stability. In the AMM case, the requirement is for continuous liquidity provisioning without active risk management. That is not mathematically sustainable for large-ticket assets. The former XTX trader is right to call the AMM model "zero" for those markets—not because AMMs are useless, but because they are structurally inferior to order books for the specific use case of high-volume, low-volatility assets. The proof is in the logic: the constant product formula cannot compete with a dynamic pricing model that accounts for order flow imbalance, inventory risk, and hedging costs.
Contrarian: Where the Bulls Got It Right
Despite my skepticism, Hayden Adams' vision is not without merit. The tokenized asset market is still nascent. The volume is orders of magnitude smaller than traditional markets. In that early stage, an AMM can provide the necessary liquidity bootstrapping that order books require a professional market maker to initiate. Uniswap's v3 concentrated liquidity allows LPs to set price ranges, mimicking the behavior of a market maker to some extent. In the long tail of tokenized assets—say, a tokenized version of a small-cap stock—the AMM's passive model may be sufficient because the volume is low and the spreads are wide.
Moreover, the former XTX trader's critique assumes that the ultimate trading venue for tokenized assets will be on-chain order books or AMMs. But the future may be hybrid: an AMM as the base layer of liquidity, with a professional market maker providing additional depth through requests for quotes (RFQ) or limit orders. Uniswap v4's hooks enable exactly this kind of programmability. A hook could allow a designated market maker to manage a pool's liquidity in a dynamic way, combining the best of both worlds. The complexity is high, but Complexity is the camouflage for incompetence only if the implementation is poor. With proper design, hybrid models could emerge.
Another blind spot in the skeptic's argument is the assumption that tokenized assets will trade in the same way as their traditional counterparts. The nature of on-chain ownership may change the demand for trading. For example, if a tokenized stock is used as collateral in DeFi lending, the cross-asset trading (NVIDIA for SPY) may become a necessary risk management tool. Hayden Adams is anticipating a world where tokenized assets are not just mirrors of existing securities but programmable assets that interact with smart contracts. In that world, the AMM's composability is a feature, not a bug.
Takeaway: The Burden of Proof Has Not Been Met
This debate is a microcosm of the broader crypto-market tension: narrative-driven speculation versus data-driven analysis. Both sides have made claims that are currently unverifiable. The former XTX trader has the weight of decades of market microstructure research and practice. Hayden Adams has the weight of a protocol that has processed billions in volume and survived multiple market cycles. But neither has provided the quantitative evidence needed to settle the matter.
I want to see the data. What is the simulated slippage for a 100,000 NVDA-equivalent trade on a Uniswap v3 pool at current depth? What is the bid-ask spread on a hypothetical tokenized SPY pool compared to the commission-free Robinhood order book? What is the cost of hedging for an LP in such a pool? Without these numbers, the debate is noise. Assume malice, verify everything, trust nothing.
The market will decide, but the market is often irrational. The real test will come when the first large-scale tokenized equity hits a major AMM. Until then, I remain skeptical of both positions. The proof is in the logic, not the promise. And the logic here is incomplete. Yields are just risk wearing a tuxedo. Ownership is a ledger entry, not a feeling. The AMM may win the long tail, but the core of the market will remain with order books until the underlying mathematics changes. And that is a change I do not see happening in the next five years.
Static analysis reveals what marketing hides. The marketing here is a battle for the future of finance. The static analysis reveals a gap between vision and reality that only time—and data—can fill.