The Angle of Repose

Welcome to the first installment of The Angle of Repose, Thompson Street's home for irreverent commentary on what's happening in crypto, tech, and the global financial system. It is so named after the steepest angle at which a pile of loose, granular materials remains stable without sliding. We will aim to circulate this brief when we have something interesting to say. No more, no less.

In this issue:

• Purpose and pragmatism: Hyperliquid as the marquee case study in crypto's new willingness to make pragmatic, opinionated trade-offs.
• The (onchain) price is right: Why proprietary AMMs and other novel mechanisms have pushed onchain price discovery into the lead for certain assets.
• Whose tokenized asset is this, anyway? — Stablecoins and RWAs are growing fast, but ubiquity is a multi-decade build; the cultural shift in finance is the bigger story.

Purpose and pragmatism

• Hyperliquid's rise is the marquee case study in crypto's new willingness to be opinionated and make pragmatic trade-offs in service of a specific application.
• The cypherpunk-vs-product tension that hamstrung earlier projects' broad adoption is giving way to application-specific design, even when that means leaning into less-pure architectures.

Our first order of business is level-setting on the state of the crypto ecosystem, where we explain how we're thinking about innovations in the space and their diffusion into the broader financial system. Our views have been informed by independent research and analysis and by conversations with industry participants.

The rise of Hyperliquid has been one of the biggest stories in crypto over the past year. Hyperliquid is an L1 blockchain built by a small, self-funded team of former HFT quants operating out of Singapore. Along the way it has become, by some measures, the most-profitable company in the world on a per-worker basis (approximately $192 million in profit on a team of 11).

Under the hood, Hyperliquid is a blockchain with two separate (but connected) execution environments: HyperCore and HyperEVM. HyperCore contains a small number of financial primitives that are optimized for high-frequency, low-latency trading. HyperEVM is a general-purpose programming environment based on the Ethereum Virtual Machine (EVM). Think of it this way: HyperCore is like a GPU (rigid and optimized for speed within a narrow set of tasks) and HyperEVM is like a CPU (comparatively slow but extensible enough to handle almost any computational task). Today, Hyperliquid has a small and quasi-permissioned validator set, and much of its underlying source code remains proprietary. Its design optimizations, coupled with a clever economic model, mean that Hyperliquid can operate a true central limit-order book (CLOB) entirely onchain.

As a result, crypto price discovery on Hyperliquid leads much of the centralized exchange market. Hyperliquid is also pushing the frontier of trading for traditional assets like commodities (e.g., oil, gold) and some securities (e.g., equities, private-market securities or pre-IPO stocks) thanks to its deep liquidity and 24/7 markets. Research from blockchain analytics firm Allium finds that onchain equities perps trade closer to the market open than the previous day's close. Exchange giants CME and ICE have taken notice.

Hyperliquid Builder Codes incentivize flow by giving a cut of protocol revenue to distribution channels, such as wallets and front-ends (Data from flowscan.xyz/builders showing past 30 days; accessed May 11, 2026).

We could spend 40k words telling you about Hyperliquid and why we think it has been successful, but that's not why we're here. What we find more compelling is an observation that Hyperliquid is the most-prominent example of a new kind of pragmatism brought to the crypto space.

For much of crypto's existence, the end-user experience has been hamstrung by projects trying to balance cypherpunk culture and values with products that are mass-market ready. Yet, in practice, many of the things protocols and projects preached around decentralization and credible neutrality were a) either a complete LARP to cater to crypto's vocal community of purity-test enforcers, or b) unnecessary for what they were trying to accomplish on the product front.

The purpose of a system is what it does. Hyperliquid chose to be opinionated, prioritized trading as its primary application, and the traders (many with HFT backgrounds, including institutional players) found a home. Everything has trade offs. From a sociocultural POV, crypto seems ready to make them depending on context.

The (onchain) price is right

• Performance and permissionless are no longer antonyms: for certain asset classes, onchain price discovery now leads the centralized exchanges.
• Proprietary AMMs and other novel mechanisms are real zero-to-one innovation and a sign that elite HFT and systems talent is still choosing crypto, even as AI labs pull on the same hiring pool.

We are constantly asking ourselves whether a given project/product in crypto is a true zero-to-one innovation or just a replication of something else. One area we're excited about today is the proliferation of new mechanisms for onchain price discovery.

Traditionally, centralized exchanges and platforms have facilitated price discovery using mechanisms like CLOBs, auctions, or request-for-quote (RFQ) systems. All rely on market makers communicating bilaterally with the exchange to constantly update their quotes based on the state of the market. This kind of market microstructure leads to an arms race in speed technology and a fairly efficient market in terms of spreads.

Historically, permissionless blockchains have been limited by low throughput, high latency, and costly transaction fees. As a consequence, decentralized exchanges have relied on auctions or relatively-simple algorithms incentivized by arbitrage against centralized exchange prices to facilitate trading. Until about 12 months ago. The Solana ecosystem is now the frontier for novel mechanism design in decentralized exchange and price discovery. The most important from our vantage point are proprietary automated market makers, or propAMMs.

Traditional AMMs are public programs that run on the blockchain and define a relationship between the prices and relative quantities of two assets in a pool of liquidity. In the most basic example, known as a constant product market maker (CPMM), the quantity of asset X, multiplied by the quantity of asset Y, must always equal a constant, K. AMMs are genuinely innovative because they enable always-on liquidity and don't require traditional market makers to submit and update quotes. However, a downside of AMMs is that they spread liquidity across the entire price curve, even at prices that wouldn't clear the market.

PropAMMs are programs that run on the blockchain where the price curve is not a fixed, public formula, but rather a proprietary market-maker strategy that updates continuously based on market conditions. Unlike traditional CLOBs, propAMMs let market makers quote CLOB-like liquidity (in terms of depth and granularity) without having to constantly submit, refresh, and cancel visible quotes, which take time — even with colocation. And like a traditional AMM, the liquidity lives inside the program, so any quote is automatically executable. PropAMMs on Solana are regularly beating the most-liquid centralized exchanges on all-in costs to retail traders. (For a more detailed discussion on propAMMs, see this.)

Like the Hyperliquid example, we highlight propAMMs to make a broader point: Crypto prices are down and popular sentiment, even among many crypto natives, is at a low point, but we see genuine innovation and maturity in some places. Trading has always been crypto's first product-market fit, but advances in trading technology and market microstructure are not particularly salient to end-users who think in terms of quotes and not spread tightness and all-in cost. Past periods of innovation too often paired developments in market structure with vague cash-flow promises through protocol-affiliated tokens that could never quite deliver. The ultimate question is whether some of the recent breakthroughs can break containment and form the future trading layer for more traditional assets.

Whose tokenized asset is this, anyway?

• Stablecoins and RWAs are taking off, but they remain a long way from ubiquity — and headline volume comparisons (e.g., "stablecoins surpass ACH") flatter the reality.
• The most-profound impact of tokenization so far may be cultural: small, rebellious teams have replaced 100-person IT departments at the cutting edge of finance.

Stablecoins and real-world assets (RWAs) are getting a lot of press, but how much of the activity we're seeing onchain is going to break into the mainstream? Let it be stated for the record that we're extremely excited about stablecoins and tokenization, in part because of the innovations described earlier in this brief. With that out of the way, we're also going to be blunt: reinventing the financial system in terms of tech, regulation, and legal underpinnings is a multi-decade endeavor.

We do not refute that stablecoin volumes are up and to the right. However, their impact, at present, appears to be greatly exaggerated within the tech and crypto information bubbles. For example, we've seen numerous reports that stablecoins "recently surpassed ACH in monthly payment volume at $7.5tn." This is an apples-to-oranges comparison. Nearly every person or business in America with a bank account uses ACH on an almost-daily basis. You'd be hard pressed to find a single person who has used stablecoins in the past year unless you're talking to someone in the industry. This is not to say they are useless or that they'll never reach ubiquity. There's simply a lot of infrastructure and market structure yet to be built, and the most-compelling use-case today is for cross-border payments or for access to dollars in emerging markets, neither of which is legible to most advanced-economy participants.

The same goes for RWAs and tokenized assets. The data show persistent growth, but private conversations with non-crypto market participants are more sobering. The refrain that we keep hearing is that, while market participants have high conviction that traditional markets will be rebuilt using blockchains over the next 10+ years, there are coordination and risk challenges in the short term that may be difficult to overcome. For example, if 99% of market liquidity is on traditional venues, most institutions can't underwrite the risk of trading the same assets in a market that has 1% of liquidity on top of a number of operational and legal risks that still need to be sorted. Markets likely to find success in tokenization are not those that are deep, liquid, and highly efficient, but rather those that feature long-tail assets and fragmented trading where the switching costs are low. Promises of efficiency savings and lower back-office costs are not enough, particularly when decision-makers can't see an immediate impact on PnLs. Where this trend might not hold is at the distribution layer, where tokenization effectively unbundles trading and custody infrastructure so users' portfolios are portable across providers, unlike the traditional broker-customer relationship.

Ultimately, perhaps the most-profound impact of tokenization today is how completely it has changed the cultural DNA of the finance industry. The most-innovative teams in finance today are not led by MBAs with hundred-person IT departments. They are small teams of rebels building at the edge.