The year is missing from the transcript, and that is the first red flag. On a date recorded only as October 15, Jeff Yan, co-founder of Hyperliquid, sat down with Bloomberg's "Odd Lots" and made a claim that, if accepted, would redraw the compliance map for every perpetual DEX on earth. His argument, in essence: the blockchain is neutral infrastructure, and the regulatory obligation should sit with the service provider, not with the protocol. No validator set was disclosed. No audit was cited. No governance model was named. Just a framing โ infrastructure is neutral, operators are liable โ delivered to a mainstream finance audience.
I have learned to distrust clean framings. In late 2017, while working as a junior data analyst at a Dublin fintech, I spent forty hours auditing the distribution script of the PotCoin ICO. Buried in the logic was an integer overflow that could have drained user wallets. I filed a bug bounty report through GitHub; it was accepted; I was paid two thousand dollars in ETH. The lesson hardened into a rule I have never broken: if I cannot audit the logic, I do not trade the token. Ledgers do not lie, only the auditors do. So when a founder tells me his chain is "just pipes," I go looking for who owns the pipes and who can turn the valve. What I found inside Hyperliquid is a layer cake whose bottom tier is far thinner than the marketing suggests.
Context: The Architecture Is the Argument
Before attacking the argument, I want to lay out the architecture, because the architecture is the argument.
Hyperliquid runs a self-built Layer 1 with an on-chain central limit order book for perpetual futures. That is the technical core. It is not a rollup on Ethereum, and it is not an AMM like GMX, where trades clear against a pooled liquidity curve. It is a bespoke chain running a full order book โ bids, asks, matching, settlement โ the way a centralized exchange does, except the matching engine is on-chain. dYdX v4 made a similar migration to its own chain. Hyperliquid's differentiator is performance: the pitch is that a fully on-chain order book can behave like a centralized matching engine without the custody risk.
The team is real-named. The co-founder, Jeff Yan, is the one carrying the regulatory message. That matters. Anonymous teams cannot lobby; named founders can.
The regulatory claim breaks into three load-bearing parts. First, the blockchain network is neutral infrastructure, comparable to the internet or to AWS โ it moves data and settles state but does not itself operate a financial business. Second, the compliance obligation falls on the financial service provider: whoever actually runs the market and holds control. Third, the goal is explicit โ reduce the compliance burden of entering the US market.
That third point is not philosophical. Hyperliquid does not onboard US users. The protocol has almost certainly geo-fenced them, which means the stack can already identify where a user sits โ a detail that quietly undercuts the "we are just neutral pipes" line, because pipes do not usually know your passport. And sitting beside the argument is a live commercial example: Payward, the parent company of Kraken, plans to offer derivatives to US investors through a regulated platform, reportedly on top of Hyperliquid's technology. That is the responsibility-layering model with a body attached: technology from one party, license from another, customer relationship owned by the regulated front.
So the question is not whether the argument is clever. It is. The question is whether a self-built L1 with a deeply involved founding team can credibly claim to be neutral infrastructure. That is where the layer cake collapses.
The Layering Model Is Elegant โ and Self-Serving
Yan's model has accounting elegance. Draw a line between the settlement layer and the business layer. On one side sits a neutral chain that records state. On the other sits a licensed operator that handles KYC, AML, and the actual customer relationship. If the line holds, the protocol avoids the compliance cost and the licensed partner absorbs it. The protocol keeps the light-asset role of a technology supplier and the heavy, capital-intensive role of a regulated broker lands elsewhere.
I have run this exact structure against my own portfolio rules, and the rule that applies is simple: you can only outsource a function if the boundary is enforceable. During DeFi Summer 2020, I managed a personal book of roughly fifty thousand euros across Compound and Uniswap, tracking real-time yield APYs in a spreadsheet I built myself. When Compound introduced its COMP incentive, I rebalanced to capture a 15% annualized incentive yield before the market corrected โ not because I believed the community, but because the mechanism was auditable and the exit was pre-defined. The boundary between "my risk" and "the protocol's risk" was written in code and enforceable at the contract level. A boundary that exists only in a podcast interview is not a boundary. It is a hope.
The problem with the layering model is structural. Hyperliquid built its own L1. Vertical integration is the opposite of neutrality. When the same team owns the consensus layer, the execution layer, and the application layer, there is no independent party to point at. The infrastructure and the service provider are the same entity wearing two hats, and a regulator applying substance-over-form will notice that the hats share a head.
The Howey Test Does Not Care About Your Analogy
Run the Howey test against the platform and the result is uncomfortable.
Money investment: yes. Users deposit capital to trade and to earn. Common enterprise: yes. The platform's liquidity, order flow, and incentive programs form a shared venture whose fortunes rise and fall together. Expectation of profit: yes. Traders are there to make money; that is the entire premise of a perpetual DEX. The fourth prong โ profits from the efforts of others โ is the contested one. Yan's position is that profits come from other market participants, not from the team's managerial effort. That is the hinge the whole argument swings on.
Here is the friction. A self-built L1 whose validator set is likely permissioned, and whose roadmap is team-controlled, is not a passive marketplace. It is an operated system. If the team sets fees, ships upgrades, curates listings, and runs the incentives, a regulator will see managerial effort everywhere. The SEC's "sufficient decentralization" language gives the argument a foothold โ but that doctrine was written for systems where no single party retains effective control. Hyperliquid, by building its own chain, has made that showing harder, not easier.
The CFTC has form here, and this is where the argument gets expensive. Ooki DAO was held liable. bZx was pursued. The precedent is that "we are just a protocol" is not a defense when a controlling group can be identified. If anything, a self-built L1 makes the controlling group easier to identify, because the chain is bespoke. You cannot hide behind a shared public network you did not build. Hyperliquid built this one, so Hyperliquid answers for it.
I will be precise about my confidence here. I cannot prove the validator set is concentrated. What I can say is that self-built L1s almost universally launch with permissioned or semi-permissioned validator sets, and that nothing in the disclosed material rebuts the default. When disclosure is missing, the underwriter assumes the worse case. That is not cynicism. That is how you price an unquantified variable.
The Decentralization Paradox
This is the structural trap, and it deserves to be stated plainly.
The more centralized the L1, the weaker the "neutral infrastructure" claim. The more decentralized the L1, the more the team surrenders control over the economics it wants to capture. Hyperliquid cannot have both. A permissioned validator set that the team can influence undermines neutrality. A fully open validator set that the team cannot influence threatens the business model. This is not a solvable problem; it is a trade-off the team has to live inside, and the argument conveniently sidesteps it.
Beta is the tax you pay for ignorance โ and the tax here is paid by anyone who accepts the "infrastructure" label without checking the validator distribution. The source material flags the gap: the validator set size is unknown, no audit is mentioned, and the governance model is undisclosed. Those are not minor omissions. They are the load-bearing walls of the entire claim. A chain is only as neutral as its validator set is independent. If that set is small, correlated, or team-adjacent, then "infrastructure" is a marketing word, not a technical description.
I spent three months in 2026 stress-testing an autonomous trading agent against historical bear-market data, and I found the same pattern in a different domain: the parts of a system that are not disclosed are almost always the parts that are weakest. I rewrote the agent's core logic to enforce strict position sizing, because the original parameters were too aggressive in high volatility and would have produced a 20% drawdown in backtest. The lesson transfers. When a protocol declines to publish its validator distribution, you do not assume good faith. You assume the distribution is the reason it is unpublished.
The Payward Experiment Is the Real Test
The most interesting fact in this entire story is not the argument. It is the existence of Payward.
Payward, Kraken's parent, intends to route derivatives to US investors through a regulated platform, reportedly on top of Hyperliquid's stack. That is the layering model with a body: technology from the protocol, license from the exchange, customer relationship owned by the regulated entity. If this clears, it becomes the template โ DeFi protocols reach the US not directly, but as B2B technology vendors standing behind a licensed front.
Watch what this does to market structure. Hyperliquid stops being a consumer-facing DEX and becomes a wholesale venue. The addressable market expands, because US demand is the largest pool of trading capital on earth, and the protocol never has to hold a US license. The exchange becomes a distributor. The protocol becomes the engine. The compliance cost sits in the middle, on the party that can actually be supervised.
This is genuinely clever, and I will give it credit. It is also unproven. One case is not a precedent. The question is whether the SEC and CFTC accept the split or pierce it. If they pierce it โ if they decide the engine and the distributor are one economic unit โ the model dies, and the protocol is left holding an unlicensed-derivatives problem in its largest target market. In January 2024, after the spot Bitcoin ETF approval, I built a Python script to track the spread between the ETF's spot price and the Coinbase Premium Index, and I captured a 2% premium discrepancy for roughly twelve thousand euros of profit over two weeks. The edge existed because institutional infrastructure creates predictable inefficiencies. The Payward case is the same shape: an infrastructure change that creates a temporary, exploitable gap. But gaps close. And a gap that depends on one regulator's tolerance is not an edge; it is a bet.
The Competitive Read
Strip away the philosophy and this is a positioning move in the perpetual DEX race.
dYdX went to its own chain. GMX runs an AMM model. Hyperliquid runs a self-built L1 with a CLOB and is now arguing that the regulatory burden should slide off the protocol and onto operators. If that framing sticks, Hyperliquid gets a first-mover advantage in the compliant-adjacent niche: it can sell its engine to licensed venues while competitors stay stuck behind the US wall.
The B2B pivot is the real strategy. Yan's AWS analogy is not decoration. It is the business model. Sell compute-like infrastructure to institutions, let them carry the license, and collect the throughput. If it works, network effects compound with every downstream integrator. If it fails, the analogy becomes the trap: AWS is a utility precisely because it is indifferent to what runs on it. Hyperliquid is not indifferent. It runs the market.
Put a number on the strategic value. If US perpetual notional flow runs into the hundreds of billions annually and Hyperliquid captures even a single-digit share through licensed fronts, the incremental fee revenue dwarfs anything it earns offshore. That is why the founder went on a mainstream finance podcast instead of a crypto one. The audience was the regulator, not the trader.
What the Infrastructure Framing Really Does
There is a reason the transcript leans on the internet and AWS. Both analogies perform the same trick: they reduce a financial regulatory question to a technical neutrality question. "We are just pipes" is easier to sell to a policymaker than "we are an unregistered derivatives venue seeking a lighter regime." The framing moves the debate from whether the chain is legal to who holds the compliance duty. That reframing is the actual product being shipped.
Liquidity is the only truth in a fragmented chain โ and the truth here is that the framing, not the technology, is what is being marketed. The tech is real. The framing is the arbitrage.
The Contrarian Angle: Integration Is the Liability
The consensus read is that Hyperliquid is making a bold, savvy play for US market access. I think that read is backwards. The self-built L1 does not enable the layering argument. It sabotages it.
Here is the counterintuitive angle. Everyone assumes vertical integration is strength โ you control the full stack, you capture the full margin. In a regulatory context, vertical integration is a liability. A protocol that sits on Ethereum and merely deploys contracts has a plausible story: "we are just code on someone else's neutral chain." A protocol that built its own chain, its own order book, and its own matching engine has no one else to blame. It is the service provider. It is the exchange. The chain is a corporate subsidiary, not a public utility.
So the very thing Hyperliquid is proudest of โ the self-built L1 that gives it performance โ is the thing that most undermines its regulatory defense. The neutrality claim requires distance between the protocol and the operator. Hyperliquid has collapsed that distance to zero.
The second blind spot is the retail framing. The argument is dressed as a defense of permissionless finance. But follow the incentives and it points the other way. The beneficiaries of responsibility layering are institutions and licensed intermediaries โ the Paywards of the world โ not the retail trader who gets onboarded through a geo-fenced front end. The model routes US users not to the neutral protocol but to a supervised distributor. That is centralized finance wearing a decentralized costume. The decentralization language serves the compliance arbitrage, not the user.
Volatility is not risk; impermanent loss is. And the impermanent loss here is the gap between the narrative and the structure. The narrative says neutral infrastructure. The structure says controlled chain, controlled roadmap, controlled incentives. Anyone pricing this story at face value is paying a premium for a label the architecture does not support.
There is one more blind spot, and it is the one I keep coming back to. The argument assumes the regulator will accept the definition of "service provider" that Hyperliquid prefers. But the regulator gets to define it. If the CFTC decides the team that built the chain, sets the fees, and ships the upgrades is the service provider, then the layering model does not protect Hyperliquid โ it indicts it. The founder's framing is a request, not a ruling. The gap between the two is where the risk lives.
Takeaway
The trade here is not the token, if one exists. It is the precedent. Watch Payward's approval path, because that single case decides whether "technology plus license" becomes the template for US DeFi access or a one-off that regulators pierce. If it clears, the entire perpetual DEX sector gets a B2B on-ramp and the addressable market re-rates. If it stalls, Hyperliquid's neutrality argument loses its only living proof.
The algorithm executes, but the human decides. Right now the human deciding is a regulator who has not yet spoken. Until that voice arrives, the layer cake looks elegant. It also looks like it might be hollow โ and hollow cakes do not hold weight when the pressure comes.
Sanity checks before sanity wins.