The Entity and the Protocol: Hyperliquid's Singapore Position and the Fragile Art of Decentralized Escape

There is a phrase that has been doing more legal work this year than any smart contract I have audited, and it is only two words long. It is not "exchange." It is not "broker." It is not "company." It is "decentralized venue." Those two words appeared, quietly, in a Financial Times report about Hyperliquid and its relationship with the Monetary Authority of Singapore, and they carry more weight than any upgrade, any funding-rate mechanism, any token unlock schedule I have read in the past twelve months. Because a venue, in the language of regulators, is a place. A place can be licensed. A place can be raided. A place can be shut. But a protocol — a set of instructions executing across a permissionless network — is not a place at all. It is a condition. And you cannot license a condition. I have been watching this particular drama unfold with the peculiar attention of someone who has spent a decade watching the same argument dressed in new clothes. The report, as far as we can reconstruct it from the available information, tells us four things. Hyperliquid has a presence, an entity, a physical footprint in Singapore. It does not hold a license from the MAS. People described as familiar with the regulator's thinking believe the decentralized venue falls outside the regulator's jurisdiction. And Singapore, as a jurisdiction, draws a hard line between crypto derivatives offered through approved channels and those offered through other channels. That is the entire factual skeleton. Four bones. And yet within those four bones lives the single most consequential unresolved question in decentralized finance: when a protocol has a body, can the body be held responsible for what the protocol does? I want to be honest with you about why this matters to me, because I do not think I can write about it honestly otherwise. I lost eighty-five percent of my portfolio in 2022. I sat in a small apartment in Manila and watched numbers that represented three years of savings dissolve into something that looked like a rounding error. And the thing that kept me in this space, the thing that pulled me back from the edge of walking away forever, was not the promise of recovery. It was the promise of meaning. I believed — I still believe — that programmable money is one of the few genuinely new tools human beings have invented for redistributing power away from institutions and toward individuals. But belief without rigor is just sentimentality. And sentimentality is how people lose money. So when I read about Hyperliquid and Singapore, I do not read it as a triumph of decentralization. I read it as a stress test. From the ashes of 2022, we planted seeds for 2030 — and now, in the winter of 2025, we are finally finding out which of those seeds were real. CONTEXT To understand what Hyperliquid is doing in Singapore, you have to understand what Hyperliquid is, and you have to understand what Singapore is, and you have to understand that the two of them are engaged in a negotiation that neither side wants to describe as a negotiation. Hyperliquid is, in the simplest possible terms, an on-chain order book for perpetual futures. Perpetual futures — perps, in the vernacular — are derivative contracts with no expiry date, kept anchored to the spot price of an underlying asset through a mechanism called the funding rate. They are the single most traded instrument in crypto, the engine of leverage, the place where fortunes are made in a weekend and unmade in an hour. Most perps in crypto are traded on centralized exchanges: Binance, Bybit, OKX, and a long tail of offshore venues that operate in a regulatory gray zone so wide you could land a cargo plane in it. A much smaller number are traded on decentralized platforms, and of those, the most technically ambitious do something unusual: instead of running their matching engine on a private server, they run it on a blockchain, so that every order, every fill, every liquidation is a public, verifiable transaction. Hyperliquid takes this ambition further than almost anyone. Rather than building on Ethereum or an existing layer-two rollup, it built its own layer-one blockchain, a purpose-built chain optimized for the specific demands of high-frequency trading. The consensus mechanism, known as HyperBFT, is designed to confirm transactions in fractions of a second, fast enough to compete with the latency of a centralized venue. The order book lives on-chain. The liquidations are on-chain. The vaults, the liquidator bots, the market-making — all of it is, in theory, transparent and permissionless. If you want to understand why this is technically remarkable, consider the problem it solves. A centralized exchange can match a million orders a second because it controls a database. A decentralized exchange that wants the same throughput must achieve consensus among a distributed set of validators on every single one of those orders. Hyperliquid's answer is to make the chain so fast and so cheap that consensus becomes almost invisible. It is an elegant solution to a brutal problem, and it is the reason the platform has accumulated, at various points, a genuinely astonishing share of on-chain perpetual volume. Now consider Singapore. For most of the last decade, the city-state has positioned itself as the most sophisticated crypto jurisdiction in Asia — arguably in the world. It has a real regulator, the Monetary Authority of Singapore, which is not a figurehead but an institution with genuine technical competence and a willingness to draw lines. It has a licensing regime, the Payment Services Act, which governs digital payment token services. It has a derivatives framework that treats crypto derivatives as a regulated activity. And it has made a deliberate, repeated choice: it welcomes innovation, it welcomes capital, it welcomes the headquarters of serious projects — but it does not welcome retail speculation in leveraged derivatives offered outside its approved channels. This last point is the crux, and it is the point that most coverage of the Hyperliquid story misses. Singapore does not ban crypto derivatives. It regulates who may offer them. There is a category of approved exchanges — institutional venues, licensed, supervised — through which derivatives may be offered. And there is everything else. If you are offering retail customers leveraged exposure to crypto through a channel that is not approved, you are operating in the space that the MAS has explicitly said it does not want to bless. The distinction is not between crypto and not-crypto. It is between approved channels and unapproved ones. So when we place Hyperliquid's Singapore entity next to Singapore's regulatory posture, we are placing two things side by side that do not obviously fit together. A platform whose core product is perpetual futures. A jurisdiction that regulates perpetual futures. A physical presence inside that jurisdiction. And a claim — sourced, crucially, not to the regulator but to anonymous people familiar with the regulator's thinking — that the platform is outside the regulator's reach. I have learned, over twelve years of watching this industry, to be suspicious of claims that are too convenient. The claim "we are not regulated" is convenient for the platform. The claim "we do not regulate them" is convenient for the regulator, because it allows the jurisdiction to keep the economic benefits of hosting a major protocol without the political cost of blessing its most controversial product. Both sides have an incentive to let the ambiguity stand. And ambiguity, in regulatory matters, is not a stable state. It is a temporary equilibrium that persists only until someone — a court, a parliament, a change in the political weather — decides to break it. CORE Here is the technical heart of the matter, and I want to slow down here, because this is where the whole story actually lives. When a regulator asks whether it has jurisdiction over a decentralized platform, it is not really asking a philosophical question about decentralization. It is asking a sequence of concrete, mundane, testable questions. Who signs the contracts? Who holds the keys? Who can upgrade the code? Who can freeze the funds? Who employs the people? Who pays the taxes? Who, in the end, can be served with a legal notice and compelled to appear? Decentralization, in the abstract, is a spectrum. But jurisdiction is not a spectrum. Jurisdiction is a switch. Either a regulator can reach you or it cannot. And the way a regulator reaches you is through the parts of your operation that have a body. This is why the details that the FT report does not give us are the details that actually decide the case. Let me list them, because I think the absence of these details is itself the most important information in the story. First: the validator set. Hyperliquid's chain is secured by validators. If the validator set is genuinely permissionless — if anyone can join, if no single entity controls a majority, if the set is large and geographically dispersed and anonymous to one another — then the chain is, in a meaningful technical sense, decentralized. But if the validator set is small, if it is permissioned, if the founding team controls enough stake to determine outcomes, then the chain is a centralized system wearing a decentralized costume. And here is the part that matters for Singapore: a regulator that can identify the validators can pressure the validators. A regulator that cannot identify them cannot. So the anonymity of the validator set is not a technical detail. It is a legal firewall, or the absence of one. Second: the admin keys. Almost every protocol of this scale retains some mechanism for emergency intervention — a multisig that can pause contracts, upgrade logic, or recover funds in the event of an exploit. This is not a sin; it is prudence. But an admin key is a control point, and a control point is a handle. If a small group of people holds a multisig that can alter the core contracts of the platform, then those people are, in the eyes of a regulator, the platform. The protocol is not autonomous. It is administered. And administration is exactly the thing that jurisdiction attaches to. Third: the upgrade path. Can the core contracts be changed without a network-wide vote? Can the team push an upgrade unilaterally? If so, the protocol is not a law of nature. It is a product, and products have owners. Fourth: the interface. This is the one that the industry consistently underestimates. Even if the underlying protocol is perfectly decentralized, most users do not interact with the protocol. They interact with a website. They interact with a front end. And the front end is almost always operated by a company, hosted on a server, controlled by a domain name registered to a legal entity. Regulators around the world have figured this out. You do not have to shut down the protocol to shut down the business. You shut down the door that most people use to walk into it. Fifth: KYC and AML. Does the platform require identity verification? Does it restrict access by geography? Does it screen wallets against sanctions lists? If the answer to these questions is no, then the platform is, from the regulator's perspective, an unmonitored channel — and unmonitored channels are precisely what anti-money-laundering regimes exist to close. If the answer is yes, then the platform is admitting, implicitly, that it exercises control over who can use it, which is itself a strong indicator of centralization. This is the trap that decentralized platforms cannot easily escape. Either you know your customers, and you are centralized, or you do not, and you are a compliance risk. There is no third door, no matter how many whitepapers claim otherwise. I want to pause on that fifth point, because it connects to something I have written about before and something I believe deeply. There is a fundamental, irreducible opposition between the architecture of state surveillance and the architecture of cryptographic privacy. A central bank digital currency is, by design, a system in which every transaction is visible to the issuer. That is not a bug; it is the entire point. The issuer wants to see the flows, to control the flows, to program the flows. A public blockchain is the opposite: a system in which every transaction is visible to everyone, but no one can prevent it, reverse it, or program it at will. These are not two versions of the same thing. They are two different philosophies of money, and one of them ends in a world where the state can decide, in real time, whether you are permitted to spend. When I look at Singapore's approach to crypto derivatives, I see a jurisdiction trying to thread a needle: to keep the innovation, to keep the capital, to keep the talent, while keeping the surveillance of the parts of the system it can reach. Hyperliquid's position is, in effect, a claim that the parts Singapore can reach do not include the part that matters. That is a bold claim. And bold claims have a way of being tested. Now let me bring in a second thread, because I think it is underappreciated and it directly bears on what Hyperliquid is. The interest rate models that underpin most of decentralized finance are, to put it politely, arbitrary. I have spent years studying the mechanics of Aave and Compound, the two lending protocols that defined the genre, and the more I study them, the more convinced I am that their rate curves are not discovered from market forces. They are designed. Someone sat down and chose the shape of a curve — the utilization ratio, the slope, the kink point — and then declared that this curve represented supply and demand. It does not. It represents a guess about supply and demand, frozen into code, and updated only when governance decides to update it. Hyperliquid's perpetuals work on the same principle, dressed differently. The funding rate — the mechanism that keeps a perpetual contract tethered to the spot price — is not a market price in the way that a spot price is a market price. It is a constructed quantity. It is a formula. It tells traders how much one side of the book must pay the other, and it is calibrated, tuned, and occasionally adjusted by the people who built the system. This is not a criticism unique to Hyperliquid; it is a feature of the entire perp landscape, centralized and decentralized alike. But it matters for the regulatory story, because it shows that even the most "autonomous" protocol is full of human design decisions. The protocol is not a natural phenomenon. It is a machine built by people, with knobs that people can turn. And every knob is a potential point of regulatory leverage. I want to bring in one more technical thread, because it explains a choice Hyperliquid made that I find revealing. Most ambitious DeFi projects in recent years have chosen to build on layer-two rollups — chains like Arbitrum and Optimism and Base that inherit their security from Ethereum while offering cheaper, faster transactions. The economics of this choice were transformed by EIP-4844, the upgrade that introduced "blob" data and dramatically reduced the cost of posting transaction data back to Ethereum. For a while, it looked like the future of scaling was settled: everything would migrate to rollups, and rollups would get cheaper and cheaper as blob space expanded. I do not believe that. I have said for two years now that the blob data introduced by Dencun will saturate within roughly two years, and when it does, the cost of posting data will rise again, and rollup fees will double or worse. Blob space is a finite resource. Demand for it is growing faster than supply. The cheap-fee era of early 2024 was a window, not a permanent condition. And I suspect that Hyperliquid's founders understood this better than most. Instead of building on a rollup and inheriting its future cost curve, they built their own chain — a chain whose cost structure they control, whose throughput they can tune, whose consensus they can optimize for their specific use case. It is a more expensive, more complex, more centralizing choice. But it is also a choice that insulates them from the economics of someone else's chain. It is, in a way, the same instinct that drives their regulatory strategy: do not depend on infrastructure you do not control. That instinct — the drive to control one's own substrate — is what makes the Singapore question so interesting. Because the whole point of building your own chain is to remove points of dependence. And the whole point of having a Singapore entity is that you have not removed all of them. Let me now turn to the question of what "sufficient decentralization" actually means, because the phrase appears constantly in these debates and almost no one defines it. The concept traces back to a speech by a former official at the U.S. Securities and Exchange Commission, who suggested that a network could become so decentralized that it no longer resembled a security — that at some point, the efforts of a central promoter became immaterial to the network's success. The idea was elegant. It was also, from the beginning, an admission that decentralization is a matter of degree, and that someone — a regulator, ultimately — gets to decide when the degree is sufficient. This is the trap that I think the industry has walked into. We have spent a decade arguing that decentralization is a property of code. But in practice, decentralization is a judgment made by institutions about code. It is not something a protocol has. It is something a regulator grants. And grants can be revoked. The FT report gives us the regulator's alleged judgment in anonymous form. "People familiar with MAS's thinking believe the venue is not within its jurisdiction." I want to be very precise about what this sentence does and does not say, because precision is the only thing that protects us from wishful thinking. It does not say the MAS has ruled. It does not say the MAS has issued a guidance, a circular, a no-action letter, or any document with legal force. It says that unnamed people believe unnamed officials hold an unnamed view. That is not a regulatory position. That is a rumor with a good pedigree. And a rumor with a good pedigree is exactly the kind of thing that can evaporate overnight. Regulatory positions in Singapore, as elsewhere, are not contracts. They are policies. Policies change with leadership, with elections, with the discovery of a scandal, with a shift in the global mood. If the MAS has privately signaled tolerance, that signal carries no legal weight. If a future MAS decides that decentralized venues with Singapore entities are, in fact, within its jurisdiction, there is nothing in the current arrangement that would prevent it from saying so — and no document the platform could wave that would make the old tolerance binding. This is why I keep coming back to the phrase "decentralized venue." It is not a legal category. It is an aspiration dressed as a category. And the gap between the aspiration and the legal reality is precisely where the risk lives. Let me also address the elephant that the FT report conspicuously leaves in the room: the token. Hyperliquid has, at various points, been associated with a token and a points program and an airdrop. The report does not mention any of this. That omission could mean many things. It could mean the reporting simply focused on the regulatory question. It could mean the news cycle had moved past the token. Or it could mean that the outlet was being careful — that when you are writing about a regulatory question, you avoid entangling it with a speculative asset whose status is itself contested. I do not know which. But I know that if a token exists, then the regulatory question is not only about whether the venue can be licensed. It is about whether the token is a security, whether the token is a derivative, and whether the people who bought it have any recourse when the regulatory weather changes. And those are questions that no amount of decentralized architecture can make disappear. CONTRARIAN Here is where I want to push against the prevailing interpretation, because I think most of the commentary I have seen gets the story backwards. The consensus reading is that Hyperliquid has achieved something clever: it has found a jurisdiction that will let it operate, and it has structured itself so that the jurisdiction cannot touch it. The consensus reading is that this is a win for decentralization — proof that a sufficiently decentralized protocol can escape the reach of any single regulator. The consensus reading treats the Singapore entity as a strategic asset and the anonymous regulatory signal as a soft endorsement. I think that reading is dangerously optimistic, and I want to explain why. The first problem is that the anonymous signal is the opposite of an endorsement. An endorsement is public. An endorsement is on the record. An endorsement can be cited, relied upon, built into a business plan. What Hyperliquid allegedly has is not an endorsement. It is a rumor that a regulator might, for now, choose not to act. And a rumor that a regulator might not act is functionally identical to a rumor that a regulator might act. Both describe the same underlying state of affairs: uncertainty. The only difference is which direction the uncertainty happens to point in this particular news cycle. The second problem is that the existence of a Singapore entity is not a neutral fact. It is a liability. If you have no presence in a jurisdiction, the jurisdiction has nothing to grab. If you have a presence, you have given the jurisdiction a handle. The whole point of the offshore structure that most DeFi projects adopt — the foundation in Switzerland, the entity in the Cayman Islands, the anonymous team scattered across the world — is to minimize the number of handles. Hyperliquid has, by having a Singapore entity, created a handle. And it has done so in a jurisdiction that is not a passive offshore haven but an active, competent, and increasingly assertive regulator. That is a strange choice for a project whose entire defense rests on being unreachable. The third problem is deeper, and it is the one I most want to press on. The claim "we are decentralized, therefore we are not subject to jurisdiction" assumes that regulators accept the framing. They do not. Regulators have spent years developing the tools to pierce the decentralization narrative, and those tools are getting sharper. The European Union's Markets in Crypto-Assets regulation, the various enforcement actions in the United States, the tightening of derivatives rules across Asia — all of them converge on the same principle: the label a project applies to itself does not determine its regulatory status. What determines its regulatory status is what it actually does, who actually controls it, and who actually profits from it. So the question I would put to anyone celebrating the Hyperliquid news is this: if the MAS decides next year that a Singapore entity operating a derivatives venue is within its jurisdiction regardless of the venue's architecture, what is Hyperliquid's plan? The answer, as far as the available information goes, is that there is no public plan. There is no public legal opinion, no public compliance roadmap, no public statement of what the platform would do if the regulatory weather turned. There is only the architecture, and the hope that the architecture is enough. And architecture, by itself, is never enough. I have watched too many projects learn this the hard way. The protocol is not the company. But the company can be regulated, and when the company is regulated, the protocol — however decentralized — discovers that it still needs someone to maintain the front end, to pay the developers, to handle the customer support, to run the marketing, to do all the hundred mundane human things that no smart contract does. And every one of those mundane human things is a point of contact. Every point of contact is a lever. And every lever is a way for a regulator to apply pressure without ever having to answer the philosophical question of whether the protocol is decentralized. This is the blind spot I want to name. The industry has spent a decade building technical decentralization and calling it regulatory insulation. But technical decentralization and regulatory insulation are not the same thing. You can have a perfectly decentralized protocol that is still, in practice, controlled by a small team of people who can be found, sued, subpoenaed, and pressured. And you can have a fairly centralized protocol that is nonetheless insulated, because its operators are genuinely unreachable. What matters is not the degree of decentralization in the code. What matters is the degree of reachability of the people. And reachability is a function of legal structure, physical presence, and identity — not of consensus mechanisms. I think this is the insight that the Hyperliquid story, properly read, actually offers us. It is not a story about a protocol that escaped regulation. It is a story about a protocol that has, for now, arranged its affairs so that regulation is not being applied. Those are different things. The first is a permanent condition. The second is a temporary state of affairs that lasts exactly as long as everyone involved prefers to maintain it. And the moment the political or economic calculus changes — the moment a consumer loses money in a way that becomes a scandal, or a jurisdiction decides that hosting a derivatives venue without a license is an embarrassment it can no longer tolerate — the temporary state of affairs ends. I have said before that trust is built in the bear and sold in the bull, and I think the inverse is also true: regulatory tolerance is granted in the calm and withdrawn in the storm. Right now, the market is calm relative to the mania of 2021. The stories that would generate outrage are not being told. The retail losses are not front-page news. But bear markets end. Bubbles return. And when they do, the venues that were tolerated in the quiet will be scrutinized in the noise. Hyperliquid's Singapore position is not a fortress. It is a bet — a bet that the quiet will last. And bets, by definition, can be lost. Let me also say something about what I think this means for the broader DeFi derivatives sector, because the contagion runs in both directions. If Singapore's alleged tolerance is real, and if it becomes known, then other derivatives platforms will look at Singapore and see an opportunity. They will incorporate entities there. They will hire there. They will build the same architecture and hope for the same result. And the more platforms that do this, the more pressure there will be on the MAS to clarify its position — because a regulator cannot indefinitely maintain a policy of studied ambiguity when a growing share of a major industry is structured around it. The more successful the strategy becomes, the more likely it is to be shut down. That is the paradox at the heart of regulatory arbitrage: it works best when it is rare, and it destroys itself by succeeding. TAKEAWAY So where does this leave us? I want to end not with a prediction but with a question, because I think the question is more useful than any answer I could offer. The question is this: what does it mean to build a financial system that cannot be captured? For a decade, the answer has been technical — build it on a blockchain, make it permissionless, remove the trusted intermediary. But the Hyperliquid story suggests that technical capture-resistance is only half the problem. The other half is legal and human. You can build a protocol that no one controls, and then hire a team in a jurisdiction that regulates what the protocol does, and discover that the protocol's capture-resistance is irrelevant to your own reachability. The code is free. The people are not. I think the next decade of this industry will be defined by how it answers that question. Not by how decentralized we can make the software, but by how honestly we can confront the fact that software does not act in a vacuum. There are always people. There are always jurisdictions. There are always handles. The mature response is not to pretend the handles do not exist. It is to design for them — to build legal structures that are as deliberate as the technical ones, to disclose what can be disclosed, to prepare for the storm while the weather is fair. I do not know how the Hyperliquid story ends. I do not know whether the anonymous tolerance will hold, whether the Singapore entity will become a liability, whether the protocol's architecture will prove sufficient or insufficient when it is finally tested. What I do know is that the way we talk about this story matters. If we tell it as a triumph — as proof that decentralization defeats regulation — we will learn the wrong lesson, and we will build the next platform on the same fragile foundation. If we tell it honestly — as a case study in the difference between being decentralized and being unreachable — we might actually learn something that helps us build better. The seeds we planted in the ashes of 2022 are growing now, in the cold, in the quiet. Some of them will become forests. Some of them will turn out to have been annuals, flowering once and dying. The only way to know which is which is to keep watching, keep testing, and refuse to mistake a temporary equilibrium for a permanent truth. Stay jagged. Stay honest. And never confuse the protocol with the people who built it, because it is the people, in the end, who will be asked to answer for what the protocol does.