Jeff Yan's “Hyper Life”: Demystifying the underlying logic of the world's most profitable company per capita

This article is from:Colossos; Original author: Dom Cooke (@domcooke）

Compiled by Odaily Planet Daily (@OdailyChina); Translator | Azuma (@azuma_eth）

Original title: Jeff Yan's “Hyper Life”

A 43-year-old man was taken from his home in Saint-Léger-sous-Cholet in western France one Friday in January, before it was bright. He was driven to the small town of Basse-Goulaine about 30 miles away, where he was beaten, tied up, and abandoned. Twelve hours later, as night fell on the outskirts of Paris, three men with only one pistol broke into a house in Verneuil-sur-Seine. They beat a couple in front of the child, tied up all four members of the family with tape, toured the whole house, and left for the train station.

This is the 70th similar attack worldwide in less than a year. Two days later, I boarded a flight to Singapore.

I was on this trip to visit a team of only 11 people, but the first person I saw in their office wasn't part of this team. He is a sturdy American with short hair and stubble, sitting at a small table in the corner of the rest area with an Apple laptop in front of him. His physique showed that he wasn't here to write code, but rather a bodyguard.

One of the company's co-founders (screen name iliensinc, short for Aliens Incorporated) accompanied me all the way from my hotel to my office. As we walked through streets shaded by rain trees, she told me they hadn't been in this part of Singapore all the time. The company was initially based in a shared office space in the financial district, but her other co-founder — the only person on the team who didn't use a pseudonym — began to attract attention. At first it was just some attention, and some people tried to recognize his face; then, strangers took the initiative to chat; later, someone followed him into the elevator of his apartment. As a result, the company moved to a quieter location, a building where no one would have thought of looking for them.

Even the company's cleaning didn't know their real business. In her opinion, this is a peripheral goods company that makes stuffed cats. There are actually 34 stuffed toys in the office, so this misunderstanding isn't hard to understand. The company's mascot is a cat named Hypurr, of which 12 are on the cabinet, but in addition, there are sharks, lizards, koalas, penguins, and dragons, some of which are attached to the display like furry gargoyles. Most of the dolls came from an engineer — his wife didn't allow him to bring any more dolls home, so he brought them to his office. The team didn't correct this misunderstanding on the part of the cleaning staff.

This is because Hyperliquid — a cryptocurrency on-chain trading platform — is one of the most profitable companies in the world per capita. The company's only 11 employees generated more than $900 million in profits last year. The company was founded only three years ago, has reached a market capitalization of $10 billion, and has never received a dollar in venture capital. The central figure behind it, 31-year-old Jeffrey Yan, has become an increasingly prominent face in an industry where success often means a higher risk of being kidnapped when not entirely voluntary.

Before starting Hyperliquid, Jeff lived in Puerto Rico and almost independently operated one of the largest anonymous trading teams in the cryptocurrency industry. The team is called Chameleon Trading — “Chameleon” was his nickname when playing games in middle school. He started with his savings of $10,000 and achieved annual growth of thousands of percentage points over two and a half years. When he mentioned his benefits to me, he immediately tried to convince me not to think it was amazing. I've noticed his objections, and I've also noticed that Chameleon has made him extremely rich. He was 27 at the time and financially free. In the eyes of San Juan surfers, bar bartenders, and waitresses, he's just an ordinary young man in surf shorts.

Now, in a heavily guarded office in Singapore, Jeff is sitting barefoot in a grey armchair, wearing black shorts and a dark blue t-shirt, explaining to me why the entire financial system needs to be rebuilt from scratch. The question I really want to know is, why did he use the first kind of life in exchange for his second life now?

“It's not about the money,” Jeff said. Jeff didn't come from a wealthy family, and he doesn't see any interest in “rich people's lives” in his life today. He wears the same Lululemon shorts and t-shirts every day—15 shorts and 10 t-shirts, each in three colors. Looking around his office, there was almost no trace of his wealth. The furniture was all left by the previous tenant, and the only things the team added were two sets of board games in the lounge area, NFTs on the wall, and those stuffed cats.

This was confirmed once again when I saw the four books on the shelf. One of them was Frank Slootman's “Amp It Up,” a management book whose core idea is that most people don't work hard enough. I mentioned this book to iliensinc, and she just shrugged it off — the idea of “work hard” was their own and didn't come from this book. So are the three bottles of Grey Goose and Macallan in the kitchen, which have remained untouched since a community event where minimum consumption was not met two years ago. This team is more accustomed to drinking tea.

Nor is it out of love for the cryptocurrency industry. Bitcoin, the industry's trendsetter, has fallen sharply after its high in early October; while gold, which was supposed to be “replaced” by it, rose 7% over the same period; most other coins performed worse. When I asked Jeff what he thought of the negative sentiment surrounding the industry as a whole, he didn't defend it. “There's really a lot of unclean behavior in this field,” he said. “People are gradually realising that this stuff isn't what they claim, which is probably a good thing to some extent.”

Jeff also doesn't consider Hyperliquid a “cryptocurrency company.” “Nobody calls the company an 'internet company' anymore,” he said. “We use cryptography, but that doesn't define us.”

Prior to joining Hyperliquid, only two people on the team (Jeff included) had experience in the cryptocurrency industry. To a certain extent, this was intentional. According to him, most early cryptocurrency practitioners were more concerned with making money quickly; what he built was a long-term project, so it was more suitable for those who think closer to technical people than traders. But it's also a supply issue. Hyperliquid's recruiters often come from the podium of the International Mathematical Olympiad. Jeff won a gold medal in physics at age 18, some of his engineers won silver medals in informatics, and others trained for the US national team. Jeff hopes to recruit more people like this — in fact, he has added 2 new members since I visited earlier this year — but the pool of talents willing to join the cryptocurrency industry and that level has been shrinking due to years of scams, loss of trust, and recent attraction in the field of artificial intelligence.

So why does Jeff, who has already earned enough wealth to do whatever he wants, stay here?

At least in the eyes of the outside world, the answer is becoming increasingly clear.

Hyperliquid is essentially a blockchain with its own exchange built on top of it. In traditional exchanges, the company controls your capital and controls the infrastructure; on Hyperliquid, users control their own assets, and the platform is public. Jeff's vision for it — he said it without being ironic — was to host the entire financial system. You can think of this as ambitious, or absurd, depending on whether you're looking at those plush cats or the platform's data. Because in the few months after my visit, some trading markets that have continued for hundreds of years have begun to deflect in a small but quantifiable way.

Hyperliquid started with perpetual contracts in 2023, a derivative and the largest market in the crypto industry. A so-called “perpetual contract” is a bet on the price of an asset. The trader does not actually hold the asset. Unlike traditional futures, it has no expiration date. The market size of this type of trading is 6 to 8 times that of the spot trading market, with a monthly transaction volume of about 7 trillion US dollars. Until recently, almost all of them were carried by centralized exchanges. The largest of these is Binance, which is far ahead. Previously, no decentralized platform had a real impact on it, and Hyperliquid was the first to do this, and its market share once grew to around 14% of Binance.

Then, in October 2025, Hyperliquid did something that centralized exchanges couldn't: it allowed anyone on the platform to create a new perpetual market for any data asset with a price predictor. An independent team called Trade.xyz became the most active builder among them. It was first launched on the silver market, and by January of the following year, its 24-hour trading volume had reached about 2% of the corresponding market of the Chicago Mercantile Exchange (Chicago Mercantile Exchange, founded in 1898, the world's largest derivatives exchange). Trade.xyz then launched the crude oil market. Crude oil trading has long relied on traditional markets that are closed on weekends, but on a Saturday in late February, the US and Israel began air raids on Iran. CME was shut down at the time, and Hyperliquid didn't shut down. The daily trading volume of crude oil jumped from $21 million to $3.7 billion. A month later, Trade.xyz launched the S&P 500 perpetual contract, which was officially authorized by S&P Dow Jones Indices and enabled 7×24 hour trading, including weekends.

Today, the most influential products on Hyperliquid are being built by developers who neither work for Jeff nor will ever work for him.

The founder of Trade.xyz (anonymity requested) bought his first Bitcoin in 2013 for $66 and has been active as an investor rather than a builder ever since. He had no intention of starting a business. He told me that if it wasn't for Jeff, he would probably have left the crypto industry a long time ago. “Hyperliquid has an opportunity to save the crypto industry,” he said.

However, all of this still doesn't fully explain why Hyperliquid is actually likely to become what Jeff envisioned — especially in an industry that “seems about to succeed and then collapse” — nor why he would give up his life in Puerto Rico to pursue it all. With these questions in mind, the first afternoon I arrived at the office and sat in the lounge area talking with iliensinc. There was a stuffed cat on the table, and the air was still filled with the smell of ginger and sesame left over from lunch. She told me that the team had asked Jeff the same question three years ago when he announced the shutdown of Chameleon. Her answer didn't start with the cryptocurrency industry, but rather with Jeff himself.

“You should ask his mother,” she said.

Jeff preferred to schedule meetings outdoors. We sat on a covered patio with four grey lounge chairs and a coffee table. Cars pass by on the downstairs street from time to time. Every few minutes, a gardener starts the lawnmower. The warning sounds of pedestrian crossings are pouring in one after another.

Jeff sat with his feet tucked under his body. When I asked about his mother, he was silent for a moment. He said that she used to say a phrase — a Chinese idiom: “There are people outside of the world, there are heavens outside the sky.” The idea is that no matter how good you think you are, there are always stronger people and more unknown worlds out there. She's not the kind of mother who strongly pushes her child, but she wants him to understand that what she sees is only a small part of the world.

Raised alone by his mother, he and his sister live in Redwood Shores, located between San Francisco and Palo Alto, in the center of one of the most valuable geographic regions in American business history. Oracle's headquarters building with a mirrored glass façade stands above the community. Most of the neighbors are engineers and product managers. Their children were raised from an early age to move forward towards the kind of life path Jeff later took. Jeff's parents are both immigrants from China and divorced when he was in the third grade. My father left the family, and my mother is an accountant, and she has to work overtime every tax season. Jeff looked at everything. “I can feel that others are richer than us,” he said, “but I've never been resentful about it. Hanging out doesn't have to cost a lot of money.”

His school did not have a strong academic competition atmosphere. Although her mother used to say that idiom, she didn't put pressure on Jeff. No one really “forced” him to do anything until he reached puberty. He hangs out every day, goes to school, comes home, and continues to play. By the standards of his zip code region, he is an extremely rare being—a “foster” child.

When he was in eighth grade, a friend who had just transferred from private school took him to a math competition — that friend just wanted someone to be with. Jeff had never seen anything like it; math in school was completely different. There's no formula to memorize, and no complicated calculation process. You'll get a question, sometimes just a sentence, and then find your own path to the problem. The answer isn't a number; it's a piece of proof — a complete argument for why something must be true. Finally, they'll rank the contestants like a sprint race. For Jeff, it was an experience that perfectly blended “the joy of sports” and “the joy of understanding the world.”

That summer, he woke up at 5 a.m. every morning, downloaded competition questions from the internet, and studied alone in his room. He doesn't have a mentor, and he can't afford any summer training programs, and no one asks him to do that. “I only discovered later that I was actually really good at winning,” he said. “It's like a game I didn't even know existed before, and the other kids had been preparing for the rest of my life, and I was left behind.”

A year after starting to compete, when he was in the 9th grade, he was selected for the American Math Olympiad training camp, which brought together the top 50 high school students from all over the country, and he was one of the youngest members. He didn't join the national team — he said he didn't care. During those three weeks, he sat with a group of teenagers who could stare at the three sentences for five hours and unearth truths invisible to most people. Jeff told me that there are no “superstars” like Roger Federer in mathematics, but at the highest level, there does exist a realm like Federer represents: one style, one elegance — reflected in proven construction methods. And during training camp, he saw this up close for the first time. “It's like being able to play with Tom Brady,” he said, “just a more nerdy version of the experience. Most people don't feel this way.”

The following year, he lost in an intermediate selection round of the math competition. He was 16 at the time and had to wait a full year to try again. I asked him if this was the first time he experienced failure. “Failure is actually a very common experience,” he said. “Most people are losers. There's usually only one winner.”

The problem is not the failure itself, but the feeling of emptiness. “It's like there's a hole,” he said. “I should learn something.” As a result, he found physics textbooks for senior students. Although his school doesn't offer this course until his senior year of high school, he has just learned calculus and understood its uses for the first time. He discovered Feynman's lecture. “I've 'brushed' them up like watching TV shows,” he said. In less than a year, completely self-taught, he became one of the top five young physicists in the country.

He was selected for the US Physics Olympiad national team, traveled to Estonia — his first time in Europe — and received a silver medal. The following summer, in Copenhagen, he won a gold medal, ranking 24th in the world. He was 18 that year, and when he came back to the Bay Area, he already understood the true meaning of his mother's words: there really are people above him, and there are at least 23 of them.

Harvard covered all of his tuition fees. In the spring semester of his freshman year, Jeff took Computer Science 124 — Data Structures and Algorithms. This course is usually taken by sophomores and juniors, and is known for being “painful.” In Harvard's course reviews, some students called it a “necessary evil.” One comment even wrote, “Without a social life, you wouldn't be in a relationship.” The course had a total of 150 students, and as a freshman, Jeff finished first and had a clear advantage.

At Harvard, students are assigned to senior year dormitories after their freshman year. Jeff was assigned to Pforzheimer House, where he befriended Scott Wu, who was two years younger than himself. The two first met at a summer camp for Olympians. Wu represented the United States in the International Informatics Olympiad for three consecutive years, winning a perfect score at the last time, and later co-founded Cognition AI. When Wu was assigned to Pforzheimer in his sophomore year, he messaged Jeff: “Yo, I'm in Pfoho.” Jeff replied, “Let's go!”

Wu often finds Jeff in front of the grand piano in the common room — where he taught himself jazz and practiced over and over until he fully mastered it. They played chess, Go, and poker together, and spent a lot of time discussing what it meant to “be the ultimate in a field.” Jeff talks about Faker, the greatest player in League of Legends history, as well as top Go players and the best high-frequency traders. “He's always thinking about what makes a person special,” Wu said to me. “What is the nature of this field? And what does it mean to be really top?”

Wu remembers that Jeff's way of thinking was unusually “reverse.” At Harvard, most students tend to come to similar conclusions with similar information and environmental influences, but Jeff never did. Additionally, Wu mentioned that he has a great sense of humour. “That kind of expressionless humor. He says things that are completely unexpected, yet in an extremely bland tone.”

Jeff has been working throughout the summer. He interned at Google X to develop tools for an autonomous driving project that later became Waymo. He also interned at trading firm Tower Research Capital. During his senior year, he worked part-time at Nuro, another autonomous driving company, largely because he felt that four years of college was at least one more year.

In the winter of his junior year, he joined Wu in Hudson River Trading's first internship program. A total of 10 interns were selected at the time. Hudson River Trading is one of the world's most successful quantitative trading companies. Among these 10 people are Alexandr Wang, who later founded Scale AI, and Jesse Zhang, who founded Decagon. The internship took place in the form of a three-week competition, and Wu and Jeff took first and second place in each round.

Jeff graduated with a bachelor's degree in mathematics and a master's degree in computer science, and joined Hudson River Trading full time in late 2017. He was assigned to the US Stock Algorithms team. Every week, he has one-on-one meetings with his manager. The manager has brought in quite a few newcomers. This type of meeting usually has a fixed rhythm: the newcomer encounters a bottleneck in the code, the two solve it together, and then the newcomer goes back and encounters the next bottleneck. But Jeff doesn't “hit the wall,” his manager recalls that he often brings new ideas. The meeting was efficient and smooth, but the manager always felt something was wrong. It took him a while to realise the problem: Jeff did everything very well, but these things didn't seem important to him. The manager wasn't surprised when Jeff proposed leaving the job after eight months. The email he sent about his resignation was an extremely warm letter within the company.

Jeff loves Hudson River Trading. He believes trading is the purest “game” in the real world — you're either right or wrong, and the market will tell you the answer. The smartest group in the world competed against each other, and in this brutal game, they created an extremely important product for the world: a liquid and efficient market. But he realized that it took him eight months just to optimize an already excellent system, and that the system would work very well even without him. This left him unable to answer the lingering question — what value have you created for the world?

This question seemed to have found its own answer in December 2017. Bitcoin was close to $20,000 at the time, and Coinbase became the most downloaded app in the US, and billions of dollars poured into various ICO projects, such as Jesus Coin — the peak of “Crypto Christmas.” Jeff first heard about Bitcoin during an internship at Hudson River Trading when two former partners introduced the concept to the interns, but it didn't resonate. However, while working at Hudson River Trading, he read the Ethereum Yellow Book, which describes a computer “recognized by the world and which no one can turn off”. He is exposed to the financial system every day, and he knows the fundamentals of its operation, and this paper describes a way to replace trust with code. “I feel like I can build something and completely reshape finance.”

He left Hudson River Trading around April 2018 and set out to create a prediction market where users could bet on any event—such as weather, elections, or sporting events. The system will run on the blockchain, and no single entity will control the funds. Its architecture is based on an idea he and his co-founder pioneered: off-chain matchmaking, on-chain settlement — because Ethereum's speed is far insufficient to support a real exchange. The funds will be stored in smart contracts controlled by code, but the user experience will remain fast and smooth — preserving the promise of decentralization of cryptography while avoiding its frictional costs. He and his college roommate Brian Wong (who also left Hudson River Trading) co-built this product in the first Binance Labs incubation project in San Francisco, and named it Deaux.

Kalshi was founded in 2019 with a similar idea, and Polymarket followed suit in 2020. Today, the combined valuation of the two companies is over $40 billion. And Deaux only has 100 users.

When Jeff spoke about this, it suddenly rained heavily in the Singapore sky. That kind of heavy, dense raindrop can fill a gutter within a few minutes. From the terrace, we could hear the rain hitting the street violently, and the wheels glide through the accumulated water, making an even louder hiss.

“There's no way this project will be successful,” he continued. By the time Deux went live, Bitcoin had dropped more than 80% from its high point; Jesus Coin had long since fallen silent, and there was no “resurrection.” No one wants to bet on tomorrow's weather. More importantly, Jeff and Wong have hardly given serious consideration to regulatory issues. Kalshi fought for a full three years to obtain regulatory approval before launching the product.

Scott Wu was one of the few people in the world who really regretted it when Deaux shut down — he was one of only five stable users. Jeff refunded more than half of his $450,000 investment. He was still subject to competition restrictions by Hudson River Trading, so he went skiing in Lake Tahoe, California with a friend who was also competitive, until the end of the snow season. He then traveled to China, Japan, and Peru on a limited budget. He tried to convince me that being a tourist actually requires quite a bit of “skill” — and he doesn't have that ability.

In late 2019, when competition restrictions expired, Jeff moved to Puerto Rico—where people could legally cut capital gains taxes to close to zero. With only $10,000 in his hands, he had a vague sense that some kind of huge opportunity was approaching.

His partner moved with him to Puerto Rico. They shared a one-bedroom apartment by the sea for less than $2,000 a month. But “sharing a lease” meant some kind of living together, and Jeff didn't spare time for it. He didn't even have a monitor, so he just took up the TV in the living room and placed his work environment there. For the first year or so, he only gave his partner about 30 minutes a day, and the rest of the time was all part of the trading algorithm scrolling on TV.

Jeff works at least 14 hours a day, easily reaching 100 hours a week. Starting with a Python script, he wrote code to connect to major cryptocurrency exchanges and let the program trade automatically around the clock. He continuously monitors these systems, optimizes logic, tracks data, and once the results don't meet expectations, he completely reverses them.

The reason he was able to do this is because of the openness of the cryptocurrency market, something traditional finance has never had. In the stock market (such as the one he traded at Hudson River Trading), placing an order on a single exchange requires connecting to 13 “live trading exchanges” located in New Jersey's top three shared computer rooms, complying with complex US Securities Regulatory Commission (Reg NMS) rules (Reg NMS), obtaining futures data from the Chicago Chicago Mercantile Exchange through microwave links, and investing tens of millions of dollars in infrastructure costs. And in the cryptocurrency market, whether it's a Hudson River Trading employee or an individual working on TV, they all have access to the same basic HTTP infrastructure — which was originally designed for web development. You only need to rent a server on Amazon Web Services.

For almost two years, his partner had no idea what was actually happening on the other side of the TV. Their lives don't seem to have changed: same rent, same diet. She knows he's very engaged and motivated, and thinks he “should have done a pretty good job,” but there's no realistic evidence of his success. Until one Friday night in the summer of 2021, she tried to pull him out of her house to make a dinner reservation a week in advance — but he refused to leave.

“You don't understand,” he said to her, “if I don't fix this bug now, I'll lose $100,000.”

After that night, Jeff decided to make this a real business. He needs “someone who can do anything but write code.” While at Harvard, he met someone in his Pforzheimer dorm room — she seemed to be able to keep everything in her life in order at the same time, which was almost an “strange ability” for him. But he finally heard that iliensinc was in Asia at the time, working as chief of staff at a VC company, traveling back and forth between Tokyo, Seoul, and Hong Kong.

When he contacted her, it was discovered that she had returned to San Francisco. The pandemic brought cross-border travel to a standstill. This job, which originally required traveling all over Asia, turned her into a late-night phone call from her apartment. Jeff described his needs to her. He didn't provide a clear job description, title, or even little explanation of what she wanted to do. But she's been evaluating entrepreneurs for the past three years, and regardless of what Jeff is saying, she intuitively thinks this is someone worth “betting against.”

The company officially has a name — Chameleon Trading, and iliensinc began delegating teams to hold Zoom meetings with the business development teams of major exchanges, adding a layer of expertise to the organization, which in reality is just “a man living by the San Juan coast.” Beneath those giant market makers — such as Jump Trading, Tower Research Capital, Hudson River Trading, and Jane Street — there is also a layer of anonymous trading institutions whose size is difficult to be accurately measured by the outside world, and Chameleon is one of the sizeable ones.

By 2022, Jeff began to feel uneasy. He has been in the crypto industry for four years, participated in various markets — whether centralized or decentralized — and gradually began to focus on the field itself, not just his own profits and losses. Bitcoin provides the world with a way to hold and transfer funds without trusting intermediaries; Ethereum provides a “global computer” that no one can shut down. In between, almost all the foundations needed to rebuild the entire financial system have been outlined. However, the industry has done almost no substantial construction on top of this. The two largest exchanges — Binance and Coinbase — are still centralized. The crypto industry is constantly reintroducing what it was originally trying to eliminate.

That summer, iliensinc organized an offline team event at a hotel in rural England. At the time, she had already expanded Chameleon into a six-person team. Jeff gave her a budget — one bitcoin. The team flew to London, visited the British Museum, then spent a few days at a country estate. And their “leader”, away from the screen for the first time in memory, didn't seem entirely comfortable.

Once back in Puerto Rico, the deal continued, but Jeff told the team they were going to start building something brand new. He wasn't sure exactly what. He had a few ideas, but none really convinced him. The only thing he knows for sure is that Satoshi Nakamoto's initial vision for Bitcoin is quietly being buried in the industry. This incident made him uneasy, even to the extent that a “person who has already made millions of dollars from it” should have been.

To the team, Jeff seemed to be “breathing too much fresh air.”

In November 2022, FTX, the world's third-largest cryptocurrency exchange and valued at $32 billion, collapsed in just nine days. It lent funds deposited by users to Alameda Research, a trading company run by the founder's girlfriend. When users requested to withdraw their deposits, the funds were long gone. Less than six months ago, Terra, the $50 billion cryptocurrency ecosystem, also zeroed out in three days. It tried to build a stablecoin pegged to the US dollar, but what supported this system was only the system's own algorithmic logic, which was meant to maintain an anchored algorithm, which instead accelerated its collapse. Two of the biggest projects in the industry's history were destroyed one after another in less than a year.

Jeff has seen enough. He told the six-person team that they are no longer trading and that maybe everyone will disagree, but Chameleon ended it. If he makes a mistake in his judgment, he can still go back to trading in the future. It was true that there were objections on the team, and others chose to leave, but that didn't change Jeff's decision. There was no investor to consult, and no board to persuade — it was his own money, he made his own decisions, and now he has a new goal.

“I was too confident at the time and thought FTX would be the end of centralized exchanges,” Jeff told me, “but it also helped because it made me determined to challenge this huge market.”

The market he was talking about is a perpetual contract. This product stems from an insight developed by economist Robert Shiller in the 90s. Traditional futures contracts have an expiration date, and once they expire, traders either accept delivery of the underlying asset — such as oil, wheat, or pork — or close and re-open the position, paying a fee each time. Shiller asked the obvious question: if the vast majority of people who trade pork futures don't want pork at all, then why do they have to let the contract expire?

Traditional markets already have viable solutions, so there is no incentive to change. But in 2016, a cryptocurrency exchange called BitMEX first tried this model, and since then perpetual contracts have become the mainstream trading method in the crypto market. These contracts have no expiration date, and traders can use high leverage, usually up to 10 or 20 times the principal amount. The resulting fees and liquidation mechanism have made centralized crypto exchanges one of the most profitable enterprises in the industry.

By the end of 2022, no one had built a truly “usable” decentralized version. The reason is the underlying technology. In most modern markets, transactions are completed through an order book: the buyer indicates the price they are willing to pay, the seller indicates the price they are willing to accept, and when the two match, the transaction takes place. The more participants in the market, the smaller the bid-ask spread. From the New York Stock Exchange to Binance, this mechanism is generally followed, but the order book does more than match transactions; it also has to handle a continuous stream of updates — traders adjust quotes over and over, often multiple times before the actual deal is completed. However, existing blockchains are not good at handling these operations; they are too slow, too expensive, and too heavy in structure. Each update requires payment and is subject to confirmation. Running an order book on such a system is like trying to use dial-up internet to run the NYSE.

At the end of 2022, Jeff and his team looked at the chains all the other projects depended on, and none met their needs, so they decided to build their own. Within three months, Hyperliquid had a custom blockchain capable of supporting the operation of the exchange. Over the next year, Jeff was heavily active on Twitter, repeatedly explaining the benefits of Hyperliquid and why it's superior to established solutions in the industry.

But the trouble with an exchange is that until it's “useful,” it's completely “useless.” If a buyer enters an empty market, they cannot find a counterparty. The traditional solution is to pay a market maker so that anyone entering the market can find a counterparty — payment can be in the form of cash, equity, or token sharing. Indeed, several market makers have turned to Hyperliquid, and one of them even bluntly told iliensinc: “We are the ones who made kings. If you don't pay, you can't succeed.”

They didn't pay — neither to him nor to anyone. Hyperliquid went live at the end of February 2023, and in the following March and April, its users were mainly NFT collectors who had never been exposed to perpetual contracts. They tried it out with small transactions of $10 and learned the leverage mechanism through simulated trading contests. There are almost no “professional users” in the true sense of the word.

Then, in May, Jeff deployed the strategies that had made Chameleon one of the most successful anonymous trading teams in the crypto industry — an on-chain funding pool — HLP (Hyperliquidity Provider). Users can deposit $10 or $10 million. There are no processing fees, and no performance sharing. The pool runs automated trading strategies, and all profits are owned by investors. The entire account is completely public on the chain. If you deposit $10, you can see that $10 increase in real time. If FTX had been built this way back then, Alameda's funding holes would have been in full view of the world.

HLP solved two problems at the same time: exchanges gained liquidity, and users providing liquidity were exposed for the first time to opportunities never offered by traditional finance. An early Hyperliquid user described it this way — for the first time in history, ordinary people can invest in high-frequency trading strategies with zero fees.

“If I were to enter a product like this, I'd even pay Jeff a 2% management fee and 50% performance share,” he told me. “But now, an ordinary person with no background, no resources, and anywhere in the world can access one of the top market-making strategies in the crypto market. People still haven't realized how special this is.”

At the time, few people really understood this. By the fall, the price of crypto assets was rising every day, while users depositing in HLP watched their account balances continue to drop, in stark contrast to the rise in Bitcoin. The algorithm itself works well and continues to be profitable in trading, but since everything runs on the chain, it can't hedge the overall market exposure. Traditional market makers usually hedge risk in other trading venues, and HLP is unable to do this for design reasons. As a result, although every trade is profitable, it is essentially shorting a rising market. Users were outraged by this. Other projects attacked Hyperliquid on Twitter and Discord, and Jeff fought back — back then, he still took these things very personally.

But HLP is never the final answer. Jeff built it to “launch” liquidity before independent market makers enter, and he knows this is an obvious opportunity for market makers. Demand far exceeds supply, and the wide bid-ask price spread means that as long as someone is willing to make an offer, it is easy to make money. He wrote detailed documentation, posted a long article on Twitter explaining the market-making mechanism, and personally guided the agency to complete the access process. But most institutions remain wait-and-see — because almost every other exchange will pay them, and Jeff adamantly refuses to do so, and HLP itself cannot expand indefinitely to fill this gap. “Alameda is a critical component of FTX's operation,” he said, “and we don't want HLP to be the key to Hyperliquid's operation.”

Various metrics are rising, but so are complaints. Arguably, market makers should be able to enter the market at any time. But if they don't show up in time and users are lost first, then everything will end.

“You can always count on one type of person to show up” — a VC investor.

Their analysts have actually been using the exchange privately — quietly testing it in their own time — and then reporting to their partners one by one: “This stuff is really good.” As a result, these partners began to actively contact each other. Jeff and iliensinc are not doing any external expansion, nor are they preparing financing roadshow materials (pitch decks). The agreement itself is already generating fee revenue, but Jeff insisted from the beginning that this revenue won't go to the team. When the venture capital asked on the phone if there were any roadshow materials, Jeff and iliensinc simply had a conversation, and eventually the other party understood: there really were none.

By January 2024, the funders had already begun offline visits. iliensinc is very familiar with this process — she has worked as an investor. She began explaining various financing terms and rights arrangements to be aware of to Jeff. Jeff followed this pace for about two weeks. “It felt almost natural,” he said to me. “VCs are starting to get in touch, then it's probably time to raise money.”

The only condition he proposed was that only investment terms valued at $1 billion should be considered. At this point, it's less than a year since Hyperliquid went live, and the team was spending hundreds of thousands of dollars each month from Jeff's personal savings. Jeff spent the weekend thinking hard when an investor accepted this valuation.

He asked the entrepreneurs and these venture capitalists themselves for advice on what exactly is the point of financing. But no one could convince him that the money was worth “more than it itself.” At some point, he felt that refusing was the right choice. Once that feeling holds true, it's over.

“We won't accept this investment,” he told iliensinc on Monday morning.

“What the hell?” She couldn't believe it at all.

She is the one responsible for managing money and watching it continue to be consumed. Now, a fund is willing to provide an investment of about $100 million, but after she spent two weeks preparing to accept the financing, he chose to refuse it. It's just as difficult for the rest of the team to accept.

He called the fund and clearly declined. The other party was also unconvinced and thought he had just accepted the terms of another agency. But that's not the case. Hyperliquid isn't a company, it's an agreement, and its “neutrality” has been at the core from the beginning. “If Bitcoin had venture capital financing back then,” he said, “I don't think it would be the Bitcoin it is today, and its entire value proposition would be destroyed.” What's more, he's not short of money. To this day, Jeff still uses his own funds to cover much of the team's expenses.

On January 28, 2024, he posted a tweet with only four lines:

No investors (no investors).

No paid market makers (no paid market makers)

No fees to the development team (no fees go to the development team).

No insiders (no insiders).

Hyperliquid only has one meeting a day—morning meeting. I watched the process on my second day in Singapore. The team surrounded an engineer's screen with a dragon-shaped plush figure on it. They're testing a new feature called “portfolio margin” (portfolio margin). Much of the discussion revolves around potential risk, but for a long time, this wasn't even a “discussion.” Jeff would cross his arms, look down at his bare feet, and fall into contemplation; the engineer next to him would do the same. This silence was neither awkward nor short, and no one in the room thought it was anything unusual.

Part of this vibe stems from personality. The team members were between 24 and 31, and almost all were extremely intelligent introverts, but when I asked Jeff if he was a regular reader, his answers indicated that it wasn't just a question of character.

“I read far fewer books than the conventional wisdom thinks are 'optimal',” he said, smiling while wearing dark-framed glasses. “It actually takes a lot of time to actually read a book in a way that will shape you for a long time. Judging from the return on time, it's not a good deal.”

He moves his chin gently when he talks — I've grown accustomed to this habit, just like someone opening their mouth on an airplane to relieve ear pressure. One risk of writing about young techies is that sooner or later they'll tell you “I don't read very much.” So when Jeff added that he reads a book about every two months and is looking forward to one day in the future to systematically complete those unread books, I was rather relieved. But he immediately explained why now is not the right time.

“If you're not the first person to do something,” he said, “then it's probably not worth doing. I really think so. If you follow this logic, then reading doesn't really help much. Because once something has a ready-made experience and context, it's likely already been done. And since it's already been done, why would you do it?”

At the end of 2023, Hyperliquid had another problem — and the cryptocurrency industry already had a mature “routine” for this. As always, Jeff doesn't intend to copy it.

In crypto projects, tokens usually represent the holder's interest in the success of the project. How to distribute initial tokens is usually done through a “points plan” — the project announces that users can get points for their actions on the platform, and the user assumes that these points will be exchanged for tokens in the future. As a result, a large number of users poured in to try to get as many points as possible before redeeming.

The problem is that most of these “incoming users” aren't real users, but specialized teams. They will reverse analyze the rules, maximize profits through automated strategies, and leave the market quickly. Real users — that is, the people the project originally wanted to reward — only get the rest.

The Hyperliquid version went live on November 1, 2023. Users earn points every week based on their trading behavior, but the exact calculation formula has never been disclosed, and no one knows how it works. Every Friday, iliensinc will announce the points results. This gradually forms a fixed ceremony: users stare at her account “being entered” on Discord, then get together to compare points, share screenshots, and speculate on the rules. “It's important to reward real users,” Jeff said. “Although difficult to define, Hyperliquid's points program could reduce the 'brush' ratio from 99% to 20%.”

Meanwhile, market makers that Jeff previously refused to pay directly to attract began entering the market one after another. One of them — and one of the biggest market makers on Binance — has been wary of the new platform after the FTX incident. But through mutual contacts, they've heard quite a few positive comments about Jeff. The market maker first met Jeff and iliensinc at a conference in Singapore in September 2023. “Jeff is ambitious, but not arrogant,” he said. “He was very restrained in describing what he wanted to do, and it met expectations in every way.” After the meeting, he sent a message to the team: “We should join.” Two weeks later, they officially went live.

When this market maker was connected, what was discovered was consistent with the user experience: the infrastructure reflected “a design only traders would notice” in many details. Hyperliquid has a built-in mechanism similar to a “speed bump,” making it harder for the most aggressive quantitative agencies to “harvest” other market makers by rushing. This design was later widely replicated by the industry. The effect is that market makers can provide deeper liquidity without seeking extremely low latency. Jeff actually chose to sacrifice some of the exchange's trading volume (such as high-frequency trading volume from quantitative agencies hedging each other) in exchange for better prices for regular users — a trade-off that would reduce the platform's own revenue.

It was at the same meeting (Token2049) that Jeff and iliensinc decided to migrate the team. Jeff told me that the regulatory environment for cryptocurrency derivatives in the US is uncertain, and continuing construction there is an unnecessary risk. One lawyer even described that period as US regulators “used every means to try to keep this technology out of the country.” iliensinc visited Hong Kong, Switzerland, and Singapore, and ultimately chose Singapore—modern, safe, and without too many distractions.

By the spring of 2024, the team had completed the migration. This city-state is perfect for Jeff because it's “boring.” He has only two modes of life: work and exercise. He swims, runs, and does any sport that consumes physical strength without increasing the risk of injury — a principle stemming from his motorcycle accident in Puerto Rico that left a scar on his face and left him unable to work for a week. For him, the meaning of exercise is simply to empty the brain so that it can continue to build. His only “leisure” time is Sunday mornings, and the rest of the time belongs to Hyperliquid. He even cut his own hair because going to the barber shop was a waste of time.

He doesn't think it's unusual — or rather, he thinks most people have a too laid-back attitude towards work. “I think people are generally a bit too 'soft',” he said. “The brain is also an organ. If you need to work longer hours, you can train.”

However, he also learned not to impose this rhythm on the team. The team ate lunch together every day and sat around a black wooden table like a family dinner. They eat Chipotle every Thursday — Singapore doesn't have this restaurant, so they hand over the recipe to the chef, who reproduces it. The conversation at lunch usually revolves around something people have been watching or listening to recently. Jeff, on the other hand, often quiets down in moments like these, and seems like he's thinking about something else — and he's probably really thinking about something else.

Also in that spring, Hyperliquid's perpetual contract daily trading volume had surpassed $1 billion, and infrastructure began to come under pressure. One afternoon, the alarm system kept triggering, and the platform couldn't handle the sudden influx of users — this was the first time Hyperliquid went down. But outside of the office, everyone's only concern is the upcoming Hyperliquid token.

In May, Jeff posted a roadmap for the next six months on Twitter, which was full of technical plans without a single mention of tokens.

Prior to that, Hyperliquid had expanded from derivatives to spot trading. The first token to go live was Purr — named after that cat. Spot trading is a necessary step: if they want to issue Hyperliquid's tokens, the team needs an exchange market, but this also poses a problem they haven't faced before — in perpetual contracts, no one needs to hold the underlying asset; in spot trading, someone must assume asset custody. This is exactly what Jeff doesn't want to do — the core of the entire design is that users control their own assets.

To solve this problem without becoming a custodian, he realized he had to change his mindset: stop thinking of Hyperliquid as an “exchange running on a blockchain,” but rather as “a blockchain with a built-in exchange.” This blockchain, built by a team and already capable of processing hundreds of thousands of orders per second, can be further made programmable — an open system that allows anyone to write code and build financial applications on it, just like developers do on Ethereum. The difference is that Ethereum is too slow to support a truly high-performance exchange, which is why Jeff initially chose to build his own chain.

Once this chain is open, assets can enter Hyperliquid through a decentralized bridge secured by the protocol itself, without the need for any single party to host it. At the same time, all applications built on this programmable layer can directly access the exchange's order book and all the liquidity accumulated in it. Developers can build lending platforms, stablecoins, or mobile trading apps, and connect directly to markets where professional institutions quote billions of dollars every day.

Jeff doesn't like analogies. He'll tell you that Hyperliquid has no target in traditional finance, and people always tend to cram new things into the old framework to understand rather than understand them according to their own logic. This is a mistake. But for outsiders like us, this is more like Amazon initially building cloud services to support e-commerce businesses, then later discovering that the cloud business itself is a bigger process than e-commerce. In that Twitter post, Jeff first used this statement: Hyperliquid will “host the entire financial system.”

In fact, he has always resisted this transformation. He told me that in his subconscious mind, he didn't want to take on all of this. Building a virtual machine in Hyperliquid was an extremely large project, and the team wasn't sure it would actually work. They don't know how many of the underlying components have to start from scratch. But at some point, it became self-evident, and if they didn't, they would spend years piecing together a system that was “both Binance and Ethereum-like”, but in the end, both would be incomplete — and they're bound to regret it.

The community was extremely outraged. The user was expecting an airdrop, but ended up waiting for a tweet about infrastructure. One review that received thousands of likes quoted the “Breaking Bad” motto: “We had a good thing (we used to have everything beautiful).” Others said, “I hate this; you betrayed us.”

Users don't want a blockchain; what they want is money. Xulian — a member who joined the team through a user interview that was expected to last 15 minutes and extended to an hour and a half, or even “never really finished” — took on a lot of negative emotional pressure. “Jeff is thinking about the best long-term solution,” he said to me. “We don't really care if we look pretty in the short term.”

As iliensinc said, those who are the loudest will eventually get exhausted. Over the next six months, the team continued to advance spot trading, improve the programmability layer, test on independent networks, and prepare for the staking mechanism. Then, on November 29th, one Friday, HYPE officially arrived.

Hyperliquid airdropped 31% of its total supply to approximately 94,000 early users. There are no strings attached, and no lock-up period. As long as you've used the platform and earned points, you'll wake up that morning with more tokens in your wallet — your wealth has grown compared to when you went to sleep. Based on the opening price, the airdrop was worth over $1 billion; at an all-time high, its total value once reached $16 billion. It was the largest transfer of wealth in crypto history, and every dollar went to users.

The team's own distribution ratio is 23.8%, which is not only lower than the community share, but it also takes years to gradually unlock. On the day of the airdrop, they got nothing. Venture capital institutions also received no allocations — if they wanted tokens, they had to buy them on the open market for the same price as everyone else, and only on Hyperliquid because it wasn't listed on other platforms. This in itself is a kind of “price to pay.”

Jeff didn't need to explain anything on Twitter that morning.

“I woke up to find myself getting a six-digit airdrop,” one user wrote.

“HYPE changed my life today. That's enough for me to live comfortably for a few years, help my family, and invest heavily in a bull market,” another replied.

Another said, “Thank you Jeff for the seven-digit airdrop.”

“I felt great at the time,” Jeff said to me. “It's rare for early players to share the upside benefits and actually own part of the network.” I asked him how it felt after everything they built was publicly priced. His response was, “Very bad.”

On a Wednesday night in late March 2025, iliensinc's computer suddenly sounded an alarm. She was on the phone at the time, and she immediately hung up the phone. On screen, the balance of HLP — Hyperliquid's community funding pool — is declining.

A trader had been testing Hyperliquid's defense mechanisms through small, coordinated positions for the past few days. Now, the test is over. They opened three positions on a niche token called JellyJelly — the token has a market capitalization of around $15 million and a daily trading volume of only $7.2 million. One of them was a large empty order, and the other two were multiple orders. This short order is “deliberately designed to fail”. Traders are shorting an asset that is about to be boosted by themselves, and when the short position closes, the risk will be passed on to others — just as if a grenade were lifted and handed over to others.

This “other person” is exactly the HLP. On Hyperliquid, when the order book is unable to close a position, the community pool takes over the position and gradually closes the position. Under normal circumstances, this is a conventional mechanism. However, JellyJelly had almost no liquidity order book, and when HLP was stuck in a position they couldn't exit, the trader frantically bought the token in the spot market. The price increased by more than 500% in less than an hour. With every increase, the pool's losses expand.

iliensinc stared at the screen and watched the losses expand all the way from $5 million and $8 million to $12 million. There is no mechanism in the system to automatically stop this process — because no one had ever imagined that someone would use a $15 million token as a “weapon.”

At this point, validators distributed across Asia and Europe were launched one after another. Hyperliquid's blockchain is maintained by around 20 validators — they get voting rights by staking large amounts of HYPE and are responsible for verifying all transactions. Many people were users before the tokens were issued. They can see it all on the same public ledger — just like anyone in the world can see it — and they don't think it's a normal transaction. Within minutes, all validators voted unanimously to remove JellyJelly from the shelves and settle at the price before the manipulation began. All users with legal positions were compensated, and only the attackers lost.

This incident raised the question that Hyperliquid's critics have been waiting for: if more than 20 validators can reverse the market price and settle at the price they decide, then how “decentralized” is the system? Jeff didn't shy away from this question. He said the small number of validators was intentional — a system that needed to be upgraded every few weeks, and it was impossible to coordinate thousands of participants each time it was upgraded. The number of validators will increase in the future, but not at the expense of system speed.

“It took a month to fix this issue. “It's really bad to learn the lessons from the attack, but it's better than not having anyone tell you,” Jeff said. Hyperliquid never paid the market maker and never charged the team, but they are willing to pay up to $1 million in bounties for bug reports. “However, these people are clearly not here to report bugs; they are exploiting them.”

At the same time as the attack occurred, Binance and OKX — the world's two largest centralized exchanges — also launched JellyJelly perpetual contracts on their respective platforms. On Twitter, a user called out Binance Co-CEO Ho Yi to suggest that she launch the token.

“If you launch JellyJelly, Hyperliquid is probably over.” He Yi replied in Chinese: “OK, received.”

That's the price of ambition. You left a Puerto Rican beach where no one knows your name, built a system from scratch using a TV and your own savings, rejected a $100 million investment, distributed billions of dollars to strangers — what did you end up with? It's war!

In 2023 and 2024, Hyperliquid was small enough not to get much attention. But the airdrop changed everything. As its market capitalization grew from $4.2 billion to $9 billion and higher, big players across the cryptocurrency industry saw a potential future — Hyperliquid could “take their jobs.” Binance announced the launch of its own decentralized exchange; Coinbase and Robinhood began offering futures products; and new agreements continued to emerge, targeting Hyperliquid as a direct competitor. Meanwhile, someone followed Jeff into his elevator.

Such incidents may have been irrelevant, but in 2025, violent attacks against cryptocurrency holders almost doubled. In France, the co-founder of a hardware wallet company had his finger cut off and photos sent to his partner for extortion; in Canada, a family was tortured by water to extract confessions. Cryptocurrency transfers are instant, irreversible, and require no bank approval—one person with a wrench and wallet address can take away a fortune.

As a result, Jeff moved to a safer place, hired a bodyguard, and his life was somewhat confined to the “safest island city in the world.” When traveling, he will be provided with two personal security guards. iliensinc also began repeatedly training team members: if a stranger asks where they work, how to respond. That's why, in this report, almost all of the interviewees used pseudonyms.

When I asked Jeff what his toughest time in 2025 was, he didn't mention the JellyJelly attack, or rivals or bodyguards. He's talking about an API server.

Over the summer, as Bitcoin surpassed $100,000 and Hyperliquid's monthly transaction volume surpassed $400 billion, the servers connecting market makers to the blockchain began to have problems. More and more institutions are connecting, and each one is sending massive orders, cancellations, and update requests, and the infrastructure that transmits this data is unable to keep up. The order, which was supposed to be filled instantly, was delayed until 3 seconds.

The blockchain itself is not down, and user funds are always safe. But in a market where milliseconds determine victory or loss, 3 seconds is already a serious warning. “If congestion occurs even when there is no extreme fluctuation,” iliensinc said, “then when the real extreme market arrives, this is completely unacceptable.” Jeff had almost no normal sleep for weeks — he fell asleep at 1:30 a.m. and was woken up at 3 because the system had another problem. Eventually, the team rewrote the entire server system from the ground up.

That incident happened on October 10, 2025. Trump threatened to impose 100% tariffs on Chinese imports, and over $19 billion of leveraged cryptocurrency positions were forcibly closed within 24 hours, making it the largest liquidation event in industry history. More than 1.6 million traders have been involved in a self-reinforcing chain reaction: forced sell-offs depress prices, trigger more liquidations, and further depress prices.

Hyperliquid experienced no downtime or suspension of withdrawals throughout the process. The restructured server withstood the test, and the JellyJelly incident repair mechanism was just as effective. As a “last resort,” HLP handled multi-billion dollar closed positions and earned $40 million in revenue from them. But since all transactions on the Hyperliquid blockchain are public, anyone can accurately count their liquidation data. Other exchanges don't disclose data with the same accuracy; for example, Binance only publishes one liquidation record per second. The data aggregation platform that mainstream media relies on can only calculate statistics based on this incomplete data, and the results are naturally biased — media reports say that Hyperliquid's bursts have surpassed all other exchanges. It appears to be the “most dangerous place to trade,” but only because it is “the most transparent.”

Three days later, while the entire crypto market was still taking stock of losses, Jeff's team released an upgrade proposal that would define the future direction of Hyperliquid: Hyperliquid Improvement Proposal 3 (HIP-3). The proposal allows any user who has pledged 500,000 HYPE to deploy a new perpetual contract market on the platform, set their own parameters, select price predictors, and receive half of the transaction fees.

By the end of the year — the second full year of its operation — Hyperliquid had made a profit of approximately $900 million. Of this revenue, not a single dollar went to the team. 99% of the proceeds are automatically converted to HYPE and destroyed, permanently withdrawn from circulation, and almost all of the platform's value is returned to token holders. When I asked iliensinc how they would evaluate 2025, she said, “It feels like we've grown.”

On my last afternoon at the office, Jeff and I sat at the black table next to the kitchen—where the team had lunch together every day, next to those untouched bottles of whiskey. I brought up the question I had left until the end.

Over the past year, Hyperliquid has continued to “spin off” some of its capabilities. Launched long before HIP-3, Builder Codes allow independent developers to build trading applications based on the platform's order book and split them from the processing fees generated by their users. Matt Huang (co-founder of crypto investment agency Paradigm) called it “a fantastic way to franchise the user experience.” Since October 2024, these teams have accumulated over $70 million.

HIP-3 goes one step further. Within six months of going live, seven independent teams have deployed hundreds of markets, most of which are not crypto-related — including oil, gold, stock indices, and forex. The largest deployer, Trade.xyz, has maintained 38% weekly growth since October 2025, with a cumulative trading volume of over $130 billion, covering 192,000 traders. Today, marketplaces created by independent developers account for half of Hyperliquid's total trading volume. By February 2026, HIP-4 was proposed — once it goes live, anyone can deploy options or forecast markets on the platform. HIP-3 opens “any asset with a price,” while HIP-4 will open “any event with a result.”

Today, the most influential products on Hyperliquid are being built by developers who neither work for Jeff nor will ever work for him. I asked him how he felt about this — what the team should do and what should be left to others.

“It's a dynamic question with no standard answer,” he said. “The most critical question is a philosophical question. Are you building a financial superapp (like Robinhood) or a financial system?” He acknowledged that he wasn't sure which path would win, “but I think an accessible financial system would be a better outcome for the world — one built on a common track rather than one controlled by a single company.”

“To achieve this, we often think about how to enable others to succeed on Hyperliquid and own their own businesses. As people compete with each other and have their own products, the system becomes more robust and scalable.”

The easiest path, he said, was to do everything themselves and shut down within a company, and they chose the opposite direction. “This is a more difficult path, but what we care about is the process of achieving our goals, because this process ultimately determines what we build.”

The founder of Trade.xyz told me that maybe one day people won't even realize they're using Hyperliquid. “Perhaps the final form is that it's just financial infrastructure and liquidity itself,” he said. “Interactive Brokers, Phantom, and other platforms are interfaces directly facing users. It's actually pretty beautiful.”

Paradigm invested heavily in the open market shortly after the HYPE token went live. Huang said to me, “What's even more amazing about this is that it was done by a team of 11 people.” 11 people, almost no reliance on AI. In the office, there are AI notebooks dedicated to running the latest models, but they're only used to explore ideas. “We'll keep watching AI's capabilities,” Jeff said, “but it's not enough to write critical code right now.”

I asked Jeff, a question that has always loomed over it: Hyperliquid has accumulated over $4 trillion in transactions since 2023, accounting for 37% of the decentralized perpetual contract market, yet users in the world's largest capital market — the US — cannot use it.

The obstacle is the Dodd-Frank Act — a US law passed after the 2008 financial crisis that requires all derivatives transactions to be carried out through a regulated intermediary. Ironically, Hyperliquid's public ledger has achieved what the bill is trying to achieve: real-time visibility of all leverage in the system, but until the US Commodity Futures Trading Commission establishes new rules, US users are still unable to legally trade derivatives through decentralized agreements. Continuing his consistent philosophy, Jeff did not form his own policy team. A month after my visit, Hyperliquid Policy Center was founded as an independent non-profit organization led by senior crypto attorney Jake Chervinsky; Hyper Foundation (an independent organization supporting ecosystem development) provided 1 million HYPE (approximately $28 million) as start-up capital.

Jeff admits that Hyperliquid has reached a stage where “just build and wait” is no longer a strategy. “There are indeed people promoting policies in the opposite direction,” he said. “I'm not really sure what will happen in the end. But regulation ultimately reflects public will, and I am optimistic about where it is headed.”

I kept the last question for a whole week: “Do you really think Hyperliquid will host the entire financial system?”

He smiled — for someone who even cuts his own hair, this smile appears more often than you might think. “The word 'all' is really an exaggeration,” he said. “That's just our vision. But this is a very difficult matter. It can easily cross decades of goals, and it itself has a certain amount of pride.”

“It's a bit like the difference between Go and chess,” he continued. “In chess, the higher your level, the more steps you can predict; in Go, there are so many possibilities. The key is to build intuition about the next step rather than trying to deduce the whole decision tree.”

I probably wanted to hear more, so he changed his mind. He has always tried to follow a principle — to be extremely confident in the direction, but when carrying out this step, you don't need to know exactly where the end is.

The next night (Friday), the team had dinner at a Chinese restaurant located in the hotel. The engineer who turned the office into a “plush zoo” didn't show up. Everyone else came along, and me. We walked through the quiet lobby, walked into a private room with dark wooden boards and carved partitions, and sat around a round table. Before eating, we had tea in the sofa area on the other side.

The room was a bit cold, and the AC seemed to be set for hotter weather. Someone handed the youngest engineer a blanket, which he draped over his body, only to find out that it was Christian Dior's. This sparked a discussion about luxury brands, and Jeff apparently had no experience with it. They even read LVMH as “LHVM,” but no one corrected each other, and iliensinc, wearing a Ralph Lauren hat, sighed.

During the meal, the turntable rotates continuously, and a dish is placed. Until the end, a large blue and white porcelain bowl was served on the table, and everyone suddenly quieted down. The shallow water in the bowl is covered with pebbles and small leaves, like a miniature koi pond. A white bowl with noodles is placed in the center, and three small orange fish swim in the “channel” between the two bowls. The waiter explained that the fish needed to rest for 30 days and only work for 5 minutes. We watched them swim in circles, then they were taken away to begin the next round of “vacations.”

At about 9:15 p.m., we left the restaurant and walked into the drizzling rain. After saying goodbye, I took a taxi to the airport. The car drove out not far and followed a curved elevated road. When turning, the financial district caught my eye: the signs of HSBC, J.P. Morgan, Standard Chartered, Deutsche Bank, and Citi were shining in the night sky. The road then extended east, and one of these tall buildings disappeared into the rearview mirror until only a wet surface was left.

Jeff went in the opposite direction — back to the office, where his bodyguard was waiting for him.

Twitter:https://twitter.com/BitpushNewsCN

Compare the TG exchange group:https://t.me/BitPushCommunity

Compare TG subscriptions:https://t.me/bitpush