Zero VC, 200,000 Orders a Second: Jeff Yan’s Hyperliquid Bet

1. A quant team chooses the order book

Jeff Yan graduated from Harvard in 2017 after studying mathematics and computer science, then joined Hudson River Trading as a quantitative trader. In Yan’s account, the job taught him to examine how markets work and to apply quantitative reasoning to trading. He discovered Ethereum around 2018 and remained focused on crypto after leaving HRT. Those experiences—not a single dramatic event—supplied the market knowledge and technical ambition behind Hyperliquid.^[3]

By mid-2022, Yan’s small team was already trading crypto and comparing centralized and decentralized venues. Hyperliquid’s own July 2023 retrospective dates the start of order-book perpetual DEX development to September 2022, followed by an Arbitrum Goerli paper-trading competition in November and an L1 testnet in December. FTX collapsed only after development had begun. Yan later described the failure as the impetus to commit fully: it made the value of self-custody obvious while exposing how far decentralized trading still lagged in speed and usability.^[3][7]

Yeah, it has entirely been self-funded.^[3]

— Jeff Yan, Wu Blockchain interview

2. What HyperBFT actually built

Hyperliquid did not arrive at its present architecture in one step. An April 2023 engineering post described the young L1 as a Rust implementation using a tuned Tendermint consensus and processing 20,000 operations per second. The current documentation instead describes HyperBFT as a custom consensus algorithm inspired by HotStuff and its successors, with state split between HyperCore for native financial logic and HyperEVM for general-purpose smart contracts. Reading both primary records shows an evolving chain rather than a finished design appearing overnight.^[1][2][8]

HyperCore contains fully onchain order books for perpetual contracts and spot pairs. According to official technical specifications, every limit order, cancellation, trade, and liquidation settles transparently onchain with deterministic one-block finality inherited from HyperBFT. The core technical achievement is establishing a single, globally consistent transaction order for all matching and margin calculations without offchain sequencers. This architecture does not imply zero trading fees, total protection from maximal extractable value (MEV), or an absence of operational trade-offs.^[1][2]

Current mainnet documentation rates capacity at approximately 200,000 orders per second and identifies execution as the bottleneck. That is a later benchmark and a different unit from the 20,000 operations per second reported in April 2023; neither number is generic TPS. The current latency figures—0.2 seconds at the median and 0.9 seconds at the 99th percentile—also apply to geographically co-located clients. The dates, units, and test conditions turn a speed headline into a claim readers can compare.^[2][8]

3. Growth without moving the goalposts

Hyperliquid’s growth across 2024 was substantial, but early- and mid-2024 metrics are easily misunderstood when removed from their historical context. DeSpread’s research chronology ties the protocol's initial surge to its points program, and by the end of that first points season in May 2024, it ranks Hyperliquid as the largest perpetual DEX by trading volume, open interest, and daily active users. That ranking represents a dated leadership snapshot from a specific research chronology, not a permanent market condition or a statement about subsequent periods.^[4]

The same discipline applies to subsequent milestones. Around the November 2024 ecosystem expansion and HYPE genesis, DeSpread documented consecutive protocol highs, including $150 billion in cumulative trading volume and $4.3 billion in open interest. These metrics are distinct: cumulative volume aggregates all executed transactions over time, while open interest measures active, outstanding derivative contracts at a specific moment. Neither figure supports casual claims that Hyperliquid routinely processed $4 billion in daily trading volume during early or mid-2024.^[4]

CoinGecko’s 2024 Annual Crypto Industry Report provides a rigorous full-year comparison set. Among the report’s tracked top-ten decentralized perpetual protocols, Hyperliquid captured more than 55% of trading volume in Q4 2024 and reached 66% in December. Its open-interest share closed the year at 66.2%. These figures validate late-2024 leadership while correcting two common distortions: the denominator was tracked perpetual DEXs rather than all exchanges including Binance, and the percentages describe the fourth quarter and year-end rather than the present day.^[6]

4. HYPE genesis: large, not “100% community”

On November 29, 2024, Hyperliquid conducted the genesis distribution of its native token, HYPE. Users received 310 million HYPE, equal to 31% of the fixed one-billion-token supply. Decrypt reported that another 38.88% was reserved for future emissions and community rewards, with CoinGecko independently corroborating the 31% user allocation and November 29 date. It was a historic distribution model, but claiming that the entire token supply went to the community erases structural reserves and turns a defensible milestone into an inaccurate absolute.^[5][6]

Basically, no. There were no private arrangements, like profit-sharing deals or investments.^[3]

— Jeff Yan, answering a question about market makers in the Wu Blockchain interview

The distribution reporting confirmed that no HYPE was set aside for private investors, centralized exchanges, or market makers. This precise negative constraint is stronger than promotional shorthand because it defines exactly who lacked an allocation without pretending that no reserves or future emissions existed. Yan separately described the Hyperliquidity Provider (HLP) pool as a protocol-owned vault rather than a private market-making desk. Genesis design thus eliminated conventional private-investor unlock pressure without removing all long-term governance concentration or protocol execution risks.^[3][5]

5. What the experiment proves—and what it does not

Hyperliquid demonstrates that a specialized onchain venue can capture demanding derivatives volume when consensus, order matching, and risk engines are designed together from first principles. HyperCore avoids reliance on offchain sequencers, while HyperEVM provides a programmable smart-contract surface on the same chain. Nonetheless, the architecture must be evaluated across validator decentralization, client diversity, software correctness, and resilience under extreme volatility—not throughput alone. Purpose-built performance offers a clear operational advantage, but it does not eliminate the fundamental trade-offs inherent in decentralized state machines.^[1][2]

Its financing model offers an equally bounded lesson. In his interview with Wu Blockchain, Yan explained that repeated venture funding rounds felt like an unhelpful proxy for product progress, arguing that real progress occurs when users derive tangible value from what a team builds and participate in the upside. By choosing to remain entirely self-funded, the team retained independence and aligned economic incentives with active platform users. The lesson is not that every Web3 project can or should reject venture capital, but that financing is a deliberate design choice with structural consequences, not an obligatory milestone.^[3]

Real progress is users actually getting value from what you’re building, and users being able to benefit from the upside—not a bunch of investors who just threw some money in at an early stage.^[3]

— Jeff Yan, Wu Blockchain interview