Hyperliquid co-founder Jeff Yan said the biggest risk facing the crypto sector is not the usual list of regulation, competitors or market cycles, but the fact that many of the best young builders are choosing to work on AI instead.
Yan says he spends less time on dashboards than on missing products
Hyperliquid is one of the most closely watched platforms in the onchain derivatives market, where performance is often judged through trading volume, open interest, liquidity depth and market share. Yan said he does not actually spend much of his time staring at those figures.
He said the numbers matter, and noted that members of the Hyperliquid community have built third-party dashboards to track trading volume, market share, volume in specific assets, spot share and liquidity. But the question that occupies more of his time is qualitative rather than numerical: what product still needs to be built that does not exist yet?
According to Yan, Hyperliquid tries to introduce genuinely new implementations even when an idea is rooted in an existing financial product. The goal, he said, is to produce a real “zero to one” step. If a feature has already been launched by 10 teams, Hyperliquid may still add it in the interest of product completeness, but he said that kind of work usually does not deserve the team’s full attention.
He cited HIP-3 as an example, saying the architecture for permissionless market deployment did not have a mature precedent before it launched. Because the product did not already exist, the team could not simply extrapolate from past trading volume or user data to decide what to build next.
Chasing small metric gains can crowd out product innovation
Yan said trading volume, market share and liquidity comparisons still matter in the end because they reflect whether a product succeeds. His point was that the causal chain should not be reversed.
If a team spends each day thinking about how to add a few percentage points to volume, he said, it can drift toward subsidies, short-term campaigns or features copied from competitors. That approach can leave little room to rethink the structure of the market itself.
Even if builders ultimately care most about the numbers, he argued, the effective path is not constant micro-optimization of metrics. It is to keep finding “zero to one” product improvements. In that framing, data is the outcome of innovation rather than the starting point that constrains it.
He said that logic helps explain why Hyperliquid has not stopped at perpetual futures and has continued expanding into permissionless markets, spot trading, asset deployment and outcome markets.
HIP-4 outcome markets, in his view, are not just event prediction tools
Discussing Hyperliquid’s proposed HIP-4 outcome markets, Yan said they should not be understood only as prediction markets for elections, sports or other events. At the protocol level, he described outcome markets as a different margin and payout system.
Under that design, both sides of a trade post enough funds in advance to cover the maximum possible loss. Positions are fully collateralized, losses are capped and the contracts do not need forced liquidation in the way high-leverage perpetual futures do. Final settlement can be completed based on an offchain event or an oracle-fed outcome.
Yan said that could produce a derivatives structure with relatively stable behavior and clearer risk boundaries. On criticism that outcome markets may encourage speculation, he said markets and speculation cannot really be separated. Any financial market tied to future events, prices or risk will involve a gap between expected returns and actual outcomes. If speculation were removed entirely, he said, price discovery and hedging functions would be difficult to preserve.
He sees onchain options as the more important next use case
Yan said the next application worth watching for outcome-market design may be onchain options rather than event prediction alone. In traditional finance, options are an important hedging tool. Investors do not always want to sell an asset outright or build an equal and opposite position. In some cases, they simply want protection if prices fall sharply.
He said an asset holder can pay a premium to buy downside protection, limiting the worst-case loss while keeping upside exposure intact.
Yet onchain options have not produced a breakout on the scale seen in perpetual futures. Yan said one of the main reasons is the combination of capital efficiency and liquidity constraints. Options market makers have to quote a large number of contracts across different strike prices and expiries. If every quote has to be fully collateralized in advance, a large amount of capital gets tied up and the market struggles to build enough depth.
Fully collateralized design may improve resilience, but it comes with trade-offs
Yan said perpetual futures on Hyperliquid use a structure where collateral is posted upfront. Crypto users may already be used to that model, though he noted that traditional futures markets do not necessarily follow the same setup in identical form.
The benefit of prefunding, he said, is that when markets move violently, positions are already backed by collateral. That can strengthen overall system resilience and reduce counterparty non-performance risk. But if the same logic is applied directly to traditional options, capital efficiency becomes a serious problem.
Yan said HIP-4 outcome markets may offer a middle path. Rather than copying standard options contracts outright, they could create option-like nonlinear payoff structures through fully collateralized, capped-loss contracts. That could give users something resembling downside protection or event-risk hedging while avoiding unlimited losses and cascading liquidations.
He added that the final product may not be considered a standard option by traditional finance, but it could become a native version better suited to onchain capital and settlement conditions.
Why he thinks talent migration to AI matters more than direct competition
Asked about the biggest risk facing Hyperliquid and the criticism outsiders should take most seriously, Yan did not put rival perpetual platforms, blockchain competition, regulatory pressure or smart contract risk at the top of the list.
Instead, he said the deeper problem is that crypto, fintech and onchain finance still have not attracted enough top-tier entrepreneurial talent. For newly graduated computer science students and young engineers, AI now offers the most funding, the most social attention and the most startup opportunities. He said some may even leave school early to start AI companies.
By contrast, he said, very few top young builders look at crypto or fintech and see the place where they can have the greatest impact on the world.