Jeff Yan On Hyperliquid’s Plan To Bring All Finance Onchain

Jeff Yan described Hyperliquid’s ambition as building useful tools that could ultimately support all finance onchain, rather than trying to replace any one exchange. He compared the protocol to neutral public infrastructure such as the internet or Linux: existing firms and new teams should be able to build on it, with builders’ needs informing how it develops. His preference is for a small number of precise, broadly useful primitives that compose well. That approach carries risk: some primitives may fail to gain traction if their assumptions are wrong or builders do not execute.

Shonda Devins said Hyperliquid had shipped HIP4 outcomes, extended portfolio margin to native borrowing and lending, added priority fees, and put HIP3 on testnet during the year. She reported that real-world-asset open interest had grown from $300 million in January to $3.9 billion at the interview, a roughly 13-fold increase. She pointed to 24/7 trading and three IPOs as signs of increasing interest from traditional finance. Yan said continued growth also depends on assets themselves being available onchain: supporting only an asset’s price is not enough for borrowing, lending, or broader financial use.

Yan called a high-quality asset-issuance layer necessary for institutions to use Hyperliquid, while cautioning that issuing spot assets is harder to bootstrap than perps and that tokenizing real-world assets is more complex than tokenizing their prices. Portfolio margin can connect borrowing and lending with perps and spot, but its usefulness depends on the collateral it supports. HIP3 Star, which the interviewer said was on testnet, adds controls such as trader whitelisting for deployer-listed markets. Yan said it incorporates builder feedback through opt-in changes while preserving HIP3’s transparency, self-custody, redundancy, and resiliency. The interviewer said Payward hoped to use it to offer Hyperliquid to US citizens, subject to regulatory approval.

Yan sees options as a complement to perps, not a replacement for them. He said HIP4 can support options or options-like payout curves using spot order books and portfolio margin, with perps available to hedge delta. In his view, HIP4 is broader than prediction markets, although convex-payout products have not yet achieved crypto’s equivalent of their adoption in traditional finance. Yan said he would explore options as a builder, but stressed that teams creating assets or options need users; Hyperliquid’s global network of market participants could help connect them with that demand.

The discussion of priority fees centered on how they can change competition for execution. The interviewer described read priority fees as allowing traders to stream uncommitted order flow from the mempool, and write priority fees as letting traders attach HYPE fees to seek queue priority; the interviewer also cited about 180,000 HYPE burned. Yan rejected the description of these fees as “bribes.” He argued that they can internalize latency-related externalities and lessen reliance on costly advantages such as microwave towers and faster transatlantic connections, which he characterized as zero-sum or potentially negative-sum competition.

Yan argued that open order books let market participants observe activity and incorporate it into their algorithms. Hyperliquid’s onchain books expose unusually granular order-flow information, including addresses associated with orders. He cited a study reportedly finding that publicly telegraphed TWAPs executed better than hidden TWAPs, because market makers had time to absorb the flow. He expects transparency to improve efficiency and execution, including for large orders; predictable activity, such as leveraged ETFs’ recurring rebalancing, can be priced competitively when its rules and timing are known. He also said every order-of-magnitude increase in usage brings new performance challenges, and described demand for blockspace beyond current capacity as a welcome problem. For Yan, infrastructure succeeds when teams build new kinds of products and users choose to use them.