At Blockworks Digital Asset Summit Asia in Singapore on Oct. 7, 2026, Hyperliquid Labs co-founder and CEO Jeff Yan said Hyperliquid should be viewed less as an exchange competitor and more as neutral financial infrastructure, closer in spirit to the internet or Linux.
Speaking with Blockworks research analyst Shaunda Devens, Yan described the project as an effort to build a compact set of composable tools that can eventually support all forms of financial activity. The conversation covered HIP-3, HIP-4, cross margin, native lending, priority fees, permission controls and the question of whether public order flow can improve execution.
Yan also said options, spot markets and a high-quality asset issuance layer are among the most obvious areas to watch next.
Keep the base layer small, but make it composable
Devens noted that Hyperliquid this year introduced HIP-4, described as outcome contracts, cross margin, a native lending protocol built on top of that, priority fees and HIP-3* on testnet. She asked how Yan thinks about the broader strategy behind those product decisions.
Yan said the strategy is to assemble a toolkit that can eventually carry all financial activity. Repeating that goal is not enough, he said. If someone wants to build a product in the system and finds they cannot, they will build elsewhere. For that reason, the real task is to make the tools usable.
He said the team looks at what is already possible, what builders want to create on Hyperliquid and what they still lack, then tries to extract common needs from that feedback. That process helps determine where the protocol should go next.
One principle matters a lot, in his view: the base components should not multiply to the point that the system becomes hard to understand or hard to compose. Each component should be precise in design, widely useful in practice and able to work cleanly with the others.
Cross margin is one example, he said. A lending component can connect with perpetuals, spot markets and other trading components. That kind of fit gives developers more room to build while keeping the overall design simple. Finance should work that way, Yan said: not many parts, but parts that cooperate well.
HIP-3 and HIP-4 may need time before the market fully understands them
Devens said Hyperliquid often faces a lag between launching a feature and seeing the community fully grasp why it matters. She pointed to HIP-3. Even on testnet, it was already clear that the framework could create markets around any asset with a price feed, but broader understanding only came after trade[XYZ] and its founder Shoku turned that capability into a concrete product. She asked whether other features today may be in a similar phase.
Yan said yes, though it is hard to predict how long the gap will last. Some things may never take off at all, he said, because real-world conditions can differ from the original assumption.
Still, he said every foundational component has the potential to create a large new story on the scale of HIP-3. Those stories have not been written yet. They are waiting for someone to turn the raw capability into a product, and that is the hard part.
He singled out HIP-4. People often ask why they cannot trade options, he said, given how important options are in traditional finance. The answer is not that perpetuals are better than options. The two are strongly complementary and serve different needs.
HIP-4, he said, provides a base capability that could let developers build options or products with options-like payoff curves and scale them into something genuinely useful to users. If he had to name a candidate for the next still-unwritten story, similar to HIP-3, HIP-4 would be one.
RWA-related open interest rose from about 300 million to 3.9 billion
Devens said HIP-3 has already become a major growth driver for Hyperliquid. She cited open interest in real-world-asset-related markets, saying it had increased from about 300 million in January to 3.9 billion by the time of the interview, a 13-fold rise.
She added that traditional finance has also started to understand the Hyperliquid vision, especially around 24/7 trading and pre-IPO-related markets. When she began studying Hyperliquid in 2024, she said, the natural comparison points were Aevo and dYdX. Later, the comparison shifted toward large crypto exchanges. Now, in some cases, it has moved toward CME and Nasdaq. She asked whether that shift in framing changes Yan’s strategy.
Yan said those comparisons are all incomplete. In his view, the protocol is trying to do something that no one has really done before, and he does not see a clean analog anywhere around it.
That is a good sign, he said, because it suggests the team is not building something that would have happened anyway with or without Hyperliquid. The fact that the comparison set keeps changing may simply mean people are slowly recognizing that the protocol has many different sides.
What it really points to, he said, is public infrastructure more like the internet or Linux. "You don’t need 10 systems redundantly doing the same thing. You need one neutral system."
He added that the institutions often mentioned in those comparisons may not end up as competitors at all. If the technology fits their use case, they could benefit from it. The community is not trying to replace them, he said. The aim is to offer better capabilities so they can serve users better.
HIP-3* adds optional permission controls on top of HIP-3
Devens described HIP-3* on testnet as an extension of HIP-3 that still lets deployers launch their own markets but adds permission features such as whitelist-based trading access. She also noted that Kraken parent company Payward has publicly said it hopes, subject to regulatory approval, to use that mechanism to offer Hyperliquid-related services to U.S. users. She asked how Yan wants these venues to interact with the broader ecosystem.
Yan said the issue goes back to the same question of which base components the protocol should provide. Some developers want to build strong products, but the current permissionless components do not fully meet what they need. HIP-3* is an attempt to distill that feedback and satisfy it with a limited set of changes.
Where possible, he said, the team should not reinvent the wheel. HIP-3 already works well and already contains many of the properties deployers want. Regardless of the regulatory framework in which a business may operate, transparency, self-custody, redundancy and resilience all have real value.
On top of that, Hyperliquid is adding controls that deployers can choose whether to enable. That gives them a way to launch the markets they want while remaining aligned with the protocol’s existing onchain functionality.
Devens said the design reflects a view Yan has expressed before: there is no separate DeFi and TradFi, only finance, and the system Hyperliquid is building is a fundamental improvement. HIP-3*, she said, expands the set of options available to individuals and institutions without taking anything away from the original model.
She contrasted that path with many other L1 networks, which start from generality and then work toward better performance. Hyperliquid, she said, started as a high-performance L1 focused on perpetuals and then added spot markets, outcome contracts and native lending. She asked whether the next priority is to keep adding new base components or to improve the performance of what is already there.
Yan said both matter. Performance constraints will always exist, he said, and that is a healthy problem to have. Every order-of-magnitude increase in usage opens up many new possibilities and creates a new class of technical problems. He does not expect that process to end.
Any L1 would want to face the question of how to satisfy real demand for the chain and for block space, he said. That is an interesting engineering problem, and one reason the team wants to work with strong engineers.
He added that it is satisfying to build a system, discover later that it is no longer enough and realize that the reason is a jump from 1 million users to 10 million users.
On the product side, Yan said the list of foundational components should not be fixed in advance. If the goal is to support the global financial system, there is no reason to pick a day and declare the protocol complete. That would not be a practical way to evolve a system or serve users and developers.
Priority fees and the attempt to reduce the latency arms race
Devens said Hyperliquid’s recently added priority fees may still be underappreciated. Read priority fees let traders stream unconfirmed order flow from the mempool, while write priority fees let traders attach a HYPE-denominated fee to gain earlier queue position for their orders.
She said the mechanism has already burned close to 180,000 HYPE. More important, she argued, is that latency competition in traditional finance is extremely capital intensive, sometimes requiring hundreds of millions of dollars just to participate. In that sense, she said, priority fees bring that competition into Hyperliquid in a fairer way.
Yan said that framing is useful, though he would not call the fee a "bribe." He compared the idea to a change Ethereum once made to its fee structure, though he said he did not remember the exact year or the exact EIP. In the earlier system, fees largely went to validators, and the word "bribe" would fit that older setup more naturally, he said.
In his view, that older approach was unhealthy and looked in some ways like the latency arms race in traditional markets. The point of priority fees is to make markets better at executing trades so that alpha tied to technical edge does not outweigh alpha tied to genuinely differentiated information or judgment.
A market exists to let people with different views and different information trade, he said, and to aggregate those signals into what amounts to collective intelligence. A trader’s unique alpha cannot simply be absorbed into system design unless the market structure changes in a radical way.
Other advantages, though, come from microwave towers and transatlantic fiber, where the edge is simply seeing something happen in Chicago and reaching Tokyo first. That type of edge is obviously zero-sum, he said, and in some respects negative-sum because participants are forced into continual speed competition.
If the result is only that the best bid and offer, or BBO, updates 1 microsecond faster, Yan said he does not think the world gets much real value from it.
That is why he sees priority fees as an effort to create more equal participation by internalizing those negative externalities. In traditional finance, even if someone has alpha, they may not be able to express it through trading unless they first clear a minimum threshold of low-latency infrastructure spending. The market then loses that signal.
Hyperliquid wants a mature priority fee system in which anyone can enter the market with only modest infrastructure and trade from a more equal starting point, he said.
Devens summarized that as pulling value away from the capital-intensive layers of traditional finance and bringing it inside the system. Yan agreed. That value, he said, used to flow to so-called toxic participants and infrastructure providers. Toxic, in this context, is market jargon rather than a moral judgment: it refers to traders who fill before others can react.
If that competition is pushed to its equilibrium, he said, the logical endpoint may be that the people laying cable and building microwave infrastructure capture all of high-frequency trading’s alpha. That sounds absurd, he said, but it is where those incentives can lead.
If he were not building infrastructure, Yan would look at options
Devens said neutrality is one of Hyperliquid’s defining traits. The team focuses on infrastructure and creates economic incentives for others to build products on top. She pointed to participation from teams such as trade[XYZ] and Phantom, then asked what Yan would build if he were no longer allowed to work on infrastructure itself.
Yan said he would start by looking for things no one has built yet that still appear promising, while also considering what other developers are already doing. At the moment, he said, he would probably look at options because he keeps hearing demand for them.
Perpetuals satisfy many users, he said, but some traders understand and use perpetuals and still feel that those instruments cannot fully express their market views. A year ago, he was not sure whether options traders might ultimately prefer perpetuals anyway. He no longer thinks that is true.
HIP-4, in his view, is a natural path for building an options protocol. Developers can use a spot-like order book and pair it with cross margin. Market makers would also have perpetual markets in the same ecosystem to hedge option delta risk.
The next challenge, he said, is turning that product form into something with broad appeal. He does not think crypto options, or products with convex payoff profiles more generally, have really broken out at scale yet, even though they clearly have in traditional finance.
So if he were building in that area, he said, he would try to understand the gap and then tell infrastructure developers what HIP-4 still needs to add or expand in order to make that vision real.
Devens added that many people initially read HIP-4 as a prediction-market feature, but its importance is much broader. What matters is that those capabilities exist inside the same ecosystem and can be combined with perpetuals and spot markets. That composability is where the value comes from.
Can public order flow actually help execution for large trades?
Devens said there is active debate around how public and transparent blockchains should be. She referred to one study comparing hidden TWAP, or time-weighted average price, orders on Hyperliquid with TWAP orders whose execution schedule was disclosed in advance. The public orders, she said, produced better execution because market makers had time to digest the flow. She asked whether Hyperliquid needs a privacy layer.
Yan said his answer may be controversial and that he has written about it in long posts over the past few years, but he prefers to start from first principles.
Before electronic trading, he said, people shouted in trading pits to compete for execution. In many cases it was not simply first come, first served. The person who was taller or louder often got the better fill.
Electronic markets changed throughput and fairness, but also something basic: everyone could see what was happening. You no longer needed to be tall, he said. You just needed market data.
When everyone can see the order book, he argued, the order book should become more liquid because all participants know what is happening and can incorporate that information into their algorithms. That is one of the benefits of transparent, open and electronic systems. He said he sees no reason to assume in advance that transparency should stop at a particular point or that the current information shown in order books is exactly optimal.
Hyperliquid put the order book onchain from day one for another reason too, he said: order books should not be run by centralized intermediaries with special powers. Even in crypto, he said, most order books are still operated by such intermediaries.
The operator of an order book holds a great deal of power. If the goal is to build a truly global and neutral financial system, Yan said, that power should not sit with one party.
That choice also changes execution quality. Hyperliquid’s order book is more open than traditional finance, he said. Participants can see not just displayed size, order counts and specific orders, but also the address tied to each order.
Traditional finance has no single settled standard on what should be public, he said. In Brazil, for example, order-book data feeds disclose the brokers associated with many orders. That is only a rough classification, but it already creates many phenomena worth studying.
Hyperliquid goes much further by exposing information at the level of onchain addresses. From first principles, he said, that should produce a more efficient market, and he finds it interesting that research continues to produce empirical results pointing in the same direction.
He gave another example from traditional finance: leveraged ETFs that must rebalance each day to maintain 2x or 3x leverage. If prices rise, they buy more. If prices fall, they sell. That usually happens in the closing auction.
Telling the market the product rule, the execution formula and the method in advance can actually improve execution, he said. Market makers know what order flow is coming and can plan around it. Because they know the flow comes from a non-toxic source, they can also quote more competitively.
Those empirical observations and first-principles arguments, he said, help explain why transparent markets may actually be better for end users when it comes to executing large trades.
Devens said the issue is also one of fairness. On traditional platforms or centralized exchanges, the data are often not absent; rather, access is restricted and special permissions matter. Public data can be fairer. She said that contrast stands out when researching Hyperliquid versus traditional exchanges: Hyperliquid data are directly available, while traditional finance often comes with access hurdles and paid datasets.
The community should focus on users, not on beating rivals
Devens said Yan often stresses that Hyperliquid is not just Hyperliquid Labs and that the community is part of the team. She asked what feedback he has for that community as co-founder and CEO.
Yan said the community is so diverse that broad advice is hard to give. Still, he said he thinks people are doing a great job and that he is proud of what they are building.
If he had to offer one reminder, it would be to keep in mind why all of this is being built and what the underlying values are. The end goal is to help users, he said, not to get distracted by beating competitors or proving oneself right.
He also said the community should stay humble. Three years ago, Hyperliquid was only an idea, with a small group of people working day and night to make it real. The growth the project has achieved since then involves a great deal of luck.
That does not justify complacency or a sense of superiority over other protocols, he said. He was not suggesting the community has failed on that front, only that this is the principle he would emphasize if asked.
How success should be measured
Devens noted that developers and market deployers carry substantial responsibility in the HIP-3 and HIP-4 systems, especially in bringing markets and products to users. She asked how the infrastructure layer measures success and which metrics matter most.
Yan said the team does watch broad metrics, though not with the level of detail a research group would. Personally, he said, he prefers a qualitative standard: has something appeared that did not exist before?
If the answer is yes, and if the product is genuinely good and users choose it because they truly prefer it, that is the biggest success, he said. Infrastructure exists so people will use it and build on it.
Seeing that kind of zero-to-one change is what gives him the most satisfaction. He cited the 24/7 markets enabled by HIP-3 as one example, saying that several such cases have emerged in just a few years.
What comes next: options, spot and an asset issuance layer
Near the end of the discussion, Devens referred to Yan’s earlier comment that building Hyperliquid feels a bit like playing Go: no one can calculate every outcome or fully predict the path, but it is still possible to have conviction about the next one, two or three moves and proceed with instinct. She asked which directions now look most obvious.
Yan said options are one. Spot, perpetuals and options together form a natural trio: the asset itself, a leveraged linear instrument and an instrument with convex payoffs. Those three categories cover most of the basic trading forms.
Hyperliquid is best known today for perpetuals, he said, but he wants the platform to expand across the other categories as well. Spot is another clear direction with large potential.
He added that the usefulness of cross margin depends heavily on which assets the system supports. If people already hold an asset and can also use it as collateral inside Hyperliquid, that becomes powerful. The task, then, is to bring the asset itself onchain, not just its price.
That network effect is harder to start from zero, he said, because simply holding an asset does not by itself generate trading activity. It also takes more work to tokenize a real-world asset than to track its price through a perpetual market. The latter is, in some sense, cleaner and closer to a math problem.
But if Hyperliquid is meant to support all financial activity and become infrastructure that financial institutions can plug into, he said, it needs a high-quality asset issuance layer.
Devens said HIP-3 barely existed a year ago and is now one of the most discussed topics around Hyperliquid. She said she could easily imagine spot markets and options becoming the center of discussion a year from now.
Yan replied that the expectation is high, but there are already strong teams working in those areas. If teams are interested in those directions and have not yet considered deploying on Hyperliquid, he said, it is worth considering.
Whether the goal is to issue assets or design options-style contracts with convex payoff structures, what he most wants to see is people actually using the products. Another advantage of building on Hyperliquid, he added, is access to an already existing and globally distributed network of market participants.
Original video link: https://www.youtube.com/watch?v=e0xi_NeVk4w