Hyperliquid Seeks Perpetual Contracts on AI Computing Power, Requests CFTC Approval

After oil, chips. The Hyperliquid Policy Center, the lobbying arm of the on-chain derivatives exchange, submitted a comment letter to the U.S. futures regulator on Tuesday, October 20. Its request: to allow perpetual contracts in the United States based on the rental hours of a GPU, rather than bitcoin or crude oil. Hyperliquid had already advocated in August for energy perpetuals and 24/7 trading. This time, the exchange is pushing the king tool of crypto towards the raw material of artificial intelligence.

Key points of this article:

• The Hyperliquid Policy Center has asked the CFTC to authorize perpetual contracts based on GPU rental hours, an essential raw material for artificial intelligence.
• The CFTC has deemed the creation of derivatives products based on computing power crucial for American competitiveness in the AI race, opening a public consultation on this topic in August.

Why the CFTC is Interested in Compute

Let’s start from the beginning. On August 19, the CFTC (Commodity Futures Trading Commission, the U.S. derivatives watchdog) opened a public consultation on the opportunity to create derivatives products backed by computing power, in other words, the rental price of processors that train and run AI models. Its chairman, Michael Selig, stated that America "cannot win the AI race without a robust derivatives market on compute," according to Reuters. The consultation focused on four points, including market oversight and the possibility of perpetual contracts on compute.

Why is a commodities regulator taking up this issue? Because the price of compute fluctuates significantly, and no one can hedge against it. According to figures cited by the HPC, an hour of renting an Nvidia H100 chip was trading between $2.50 and $3.19 on online marketplaces at the end of 2024, then around $2 a year later. A data center that bought its cards on credit absorbs the drop without a safety net. A developer renting compute faces the rise in the same way. And unlike wheat or gas, this hour of compute cannot be stored.

The traditional market has already sniffed out the opportunity. The CME Group announced in August futures on the monthly rental of H100 and B200 GPUs, listed on the NYMEX and set for October 5. Expiry contracts, with deadlines, in the old-fashioned way.

What the Hyperliquid Policy Center is Asking the CFTC

This is exactly where the HPC's letter comes in. Expiry contracts, it essentially says, only cover part of the need. A data center rents its machines hour by hour, an inference platform pays for its compute daily, and neither of these exposures has an end date. A perpetual contract, which never expires and is maintained by a financing mechanism between buyers and sellers, fits this profile without forcing the hedger to "roll" their position from month to month.

The HPC is asking the regulator for six things. Derivatives on the two forms under which compute is sold, the GPU-hour and the inference token (the unit of text produced by a model). Clear principles for benchmark indices, with explicit support for the standard proposed by Douro Labs, the main contributor to the Pyth oracle. The right for multiple competing indices to coexist rather than a single benchmark. Position limits and enhanced oversight during the initial months. And above all, the examination of perpetuals on compute on a case-by-case basis, according to the procedure already used for crypto perpetuals since the policy declaration of May 2026.

This last point is the crux. At the end of May, the CFTC approved the first perpetual contract on bitcoin listed on a regulated U.S. market and subsequently published a doctrine that allows this type of product through individual examination. Hyperliquid wants this door, opened for cryptos, to also serve raw materials. The energy file in August was the first trial. Compute is the second, and it arrives in a space where Washington is eager to say yes: the U.S. government has been repeating for months that the AI race is won with deep markets built on its soil.

The letter also pushes a much more crypto argument. Part of the compute market is already operating on-chain: stablecoin loans secured by GPUs, tokenized inference credits, inference networks set on blockchain. If software agents buy compute continuously without human intervention tomorrow, they will need open markets operating continuously and governed by code. This closely resembles the description of a Hyperliquid order book.

The timing is not trivial for the exchange. The CME, which is launching its own compute products, has sued the CFTC for allowing Kalshi and Coinbase's perpetuals. Between the derivatives giant and the regulator, Hyperliquid chooses its side and puts its candidacy forward to provide the infrastructure. Comments are now on the table. The Commission has not set a date to decide, but its chairman promised in August to provide initial "road rules" for U.S. compute markets.

This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.