Via tripadvisor.com
The new GPU rental futures contracts will let AI companies and investors hedge against volatile compute costs on NYMEX
CME Group is bringing Wall Street’s oldest trick to Silicon Valley’s newest problem. The exchange giant plans to launch two compute futures contracts on October 5, 2026, giving AI developers, cloud providers, and institutional investors a way to hedge against the wildly unpredictable cost of renting the GPUs that power modern artificial intelligence.
The contracts, developed in partnership with Silicon Data, will track hourly rental rates for Nvidia’s H100 and B200 chips.
What the contracts actually look like #
The two products are the Silicon Data H100 Rental Index Futures and the Silicon Data B200 Rental Index Futures. Each contract represents one month’s rental of a single GPU from the respective Nvidia chip family.
They’ll be listed on the New York Mercantile Exchange, the same venue where energy and metals futures have traded for decades. The launch is pending regulatory review.
CME Group and Silicon Data first announced their partnership on May 12, 2026. Silicon Data, which is backed by trading firm DRW, provides the underlying rental rate indexes that the futures will track.
The H100 contract targets what is currently the workhorse GPU of the AI industry. Nvidia’s H100, part of its Hopper architecture, has been the chip of choice for training and running large language models since its release. The B200 contract, meanwhile, tracks Nvidia’s next-generation Blackwell chip.
Why GPU rental costs need a hedge #
Anyone who has tried to rent GPU compute over the past three years knows the pricing can be chaotic. Demand for AI training and inference capacity has surged as companies race to deploy large language models, image generators, and autonomous systems. Supply, meanwhile, has been constrained by chip shortages, export restrictions, and the sheer difficulty of building out data center capacity fast enough.
The target audience is broad: AI startups burning through GPU hours on training runs, cloud service providers managing capacity for enterprise clients, hyperscalers like Amazon and Microsoft that operate massive GPU fleets, and institutional investors looking for exposure to the AI infrastructure boom without buying Nvidia stock directly.
The bigger picture for compute as an asset class #
CME Group’s Pete Keavey, Global Head of Energy and Environmental Products, emphasized the importance of compute capacity in the current AI landscape, likening it to a currency in the AI era.
There’s also a competitive dimension. Decentralized compute networks like Akash and Render have been trying to create marketplaces for GPU resources using blockchain-based systems. CME’s entry into the space brings institutional-grade infrastructure, clearing, and regulatory oversight to a market segment that has, until now, mostly operated through bilateral deals and spot pricing on platforms with limited transparency.
The October 5 launch date, assuming regulatory approval comes through, will mark the first time a major regulated exchange has offered derivatives directly tied to GPU compute pricing.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our