CME and Silicon Data team up to turn GPU compute into a hedgeable asset

CME Group has partnered with Silicon Data to launch what they describe as the first dedicated compute futures market, planned for later this year pending regulatory approval. The contracts will be financially settled instruments referencing Silicon Data’s daily GPU rental rate indices, including benchmarks such as prices for renting NVIDIA H100-class capacity, rather than delivering physical hardware. The products are aimed at traders, financial institutions, AI developers, and cloud-service providers that need to manage sharp swings in the cost of on-demand GPU power in the rapidly expanding AI infrastructure economy.

The partnership scope is tightly centered on market infrastructure and pricing data: CME contributes the derivatives venue, clearing, and rulebook, while Silicon Data provides the reference indices that standardize GPU rental prices across providers, regions, and contract terms. Strategically, both firms present the initiative as a way to transform AI compute from an opaque operational expense into a transparent, benchmarked financial asset class, bringing tools for hedging, price discovery, and long-term planning that resemble those in oil, gas, and other commodities. For Silicon Data, the deal elevates its GPU benchmarks and forward curves as the de facto reference for institutional risk management around compute, while CME gains a first-mover position in a multi-trillion-dollar compute market that could diversify its volumes and revenue as demand for AI data centers accelerates. The collaboration also complements Silicon Data’s existing physical and forward contracting platforms, such as Compute Exchange, by cleanly separating financial hedging at CME from physical GPU procurement and delivery handled through Silicon Data-linked channels.