Compute

CME, Silicon Data to Launch Compute Futures

Futures tied to daily GPU rental benchmarks aim to let firms hedge compute price risk.

Futures tied to daily GPU rental benchmarks aim to let firms hedge compute price risk.

A dimly lit room features a desk with three monitors displaying charts in front of glass-enclosed racks of servers and graphics cards. © The GPU Trade Inc 2026


CME Group and Silicon Data announced on May 12, 2026 that they will launch a compute futures market later this year, subject to regulatory review.

The planned contracts will use Silicon Data’s indices — which the companies describe as the world’s first daily GPU benchmarks for on‑demand rental rates — as the price reference.

CME framed the move as a way to bring transparency, liquidity and risk management to what the partners call a multi‑trillion‑dollar compute market. The announcement names traders, financial institutions, AI builders and cloud providers as intended users.

CME Chairman and CEO Terry Duffy said compute is “the new oil of the 21st century” and stressed the exchange’s role in providing market structure and trusted clearing. That quote is in the companies’ joint release.

Silicon Data CEO Carmen Li said the firm built its benchmarks to standardize pricing across a fragmented compute market and to give AI teams and cloud operators tools for valuation and hedging. The startup is backed by trading firm DRW, the release noted.

Don Wilson, founder and CEO of DRW, called the futures a potential solution to the lack of hedging vehicles for rising data center and GPU spending. DRW’s involvement is highlighted in the announcement.

Under the plan, contracts would settle against Silicon Data’s daily rental‑rate indices so buyers and sellers could lock in future compute prices or create hedges against sudden spikes. CME said the products will come later in 2026 after required rule filings and regulator sign‑offs.

Industry executives and analysts have recently forecast massive, multi‑year increases in AI infrastructure spending — language that helped set the stage for this product. BlackRock CEO Larry Fink and others have publicly argued that compute could become a tradable asset class.

Proponents say a regulated futures market could standardize pricing across providers, reduce planning risk for hyperscalers and give enterprises ways to hedge operating costs. White papers and market analyses published this year project large, multi‑trillion dollar buildouts of data‑center and AI infrastructure.

Skeptics and market commentators warn new contracts often start with thin liquidity and that compute pricing varies by region, contract terms and hardware generation — complications that could slow adoption. Early trading volumes and open interest will be watched closely once contracts list.

Operationally, CME brings clearing, margining and exchange design; Silicon Data supplies the price feeds and indices. Regulators will evaluate whether the contracts meet standards for market integrity and customer protection before trading can begin.

If approved, the launch would mark a notable step in the financialization of compute — shifting some planning risk from operators to capital markets and creating new tools for hedging the costs of running AI models. Market participants now await the formal rule filings and the timetable for a launch later this year.