Silicon Data raises $30.5M to price the GPU futures market
The AI buildout's financial layer is taking shape: the firm tapped to set the reference price for CME Group's planned GPU futures market just closed a $30.5 million Series A.
Silicon Data, the New York startup building independent pricing and performance benchmarks for AI compute, announced a $30.5 million initial close of its Series A led by the Valor Atreides AI Fund, with CME Ventures, DRW, F-Prime, Samsung Next, VanEck, Further, Jump, Tectonic and Wintermute also participating. The round is the clearest sign yet that GPU compute is being treated as a commodity with a derivatives market of its own: CME plans to use Silicon Data's benchmarks as the reference price for its cash-settled GPU futures (pending regulatory approval), letting buyers and sellers hedge swings in rental prices against a published, daily index. CEO Carmen Li frames the mission as building "independent referees" for compute — benchmarks that say what something is worth, whether it performs as promised, and how risk should be measured.
The money funds four areas: benchmark pricing, performance measurement through SiliconMark, institutional market data, and risk infrastructure for derivatives, insurance and credit markets. SiliconMark is the subtle piece — it independently benchmarks how physical GPU infrastructure actually performs, on the observation that two clusters built on identical chips can deliver meaningfully different real-world output depending on networking, topology and configuration. Normalizing that performance is what makes hedging output risk possible and, the company argues, paves the way for physical delivery of compute down the line. Silicon Data already publishes nine financial-grade indices covering GPUs and LLMs and says it counts more than 1,000 registered users among semiconductor makers, model builders and financial institutions, up from a $4.7 million seed round in March 2025.
The significance here is the pattern, not the check size. This is the week finance officially started underwriting the buildout — Bank of America pledged $250 billion to data center and energy infrastructure, and CoreWeave's backlog crossed $104 billion — and the missing piece has always been price discovery. "Compute is now the seed and equipment of the AI economy, being built out at national-infrastructure scale with none of that machinery underneath it," said Gavin Baker, CIO of Atreides Management, whose Valor Atreides AI Fund led the round. If CME's GPU futures launch, the people who measure compute will be as essential to the market as the people who build it — the index business of AI, taking shape before the commodity even has a liquid price.
What to watch: whether regulators approve CME's GPU futures, and whether other exchanges build their own compute derivatives on the same benchmarks.
If GPU futures go live, does a transparent price for compute change how AI infrastructure gets financed? Tell us in the comments.
Sources: Silicon Data · CME Group · FinSMEs · Techmeme
