CME Group and Silicon Data Plan New Futures Market for Semiconductor Compute

CME Group and Silicon Data announced Tuesday a plan to create a futures market for semiconductor computing capacity. The new contracts will be based on graphics processing unit price indexes developed by Silicon Data. The joint venture is still pending regulatory review.

The market is intended to provide standardized pricing for computing power that has become increasingly expensive amid artificial intelligence expansion. Contracts can be used to lock in prices for GPU-based capacity and to hedge against rising rental rates and other operational costs associated with large-scale AI infrastructure projects.

"GPU markets ... have historically lacked standardized reference pricing," Carmen Li, chief executive of Silicon Data, said in the release. " Silicon Data provides specialized price indexes for semiconductors, including a standardized GPU price index, a RAM index and projections for GPU rental prices.

These indexes function similarly to broader economic measures such as the consumer price index.

The announcement referenced past attempts to create futures markets for technology infrastructure. During the broadband expansion in the late 1990s, the broadband services division of Enron sought to sell unused capacity on its fiber optic network before the company's failure.

GPUs and central processing units is not expected to decline soon. Memory chip prices increased in the first quarter as artificial intelligence drove higher demand for both CPUs and GPUs. Hyperscalers raised capital spending while noting concerns about memory supply constraints that have pushed input costs higher.

Memory chip makers are projecting strong profit margins for this year and next. One analysis stated that agentic AI systems will require additional racks of CPU servers operating alongside GPU infrastructure for orchestration, data processing and tool execution.