Seen as an important attempt to incorporate AI computing power into financial markets and create a new category of investable assets, the Nvidia GPU rental price futures planned by the CME Group face delays due to the extension of the review by U.S. regulators.

The CME’s original plan was to launch futures contracts in early October that would be tied to the rental prices of Nvidia H100 and B200 GPUs. However, according to a report by The Information, the U.S. Commodity Futures Trading Commission (CFTC) has extended the review deadline by 45 days to November 9, meaning the related contracts cannot be listed on schedule in October as originally planned. The CFTC said this case involves “novel or complex” issues that require further research into how the contracts would operate, as well as the potential risk of market manipulation.

The core purpose of such computing power futures is to enable AI companies, GPU compute providers, and financial institutions to hedge fluctuations in GPU lease prices. With AI model training and inference demand growing rapidly, GPU compute leasing has gradually become a large-scale market. If futures contracts can be launched smoothly, participants could have an opportunity to manage compute costs through financial instruments, and further establish a price discovery mechanism for AI compute power.

However, the CFTC believes that the GPU compute market is still highly fragmented and lacks transparency. A large portion of current GPU leasing prices comes from private contracts, and price differences may arise from varying suppliers, GPU models, leasing periods, and service terms. As a result, it is not easy to build a reliable and representative pricing benchmark. This also requires regulators to further evaluate whether futures prices can accurately reflect the spot compute market, and whether the contracts are susceptible to manipulation.

The CFTC said that the financialized computing power market could help support the development of the U.S. AI industry. However, because the underlying market is still in an early stage, regulators want to hear further from industry stakeholders before approving the related products. This extension of the review period also suggests that the process of AI compute power moving from purely infrastructure services to financial assets still faces regulatory issues such as price transparency, benchmark establishment, and market integrity.

If the CME receives final approval, H100 and B200 GPU leasing price futures will become an important attempt at linking AI infrastructure with traditional financial markets—allowing GPUs not only to serve as capital expenditures for AI companies, but potentially to gradually form a financial market that can be traded, hedged, and priced.

Source