Jump Trading, DRW Forge Path for Tradable GPU Compute Market

Two prominent proprietary trading firms, Jump Trading and DRW, are developing financial contracts and infrastructure to create a tradable commodity market for GPU computing power, aiming to bring price discovery, hedging, and trading capabilities to AI-driven compute resources.

Jump Trading has established a dedicated compute trading desk by hiring the founding team of Itô, a firm specializing in compute and inference trading. This strategic move allows Jump Trading to focus on bilateral contracts, which are private, two-party agreements. These deals typically range from three to five years in duration and are valued in the tens to low hundreds of millions of dollars, catering to large, sophisticated counterparties.

DRW has adopted a broader strategy, building out the foundational infrastructure for a GPU compute market. This includes incubating Compute Exchange, a marketplace for GPU capacity, and Silicon Data, a firm that provides pricing benchmarks and indices for GPU compute rental. In August 2026, DRW entered into a GPU-as-a-Service agreement with QumulusAI for a dedicated Nvidia Blackwell B300 cluster. This contract has an initial term of one year with options to renew for up to four additional years.

CME Group is planning to launch futures contracts based on Silicon Data’s indices for H100 and B200 GPU rentals, with each planned contract representing 730 GPU-hours. As of mid-October 2026, these futures contracts were awaiting approval from the Commodity Futures Trading Commission (CFTC). Regulatory approval could transform the market by introducing standardized, exchange-listed products, potentially increasing accessibility beyond large counterparties involved in bilateral deals. ICE and Nodal are also reportedly developing futures or working with indices for GPU compute.

The development of these financial instruments is driven by the need for risk management in the rapidly growing AI sector. Companies relying on GPUs face significant price volatility for rental services. The ability to lock in costs through hedging or sell future capacity at a known price offers a crucial risk management tool for both buyers and data center operators.

However, the evolving nature of GPU hardware presents challenges. Newer generations of chips are released frequently, which could impact the long-term relevance of indices and contracts based on specific hardware models. A long-term bilateral contract, for instance, could potentially outlast the relevance of the chip it was originally priced around. Furthermore, when a reference price underpins futures contracts, questions regarding the calculation methodology and potential conflicts of interest for index providers can arise.

As of mid-October 2026, the CFTC’s approval remains pending for CME Group’s planned futures contracts. The market is currently characterized by bilateral forward and structured contracts, with spot indices tracking current rental prices. The potential launch of exchange-traded futures represents the next layer of market development.

Why This Matters

The materials describe a narrow update: Jump Trading has established a dedicated compute trading desk by hiring the founding team of Itô, focusing on bilateral contracts. Whether the CFTC will approve CME Group’s planned futures contracts.

Broader Context

Source materials place the factual news in this context: The AI boom runs on GPUs.