Companies building artificial intelligence applications frequently rent computing hardware rather than purchasing expensive graphics processing units, or GPUs. While lower rental prices make AI software cheaper to operate, they create financial pressure for infrastructure owners who financed roomfuls of equipment assuming higher hourly rates.
To combat this risk, financial contracts known as AI compute derivatives are stepping in to let businesses trade their exposure to computing prices separately from the physical rental of the computers. Luxor, a company known for providing services and financial products to Bitcoin miners, has expanded into the AI sector by brokering agreements between owners of computing capacity and users.
However, the cash-settled derivatives market remains in its early stages. Luxor noted that a liquid market has not yet formed, and current trading volumes or specific customer hedge examples are not yet available. Developing this market requires persuading counterparties to accept losses that other businesses seek to avoid, while relying on dependable price benchmarks and reliable payment parties.
Traditional methods for stabilizing rental income involve signing customers to longer-term contracts at fixed prices. However, because customers cannot always predict their future computing needs and operators often prefer flexibility, cash-settled derivatives offer an alternative. These contracts pay out based on a price formula without requiring the physical exchange of computing capacity, allowing operators to offset movements in market rental rates while continuing to rent out their GPUs.
Major institutions are also exploring this space. CME Group has announced plans for exchange-traded H100 and B200 rental-index futures tied to Silicon Data benchmarks, subject to regulatory review. Despite these developments, challenges remain regarding basis risk—the mismatch between the benchmark price and the actual rates an operator receives—as well as variations in equipment types, interruptibility, and commitment lengths.
Furthermore, reliance on a counterparty introduces credit and cash-flow risks. If rental prices fall across the board, a counterparty heavily invested in AI infrastructure may face financial distress simultaneously, affecting their ability to pay. While collateral requirements and standardized benchmarks can help mitigate these issues, building a functional, liquid market requires balancing tailored contracts with broad trading participation.
