Compute Perpetuals: GPU Rental Price Indices and Cash Settlement
Summary
The document introduces compute perpetuals as cash-settled derivatives tied to indices of GPU rental prices rather than ownership or delivery of hardware. It describes H100 and B200 contracts linked to hourly rental price benchmarks compiled from supplier quotes and standardized across factors such as region and rental terms. The contracts let traders express views on the cost of high-end computing and the balance of AI compute supply and demand.
The article contrasts these products with crypto futures and conventional commodity markets, noting their USDT settlement and continuous trading. It also identifies possible price drivers, including AI demand, chip supply, and electricity costs, and distinguishes this exposure from owning shares in a GPU manufacturer. The piece is a product overview, not an independent analysis of index methodology, liquidity, tracking error, or historical performance. It notes that leverage increases risk and that the contracts convey price exposure without physical delivery.
Key ideas
- Compute perpetuals provide price exposure to GPU rental costs without transferring physical hardware.
- The H100 and B200 benchmarks aim to represent standardized hourly rental prices using quotes from multiple suppliers.
- These contracts settle profit and loss in USDT and are described as trading continuously.
- AI demand, chip supply, and electricity costs are among the stated influences on compute pricing.
- Compute price exposure may differ from the performance of shares in a GPU maker, and leverage adds risk.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.