
The CFTC-regulated exchange said its new curves derive implied future rental prices for Nvidia B200, H200 and A100 GPUs from event-contract trading out to one year.
Kalshi has launched compute forward curves that let traders and businesses track implied future prices for renting Nvidia GPU capacity up to a year ahead, creating a market-based reference for one of AI's most important variable costs. The CFTC-regulated prediction market exchange said the curves are derived from trading in its event contracts and cover Nvidia's B200, H200 and A100 chips. The move gives AI companies a potential hedging tool for future compute needs while offering investors a new way to trade views on supply, demand and pricing in the GPU rental market. The launch also comes as CME Group said on May 12 that it planned compute futures with Silicon Data, signaling broader efforts to turn GPU rental prices into a standalone tradable asset class. Kalshi's compute contracts do not use crypto tokens, but the development may still draw attention in digital-asset markets because the exchange launched tokenized news prediction markets on Solana in December 2025 and already supports markets tied to Bitcoin and Ethereum.