Kalshi Extends Its Exchange Model to GPU Pricing

Kalshi has launched GPU forward curves, extending its exchange infrastructure beyond prediction markets into AI compute. The new benchmarks are designed to help price future GPU capacity and could serve as a reference point for over-the-counter compute transactions and other institutional risk-management products.

The curves cover Nvidia B200, H200 and A100 chips, providing forward estimates for the hourly cost of GPU capacity weeks and months ahead. GPU capacity has become an increasingly important input for AI developers, cloud providers and data center operators, with demand often moving faster than available supply.

Kalshi argues the market lacks transparent forward pricing, making it difficult for participants to estimate future compute costs or structure long-term agreements. "Compute is the new oil. Like every commodity before it, it needs a real derivatives market," said Tarek Mansour, CEO of Kalshi. "Kalshi intends to be the exchange where all future buyers and sellers manage their risk."

According to the company, the benchmarks are intended for participants looking to price or hedge future GPU demand, including AI labs, cloud providers and data center operators. The forward curves themselves are not tradable products; Kalshi says the curves can be used to price OTC compute deals, with direct exposure available through the underlying GPU contracts and block trades.

Kalshi is not the only exchange moving in this direction. CME Group said in May it would launch computing-power futures linked to an index developed by Silicon Data, while Intercontinental Exchange is working with financial infrastructure firm Ornn on futures contracts tied to GPU capacity. Together, the projects suggest exchange operators increasingly see AI compute as a market that could support standardized pricing and risk-transfer tools.

For brokers, exchanges and infrastructure providers, the announcement highlights another potential application for exchange technology. Stable reference prices, standardized contracts and block trading are familiar tools in commodity and derivatives markets. Kalshi is now applying the same market structure to GPU capacity. Whether compute develops into a liquid derivatives market remains uncertain, but the launch shows how exchange infrastructure built for event contracts can be adapted to entirely different types of underlying assets.