Kalshi develops a forward curve for computing power as exchanges compete to transform GPUs into a tradable asset.

      Kalshi has created a forward curve for GPU computing costs utilizing prediction market contracts, joining CME and ICE in the financialisation of AI infrastructure. The prediction markets exchange has developed a forward curve that monitors the anticipated pricing of computing power, becoming part of the increasing number of exchanges and index operators aiming to standardize GPU rental costs into a financial instrument. This tool employs weekly and monthly event contracts concerning compute prices, extending up to a year ahead. An algorithm integrates these contracts into a single curve that can be referenced for futures, options, and other derivatives.

      Udesh Jha, Kalshi’s chief risk officer, stated to Bloomberg, “We are employing prediction markets to create the forward curve, providing the market insight into the future compute costs for various grades and time-frames of GPUs.” Forward curves are fundamental in commodity markets, used to illustrate anticipated future prices for products such as crude oil, natural gas, and interest rates. The emergence of a forward curve for GPU rental costs indicates significant progress in treating compute as a commodity.

      Kalshi is not alone in this push for compute. In May, CME Group introduced compute futures in collaboration with Silicon Data to develop contracts based on an index monitoring the hourly expenses of renting high-performance GPUs. Shortly after, Intercontinental Exchange announced a partnership with Ornn to initiate its own cash-settled compute futures, marking at least three major participants striving to create a benchmark contract for AI computing power.

      Kalshi's methodology stands out from its larger competitors in one crucial way. While CME and ICE are constructing traditional futures contracts that necessitate regulatory approval, Kalshi is leveraging its existing prediction market structure to form the curve from already traded event contracts. Jha highlighted this approach as a significant enabler for hedging, risk management, and speculative endeavors.

      The driving force behind all three initiatives is consistent. AI infrastructure expenditures are expected to reach trillions of dollars over the next decade, and businesses engaged in GPU capacity transactions currently lack a standardized method to hedge against price fluctuations. The rental rates for GPUs have been unpredictable, and the computing market remains splintered among cloud providers, data center operators, and GPU brokers, who each set prices through bilateral agreements lacking transparency.

      A viable forward curve provides both buyers and sellers with a common perspective on future pricing, serving as the foundation for hedging and risk management strategies. The ultimate distribution of liquidity among Kalshi, CME, or ICE will determine which curve becomes the industry standard, similar to how oil contracts eventually settled into the Brent and WTI duopoly that currently defines energy markets. For an asset class that barely existed two years ago, the financial infrastructure is developing at an impressive pace.