Industry

AI-generated text

Exchanges Race to Price GPU Compute, but Secrecy and Washington Could Stall Markets

Exchange operators are attempting to turn GPU-based AI compute into a tradable commodity, but a lack of transparent spot pricing and political pressure from Washington present obstacles.

Exchanges Race to Price GPU Compute, but Secrecy and Washington Could Stall Markets

Exchange operators are trying to create a market where GPU-based AI compute can be traded like traditional commodities such as oil or wheat. That ambition faces a practical hurdle: the largest purchasers — AI hyperscalers — buy under confidential contracts, so most publicly available information about compute pricing comes from sporadic disclosures by smaller "neocloud" providers.

Intercontinental Exchange's view

Jeff Sprecher, chief executive of Intercontinental Exchange (ICE), told an interviewer at the futures industry’s annual conference in Chicago that he was “bullish but cautious.” He noted that futures markets typically develop after a well-established spot market exists, and that compute lacks a clear spot market, meaning exchanges are in a sense “working backwards.”

AI debt and financing could accelerate markets

Sprecher suggested that the surge in AI-related borrowing might force the development of tradable compute prices. Lenders growing uneasy about financing data centers could require companies to lock in GPU prices for the duration of loans. Sprecher speculated that this kind of capital-market pressure might prompt large trades from major industry players.

Washington as a wildcard

Political factors add uncertainty. Semafor reported that pressure from the Trump administration led Kalshi to scrap one of its GPU price trackers in September, an example of how government concern can reduce pricing transparency. Sprecher attributed such moves to worries that transparent AI pricing might disadvantage the United States in competition with China. He observed that the U.S. free-market system has historically been an advantage versus China, and suggested that policy decisions will likely reflect that tension — while market participants wait to see the outcome.

Why this matters

If a liquid market for GPU compute develops, it could reshape how AI projects are financed, how companies hedge price risk, and how capital is allocated across the industry. At the same time, confidential procurement practices among the largest buyers and potential political constraints from policymakers in Washington could limit the emergence of a transparent, liquid market.

(This article is based on reporting by Liz Hoffman, including an interview with Jeff Sprecher at the futures industry's annual conference in Chicago.)