Wall Street's push to trade AI compute hits Washington roadblock

Wall Street wants a price for AI compute. Washington may not.

Exchange operators are racing to make GPU compute a tradable commodity like oil or wheat. But that effort is complicated by the fact that the biggest buyers, AI hyperscalers, pay under secret contracts, leaving most of the public data on how much people are paying for compute coming from the spotty offerings from neoclouds.

Intercontinental Exchange CEO Jeff Sprecher told me when I interviewed him this week at the futures industry’s annual confab in Chicago that he was “bullish but cautious.” Futures normally follow an established spot market, and compute doesn’t really have one (see this robust X debate ). “We’re kind of [working] backwards,” Sprecher said.

The AI debt boom might just bring the push the markets need. Sprecher said lenders that are increasingly nervous about financing data centers could start demanding that companies lock in their GPU prices for as long as they’re asking to borrow. “It may be that the capital markets require that, given how much money is being lent,” he said. “And then Zuckerberg comes over and says, ‘I need to do a big trade,’” he added with a smile.

The bigger wild card is Washington. Pressure from the Trump administration forced Kalshi to scrap one of its GPU price trackers in September, Semafor reported, which Sprecher chalked up to concerns that transparent pricing for AI could disadvantage the US in its competition with China. “The United States is a free market and that has always advantaged us vis-à-vis China, so I suspect that’s where they’ll come out. But we’re all waiting with bated breath.”