ICE's Sprecher "bullish but cautious" on GPU compute futures

The push to trade AI compute is running into an awkward sequencing problem. Futures contracts are typically priced off an established spot market, but compute doesn't really have one — the largest buyers, AI hyperscalers, pay under secret contracts, and most public data comes from the spotty offerings of neoclouds. That is why Jeff Sprecher, speaking to Semafor at the futures industry's annual confab in Chicago, described the effort as working backwards.

Two forces could change that. Sprecher suggested lenders increasingly nervous about financing data centers might start demanding that companies lock in their GPU prices for as long as they are asking to borrow, with the capital markets potentially requiring it given how much money is being lent. He also floated the idea, with a smile, of Zuckerberg arriving to do "a big trade."

The other variable sits outside the market itself. Pressure from the Trump administration forced Kalshi to scrap one of its GPU price trackers in September, which Sprecher chalked up to concerns that transparent pricing for AI could disadvantage the US in its competition with China. He said he suspects the US will land on the side of free markets, which he said has always advantaged it vis-à-vis China — but added that everyone is waiting with bated breath.