CME Group Inc. (NASDAQ:CME) said Tuesday it will launch AI compute futures on Oct. 5, pending regulatory review, with contracts tracking monthly rental prices for NVIDIA Corp. (NASDAQ:NVDA) H100 and B200 chips.
The exchange wants to turn compute into a “standardized tradable commodity,” giving AI labs, hyperscalers and data center operators a way to hedge swings in computing costs.
The plan rests on two awkward questions. Is compute uniform enough to trade like a commodity? And can a real market form around a product dominated by one company?
Compute Isn’t Oil
Compute is not naturally interchangeable. Chips age quickly, generations differ and the economics shift with power, networking and location.
Semafor business reporter Rohan Goswami, speaking Thursday on the Prof G Markets podcast, said exchanges have handled that problem before. Not all steel is identical either, yet hot-rolled steel trades as a single benchmark.
CME’s answer is similar. Its initial contracts will benchmark specific chips, Nvidia’s H100 and B200.
Kalshi already trades a range of GPU rental-price markets, including one on where Nvidia B200 compute ends the year. Those markets underpin its market-implied compute forward curves.
The Bigger Problem Is Nvidia
Standardization may be the easy part. Goswami said compute futures may never function like a conventional commodity market because Nvidia has enormous influence over the supply and pricing of the chips underlying the market.
“There’s no other game in town except for Nvidia,” he said, likening the setup to an oil market run entirely by Standard Oil.
Goswami argued Nvidia has little incentive to encourage greater price transparency or falling compute costs, since cheaper compute could reduce the value of its chips. He added he does not believe compute futures will ever become a “real market.”
Wall Street Is Financializing the AI Buildout
The contracts fit a broader push to price, hedge and finance the AI boom. Nvidia said this week that Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR will establish financing platforms intended to mobilize more than $500 billion for AI infrastructure.
Goswami called the initiative a risk-spreading mechanism, one that spreads the financial risk while giving Wall Street greater exposure to the AI buildout.
The Oct. 5 launch will test whether a tradable price for Nvidia compute can mature into a genuine commodity market.
Image: Shutterstock
Login to comment