The next AI computing futures market

Unlock Editor’s Digest for free

Roula Khalaf, editor-in-chief of the FT, selects her favorite stories in this weekly newsletter.

The writer is a former hedge fund manager and author of the Net Interest newsletter.

You can trade all sorts of things on the Chicago Mercantile Exchange: soybeans, precious metals, lean pigs, crude oil, interest rates – all the commodities that support the global economy. Now, CME Group plans to introduce a new product to its roster: computing, which it calls “the new currency of the AI ​​economy.”

“Just as oil fueled the 20th century economy and evolved from spot trading to a global derivatives market, our futures contracts will now transform computing into a standardized, tradable commodity,” Pete Keavey, executive head of CME, said when announcing the product in August. Others on Wall Street are equally optimistic. “A new asset class will be buying calculus futures,” Larry Fink, chief executive of BlackRock, said in May.

For an exchange, it makes sense to tackle new products, and the math is potentially large. The Boston Consulting Group estimates that the AI ​​computing market will grow from $360 billion in 2025 to about $2.3 billion in 2030. If CME can capture even a fraction of these flows, it will add a lucrative revenue stream.

But just because a market is large doesn’t mean it’s ready to trade. Research into the origins of futures markets reveals that between two-thirds and three-quarters of new contracts fail to attract and maintain a profitable level of trading volume. For IT derivatives to thrive like oil derivatives, they must meet a number of conditions.

First, the underlying prices must be volatile enough to make hedging worthwhile. Speculators may provide liquidity to a futures market, but it is commercial hedgers who give them their economic focus. On this test, the arguments in favor of calculation are quite strong. Long lead times for bringing new capacity into service can make prices very volatile.

The construction of Oracle’s Project Jupiter data center illustrates this point. The facility is expected to come online in 2028 but has faced setbacks, including delays to the gas pipeline project and legal challenges over water and air quality permits. More broadly, when chip bottlenecks were tightest in early 2024, hourly rental rates for one of Nvidia’s H100 chips soared as high as $8, before falling below $2 in late 2025. That kind of gap is something participants may want to cover.

Second, participants must agree on a standardized contract. When oil futures were publicly traded in 1983, they were designed to reference West Texas Intermediate, with Cushing, Oklahoma, chosen as the physical delivery point. Although WTI represents only a small percentage of global crude production, it serves as a pricing benchmark for the rest of the market. In contrast, the pioneers of the bandwidth business in 1999 struggled to converge on a standard. The reliability, quality and bandwidth of services varied too widely and attempts to create a market collapsed with its champion, Enron.

Working with index provider Silicon Data, CME Group chose Nvidia’s H100 and B200 chips as the benchmark for its contracts. According to Silicon Data, the capacity of a B200 can be rented for $5.86 per hour; $2.77 on the old H100. CME plans to list contracts that reflect the future value of these rental rates, for a duration of 36 months.

But even GPU hour isn’t fully standardized. Performance varies depending on cluster configuration, networking, software, and location. Rapid obsolescence further fragments the market: an H100 hour is not interchangeable with a B200 hour, which is why CME needs separate contracts for each. There is also no consensus on how to even measure these rental rates. Another index provider, Ornn, has its own benchmark, and the two don’t always align. Until a consensus forms, building a derivatives market could be slow.

Finally, a vibrant derivatives market requires a diverse group of buyers and sellers. Onion futures were explicitly banned in 1958 in the United States after two traders cornered the market. Computing is unlikely to be cornered in the same way, but it is already very focused. Nvidia dominates chip supply, hyperscalers control much of the available capacity, and a handful of AI labs account for much of the demand.

CME Group’s IT futures contracts will attract a lot of attention when they are listed. The price of computing is increasingly weighing on the AI ​​economy, the value of its infrastructure, and the vast investment boom that surrounds it. This gives investors good reason to want a benchmark. Bears may remember the ABX subprime index, which illuminated a previously opaque aspect of mortgage financing – and became a focal point of the unraveling that followed. Whether it validates the boom or breaks it, the futures calculation may ultimately matter more as a signal than as a market.