Compute Exchange: Carmen Li is building a GPU compute market bigger than the oil market

In a new episode of Bloomberg's Odd Lots podcast, Carmen Li, co-founder and CEO of Compute Exchange and Silicon Data, told hosts Joe Weisenthal and Tracy Alloway how her team is trying to build a market for computing power (compute) that could grow bigger than the oil market, and what it takes to create a GPU index.

Why compute is becoming a commodity like oil

The generative AI boom has sharply driven up demand for scarce GPUs to train and run inference on large models. Data center compute has become a scarce resource that AI labs, cloud providers, and startups all compete for, much as companies have historically competed for supplies of oil and other commodities. The GPU shortage has been one of the AI industry's defining stories in recent years: the largest cloud providers spend tens of billions of dollars a year on data centers and Nvidia chips, and Nvidia itself has become one of the world's most valuable public companies, largely on the back of that demand.

Compute Exchange and similar projects want to turn this scarcity into a tradable financial instrument: indexes and, eventually, futures on the price and availability of GPU hours.

How the Compute Exchange project works

• The episode's guest is Carmen Li, co-founder and CEO of Compute Exchange and Silicon Data.
• The Odd Lots podcast is hosted by Bloomberg staff journalists Joe Weisenthal and Tracy Alloway.
• The project's goal is to build a compute market that could eventually exceed the oil market in size.
• Its key instrument is a GPU index that tracks the price and availability of computing power on the market.

Why investors and AI companies would want such a market

A formal compute market would give AI companies a way to hedge against sudden spikes in GPU rental prices, much as airlines hedge fuel costs with oil futures. For investors, a compute index would be a new asset class tied directly to the AI industry's growth rate rather than to individual chipmaker or cloud provider stocks. Turning access to this capacity into a tradable asset is a logical next step in the financialization of AI infrastructure, a trend that in recent years has already produced specialized GPU cloud providers and intermediary marketplaces for renting compute.

Odd Lots is one of Bloomberg's best-known finance podcasts, where hosts Joe Weisenthal and Tracy Alloway regularly walk guests through how new or little-understood markets work, from commodity derivatives to complex financial instruments. The fact that the show devoted an episode to the compute market shows that the topic has moved beyond a narrow circle of AI infrastructure specialists and now interests a broad financial audience.

What it means

The emergence of tradable indexes, and eventually futures, on GPU capacity signals that compute is completing its shift from a technical data center metric into a standalone financial market, with new infrastructure, from exchanges to derivatives, forming around it.

If such a market takes hold, it will also change how AI companies plan spending: instead of signing long-term contracts with a single cloud provider, they will be able to buy GPU capacity on the open market and respond to price swings much the way industrial companies manage energy and raw material purchases today.