TechFlow Logo
Login/ Sign up
ETH Gas
Gwei
Fear
gas
CME Group Bets on Compute Futures, BlackRock CEO: This Is the Next Trillion-Dollar Asset

CME Group Bets on Compute Futures, BlackRock CEO: This Is the Next Trillion-Dollar Asset

2026.08.17
Share
For compute futures to work, the two problems of concentration and interchangeability must be solved first.

Author: Chamath Palihapitiya

Compiled by: TechFlow

TechFlow Editor's Note: AI capital expenditure has surpassed oil and gas for the first time, but compute prices are highly volatile, and the market lacks almost any hedging tools. The CME Group plans to launch compute futures, behind which lies a key experiment on whether compute can become the next trillion-dollar asset class. This article points out that for compute futures to work, two issues must be addressed first: concentration and fungibility. These are risk variables that all investors deploying AI infrastructure must understand.

"I actually believe a new asset class will emerge, which is buying compute futures. We simply don't have enough compute right now." — BlackRock CEO Larry Fink

This week, he was proven right.

CME Group, the world's leading derivatives market, jointly announced with industry leader Silicon Data in the field of GPU market intelligence and benchmarking that it plans to launch compute futures contracts on October 5, 2026, pending regulatory review.

Why Does Compute Need a Financial Market?

In 2026, AI capital expenditure reached $765 billion, exceeding oil and gas's $681 billion for the first time. By 2031, it is expected to nearly double. Morgan Stanley estimates that the proliferation of AI in the global economy will create a $40 trillion opportunity. And this opportunity relies on a key resource: compute.

Silicon Data's index shows that since the beginning of this year, demand for compute has surged sharply, even for older generations of GPUs:

When so much capital floods into an industry, spenders need a way to protect themselves from adverse price fluctuations.

Today, producers selling oil can buy futures contracts to lock in prices before delivery. If spot prices fall, the contracts keep revenue stable. Buyers on the other side use the same market to cap fuel costs. Both parties strip price volatility from their businesses.

Compute does not yet have such tools, leaving anyone building or buying AI infrastructure facing three risk exposures:

GPU lease prices fluctuate violently, soaring during demand surges, or plummeting when supply is loose or new chips are released, making it impossible for AI companies to budget accurately for their largest cost item.

Whenever NVIDIA releases faster chips, the lease value of the previous generation chips declines, and the collateral behind hardware loans shrinks accordingly.

A data center takes two to three years to build, but developers have very limited means to lock in their compute costs or revenues. Every such decision is a bet of billions of dollars.

These risk exposures create demand for compute futures. But before this market can scale, it must face the same two issues that once limited other futures markets: concentration and fungibility.

Attempts to build futures markets around onions, uranium, DRAM memory chips, and bandwidth have encountered one or both of these issues.

For compute, the concentration issue is more complex. Buyers are becoming dispersed, as inference demand is spread across thousands of companies running production workloads. The seller camp is broad and still growing, with revenue from new cloud providers exceeding $25 billion in 2025, covering over 60 providers. But its underlying layer remains highly concentrated, with NVIDIA supplying most AI chips.

The second issue is fungibility.

Today, compute prices are quoted in GPU hours, i.e., the cost to rent a single GPU for one hour. But two GPUs of the same model may provide different compute within one hour.

Silicon Data, together with academic partners, ran the same workloads on 3,500 GPUs across 11 cloud providers. Even within the same chip model, they found significant differences. In one test, H100 performance varied by up to 34.5%, and the largest gap in the entire study reached 38%.

The first sustainable contracts may need to define several grades, just as energy markets use different fuels, locations, and delivery periods.

But the bigger question is, what happens if compute futures work. Can compute become the next asset class with a notional trading volume of trillions of dollars, and accelerate the AI economy?