CME Bets on Compute Futures; BlackRock CEO: This Is the Next Trillion-Dollar Asset Class
- Core View: CME Group plans to launch compute futures contracts in 2026, aiming to provide price hedging tools for the AI industry—whose capital expenditure is set to exceed that of oil and gas for the first time (projected at $765 billion in 2026)—but its success hinges on resolving two major challenges: market concentration and product fungibility.
- Key Elements:
- CME Group, in partnership with Silicon Data, plans to launch compute futures contracts on October 5, 2026, pending regulatory review. This marks a critical experiment in compute's journey toward becoming a trillion-dollar asset class.
- AI capital expenditure is projected to reach $765 billion in 2026, surpassing oil and gas at $681 billion for the first time. Morgan Stanley estimates AI will create a $40 trillion economic opportunity.
- The current compute market lacks hedging tools, exposing participants to three major risks: sharp fluctuations in GPU rental prices, collateral value depreciation of older hardware following new chip releases, and multi-billion-dollar investment bets driven by long data center construction cycles.
- Concentration issue: While both buyers and sellers are relatively fragmented (new cloud providers' revenue surpassed $25 billion in 2025, covering over 60 providers), the underlying chip supply remains highly concentrated in NVIDIA.
- Fungibility challenge: Silicon Data's testing across 3,500 GPUs from 11 cloud providers found performance differences of up to 34.5% for the same model (e.g., H100), with the most extreme gap in the research reaching 38%, complicating standardized futures pricing.
- Historical precedents show that attempts to establish futures markets around onions, uranium, DRAM chips, and bandwidth all failed due to concentration or fungibility issues. Compute futures must avoid these pitfalls by defining multiple grades similar to energy markets.
Original Author: Chamath Palihapitiya
Original Translation: TechFlow
TechFlow Brief: AI capital expenditure has surpassed oil and gas for the first time, but compute prices are highly volatile, and the market has virtually no hedging tools. The CME Group's plan to launch compute futures represents a critical experiment in whether compute can become the next trillion-dollar asset class. This article points out that for compute futures to succeed, two issues—concentration and fungibility—must first be resolved, and these are risk variables that every investor building AI infrastructure must understand.
"I actually believe a new asset class will emerge—purchasing compute futures. We simply don't have enough compute right now." — Larry Fink, CEO of BlackRock
This week, he was proven right.
CME Group, the world's leading derivatives marketplace, jointly announced with Silicon Data, an industry leader in GPU market intelligence and benchmarking, its plan to launch compute futures contracts on October 5, 2026, subject to regulatory review.
Why does compute need a financial market?
In 2026, AI capital expenditure will reach $765 billion, surpassing oil and gas at $681 billion for the first time. By 2031, it is expected to nearly double. Morgan Stanley projects that AI's diffusion across the global economy will create a $40 trillion opportunity. And that opportunity depends on one critical resource: compute.
According to Silicon Data's index, since the beginning of this year, demand for compute has risen sharply even for older generations of GPUs:
When that much capital flows into an industry, the people spending it need a way to protect themselves from adverse price movements.
Today, oil producers 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 their fuel costs. Both sides strip price volatility out of their businesses.
Compute has no such tool yet, leaving anyone building or purchasing AI infrastructure exposed to three types of risk:
GPU rental prices fluctuate dramatically—spiking during demand surges or crashing when supply loosens or new chips launch—making it impossible for AI companies to budget accurately for their largest cost item.
Whenever Nvidia releases faster chips, the rental value of previous-generation chips declines, and the collateral behind hardware loans shrinks accordingly.
A data center takes two to three years to build, yet developers have very limited means to lock in their compute costs or revenues. Every one of these decisions is a multi-billion-dollar bet.
These risk exposures create demand for compute futures. But before this market can scale, it must confront the same two problems that have constrained other futures markets: concentration and fungibility.
Attempts to build futures markets around onions, uranium, DRAM memory chips, and bandwidth have all run into one or both of these issues.
For compute, the concentration problem is complex. The buy side is becoming fragmented, as inference demand spans thousands of companies running production workloads. The sell side is broad and still growing—new cloud provider revenues surpassed $25 billion in 2025, spanning over 60 providers. But underneath, it remains highly concentrated: Nvidia supplies most AI chips.
The second problem is fungibility.
Today, compute prices are quoted in GPU-hours—the cost of renting a single GPU for one hour. But two GPUs of the same model can deliver different amounts of compute in an hour.
Silicon Data, together with academic collaborators, ran identical workloads across 3,500 GPUs from 11 cloud providers. Even within the same chip model, they found significant variance. In one test, H100 performance differed by up to 34.5%, and the largest gap across the entire study reached 38%.
The first contracts that can sustain themselves may need to define several grades, much like energy markets differentiate by fuel type, location, and delivery period.
But the bigger question is what happens if compute futures work. Can compute become the next asset class with trillions in notional trading volume, accelerating the AI economy?
