Test report DSG-6841 · Rev D · tested November 17, 2026
AI Datacenter InfrastructureDevice under test
CFTC Extends GPU Futures Review; Trading Slips to Early 2027
The CFTC will extend its review of GPU rental futures by at least 60 days, pushing CME’s planned H100 and B200 launch to early 2027 at the earliest amid manipulation concerns.
- Read
- 3 min
- Words
- 635
- Node
- 10nm
- Operator
- Amara Osei
Spec summary
- The CFTC will extend the GPU futures review by at least 60 days.
- Actual trading is expected no earlier than early 2027.
- CME's original launch window was early October.
- The contracts reference Nvidia H100 and B200 GPU rental prices.
- The Financial Times reported the CFTC decision on the 11th.

By Amara Osei
The U.S. Commodity Futures Trading Commission will extend its review of futures contracts that reference AI-chip rental prices by at least 60 days, pushing the start of actual trading to early 2027 at the earliest. The Financial Times reported the decision on the 11th.
CME Group, one of the world's largest derivatives exchanges, had scheduled futures products for an early-October launch. The contracts would reference rental prices for Nvidia's high-performance H100 and B200 GPUs and rely on a price index developed by AI startup Silicon Data. No public futures market currently trades future GPU rental prices.
What changes for CME's H100 and B200 contracts?
The extension removes early October from CME's timetable and places early 2027 at the earliest expected start of trading. It gives the CFTC at least another 60 days to examine the proposed Silicon Data benchmark and the market conditions behind it. Until the review ends, the products cannot offer investors an exchange-traded way to hedge or lock GPU rental rates.
Takeaway: CME's near-term listing plan has become an early-2027 earliest case.
What would GPU futures be used for?
Demand for AI computing has surged, and Wall Street wants to treat computing capacity as a new financial asset, much like crude oil. AI companies expect a futures market to serve three main purposes:
- Rate locking: AI companies could lock GPU rental rates in advance and reduce their exposure to rising costs.
- Price hedging: Companies that own GPUs could hedge against a decline in chip values.
- Collateral valuation: Financial institutions that lend against GPUs could use market prices to assess the value of that collateral.
A spot market already exists for short-term trading in GPU computing power, but no public futures market lets participants trade future rental prices. CME's proposed contracts would fill that gap with a forward rate based on the Silicon Data index.
Takeaway: The proposed market would turn short-term GPU capacity prices into a forward benchmark for costs, chip values and collateral.
Who is competing for the GPU futures market?
Competition to establish the market is intense. Beyond CME, the Intercontinental Exchange and prediction-market operator Kalshi are pursuing similar products. Support for a GPU futures market also comes from global trading firms DRW and Jump Trading and venture capital firm Andreessen Horowitz.
Some investment firms expect an AI-computing futures market could eventually rival the crude oil futures market in size. That expectation places the proposed contracts within Wall Street's effort to build a commodity-style market around scarce AI capacity, even as the CFTC examines whether its price data can support one.
Takeaway: Exchanges, trading firms and venture investors are competing over a market that has not yet cleared regulatory review.
Why did the CFTC extend the review?
The CFTC's concerns center on two features of the underlying market: GPU rental prices remain opaque, and the market is concentrated among a small number of suppliers. The regulator sees those conditions as creating a risk of manipulation.
The Financial Times noted: “If price indexes are calculated from quotes on online platforms with thin actual trading volume or from private contracts, a handful of trades could move prices artificially.” That risk bears directly on the Silicon Data index because the contracts need a reference price that reflects more than isolated trades.
Columbia Business School professor Tomasz Piskorski said: “Transactions between affiliated companies could be distorted to make demand appear stronger than it is.” Such transactions would weaken the same index by overstating activity in the rental market.
Takeaway: The delay rests on whether the proposed index can measure real GPU demand without accepting manipulated prices.
GPU futures remain under CFTC review, with actual trading expected no earlier than early 2027.
via en.sedaily.com (Source)
More from Amara Osei
Show full bio
Staff writer covering money, funding and market shifts around hardware layer of artificial intelligence at Die Signal.
698 articles
Same lot · LOT-362D
- DSG-345545nmSpaceX strikes $40 billion Apollo lease for Nvidia AI chips
- DSG-194010nmMicron raises 2027 HBM prices after missing 2026 surge
- DSG-61675nmAMD Shares Up 149% in 2026; Analysts See Path to $800 in 2027
- DSG-350528nmTencent to rent 100,000 offshore AI chips from Oracle for $7 billion
- DSG-789620nmBloomberg Lifts 2030 AI Accelerator Forecast to $1.4 Trillion