Wall Street is looking to trade futures on AI computing power, while Washington is seeking more time to consider the idea.
The Commodity Futures Trading Commission (CFTC) has extended its review of two compute-power futures contracts by 60 days. The new deadline is January 8, 2027, replacing the previous deadline of November 9, 2026.
The contracts were developed by NYMEX, a subsidiary of CME Group. They would allow traders to bet on or hedge against the hourly cost of renting Nvidia GPUs in the United States.
What is actually under review
The contracts are linked to two benchmarks created by Silicon Data: the H100 Rental Index and the B200 Rental Index.
Both indices track the hourly on-demand rental price of Nvidia GPUs in the US.
Each futures contract covers 730 GPU-hours and settles in cash, using the monthly average of the relevant index. Prices are quoted in US dollars per GPU-hour.
No graphics cards change hands when the contracts expire. They involve no physical hardware delivery and are purely financial instruments pegged to a rental price.
A timeline that keeps stretching
NYMEX initially submitted the contracts to the CFTC on August 11, 2026. The agency then extended the original review period by 45 days, setting a deadline of November 9, 2026.
The CFTC has now granted a second extension of 60 days, moving the decision to January 8, 2027.
The extension is voluntary, and NYMEX agreed to it with a notable condition: while the review is ongoing, no other registrants may self-certify or list similar contracts.
Why regulators are taking their time
The CFTC has recognized that these products present complex issues. The agency has specifically raised questions about liquidity and manipulation risk, and is seeking industry input before approving them.
Liquidity concerns whether enough buyers and sellers will participate to produce meaningful prices. A contract with little trading can swing sharply on just a few orders.
Manipulation risk is more complicated. Since the futures settle against an index, the product’s integrity depends on how that index is constructed. If a small number of participants could influence a benchmark of hourly GPU rental rates, they could potentially influence the futures linked to it as well.
What this means for the AI economy
For companies developing AI models, computing costs are among the biggest budget items. A futures market would give them a way to lock in costs in advance. A startup preparing a large training run could hedge against rental prices rising before it begins, while a cloud provider with unused capacity could hedge against prices falling.
If the CFTC approves the contracts, they could help establish a recognized pricing standard for AI compute. A publicly traded exchange price would make it easier for buyers and sellers to compare deals.
For now, the key date is January 8, 2027. The industry feedback the CFTC gathers on liquidity and manipulation between now and then will likely influence whether the contracts launch as filed, launch with changes, or do not launch at all.
Source: cryptobriefing.com. Adapted by NBTC.
Add A Comment
Comments are closed.