← All posts / Policy

Computing Power Futures Shut Down: Commerce Orders Kalshi to Pull Its AI Compute Price Tracker

The US Commerce Department quietly ordered prediction-market operator Kalshi to take down its AI compute forward-curve product, and pressed the CFTC to freeze new compute contracts for 60 days — an uncharacteristic intervention that stunned an industry used to White House AI enthusiasm.

In a rare intervention that surprised an industry accustomed to White House enthusiasm for both AI and financial-market innovation, the US Commerce Department last month quietly ordered prediction-market operator Kalshi to take down one of its products tracking the price of AI compute — the crucial resource from data centers that is driving the entire artificial intelligence boom.

Semafor first reported the story on September 15, 2026, citing people familiar with the matter. Commerce officials cited national security concerns when they told Kalshi to unpublish its AI compute future curve, a product that aggregates data from several markets allowing users to bet on the cost of renting Nvidia chips, creating an overall picture of where AI compute costs are heading. Kalshi quietly complied, though many of its underlying betting markets remain live. Separately, Commerce also pushed the Commodity Futures Trading Commission (CFTC), which oversees prediction and futures markets, to freeze approval of new compute contracts for 60 days, some of the people said.

What Was Kalshi Actually Selling?

Kalshi’s compute product aimed to do for AI computing what oil futures do for crude — let buyers and sellers of compute lock in prices, and give traders a way to bet on where those prices go. The company launched its forward curves in July 2026, becoming what it billed as “the exchange for the AI economy.” The curves measure the implied future price of renting one hour of a specific GPU, and went live for Nvidia’s B200, H200, and A100 chips, with contracts tied to additional chips like the H100 and RTX 5090 following.

The product had real traction. Kalshi’s GPU rental markets recorded $4.4 million in notional volume through July 27 — roughly 15 times more than rival Polymarket’s $285,000, according to data analytics firm Allium. That summer surge attracted heavyweight interest: Fortune reported in August that Kalshi’s CEO was racing to build a full futures market for AI’s most critical commodity, with counterparties including CoreWeave, CME, and Intercontinental Exchange (ICE), the parent of the NYSE. Betting markets showed the market splitting around $0.79/hour on the B200 at year-end.

Why the Sudden Concern?

It is unclear exactly why Commerce is worried about the product. One potential reason floated to Semafor by market participants is that compute futures could be manipulated to show a sharp drop in the cost of older chips — which might destabilize AI stocks and debt markets. Some of these markets are thinly traded, which could lead to volatility even without bad actors.

The stakes are enormous because the cost of compute has become one of the most important numbers in the US economy. On one side of the debate are fears that older chips, which serve as collateral for billions of dollars of borrowing by neoclouds like CoreWeave and which underpin data-center deals, will become obsolete and decline in value. On the other are concerns from big companies adopting AI that shortages of power and infrastructure will send token prices soaring.

That uncertainty is precisely what gave rise to the futures market in the first place — and what made it start to take off this summer.

A Denial Wrapped in a Takedown

The official response to the story has been contradictory. Kalshi declined to comment. A Commerce spokesperson said the department “has never once asked Kalshi to take down this market or any other markets,” and a spokesman called the story “false” — even as the product remains unpublished and the CFTC’s freeze, per Semafor’s sources, was real. The CFTC did not respond to a request for comment.

The 60-day pause on new compute contracts could have concrete casualties. It could delay plans by exchange operators like CME and NYSE-parent Intercontinental Exchange, along with upstarts like Architect Financial Technologies, to list their own two-sided compute betting parlors.

The Bigger Picture: When AI Becomes Systemically Important

The intervention is telling for what it reveals about how Washington now sees AI infrastructure. This is an administration that has championed AI dominance as national strategy, promoting everything from the AI race with China to financial innovation in prediction markets. Yet when a market emerged that could publicly price the industry’s core input — compute — someone inside Commerce decided the signal itself was dangerous.

There are real echoes here of the early oil futures markets, and of the debates over whether speculators destabilize strategically important commodities. But there is also a more modern dynamic: if older GPUs are collateral for billions in neocloud debt, then a public, tradeable price curve for those GPUs is effectively a live mark on the collateral of an entire financing ecosystem. A sharp, manipulated, or even merely noisy downward move could trigger margin calls, covenant breaches, and a repricing across AI equities and credit. Regulating that price discovery out of existence doesn’t remove the risk — it just makes it invisible.

For now, the industry is left reading an unusual signal: the US government appears to believe that the price of AI compute is now a matter of national security — so sensitive that publishing where the market thinks it is heading had to be stopped. Whether the pause is a temporary precaution or the first move toward a broader regime of compute-market oversight remains to be seen. Either way, the era when AI compute was just another cloud billing line item is definitively over.