Kalshi thinks the most important commodity of the AI era is compute. The CFTC-regulated prediction market just launched what it calls the first market-driven forward curve for GPU computing power, a way to bet on where the price of AI processing is headed.
The product was unveiled in a Bloomberg exclusive by Uday Shah, Kalshi’s newly appointed Chief Risk Officer and a 16-year veteran of CME Group. It plots future prices of computing power and positions Kalshi squarely in a brewing fight with the biggest names in derivatives. Both CME Group (NASDAQ:CME | CME Price Prediction) and Intercontinental Exchange (NYSE:ICE) have announced their own compute futures products.
The Pitch: Compute Is the New Oil
Kalshi CEO Tarek Mansour has been direct about the ambition. “Compute is the new oil,” he has said, adding that “compute futures will eventually dwarf oil futures.” The numbers behind the claim are staggering. Hyperscalers have committed “north of $500 to $600 billion just for 2026” to computing infrastructure, according to Kalshi, with total addressable market estimates stretching into the trillions.
The launch follows a May 2024 prediction from BlackRock CEO Larry Fink at the Milken Institute that “a new asset class will be buying futures of compute.” Two years later, that new asset class is being built in real time.
Why Kalshi Thinks It Has an Edge
Shah’s core argument is that the GPU market is fragmented, and the fragmentation is Kalshi’s opportunity. Computing power lacks the standardization of a barrel of oil. Prices vary by chip grade, by location, and by use case, which makes a single clean index hard to pin down. Traditional exchanges focus on specific indices; Kalshi’s prediction-market structure lets it list many contracts simultaneously to capture that fragmentation.
He also draws a sharp line on price discovery. Kalshi’s forward curve is “true market-driven,” built from actual trading, while competing curves lean on OTC deals or bilateral contracts negotiated privately between parties. In a market this new, whose price is the “real” price is a live question.
The Competitive Subplot
A pointed rivalry sits underneath all of this. CME is currently suing the CFTC to block Kalshi’s perpetual futures. Shah’s move is especially loaded: he defected from CME, the very exchange trying to box Kalshi in, to build Kalshi’s compute product. A 16-year CME insider now helping the challenger draw the map is a direct competitive signal.
Both incumbents are already priming the ground. CME reported Q1 2026 revenue of $1.88 billion and record ADV of 36.2 million contracts, with CEO Terry Duffy citing “expanding access through initiatives such as U.S. Treasury clearing, 24/7 cryptocurrency trading and prediction markets” as core growth vectors. ICE posted Q1 2026 revenue of $2.98 billion and last October announced a strategic investment in Polymarket, a leading prediction market platform, expanding its footprint into decentralized prediction markets.
Why It Matters for Investors
This is a land grab for what could become an entirely new derivatives category. If Mansour is right, the exchanges that establish standard contracts and capture volume stand to win a durable, high-margin business. CME’s operating margin sits at 69.8%; ICE’s adjusted operating margin expanded to 65% in Q1 2026. That is the prize Kalshi is aiming at.
CME trades near $243.61, down 8.16% year to date, while ICE sits at $137.61, down 14.44% YTD. Both remain the public-market vehicles for exposure to this category. Kalshi, still private, is the disruptor trying to define the standard before the incumbents lock it down.
A caveat: this is early and unproven. A forward curve for compute is only as useful as the liquidity and standardization behind it, and the GPU market has neither in abundance today. New derivatives categories launch frequently, and most never reach the scale their founders promise.
The direction is clear, though. When the CEO of the world’s largest asset manager predicts a new asset class, and a CFTC-regulated market plus the two biggest names in derivatives all race to build it within two years, the question is no longer whether compute becomes a tradable commodity. It’s who owns the market when it does.
Contact [email protected] for any questions or corrections.
Top Gaining Stocks
Top Losing Stocks