How AI Computing Power Is Becoming a Traded Commodity

Major investment exchanges are building standardized markets where traders can buy and sell futures contracts — essentially bets on future prices — tied to the cost of AI computing power. This shift could change how companies pay for and manage the expensive compute infrastructure that powers artificial intelligence systems. The move comes as U.S. and Chinese regulators begin writing the rules for these new markets.

CME Group, one of the world's largest futures exchanges, has partnered with a pricing firm called Silicon Data to launch futures contracts based on the cost of graphics processing units — the specialized chips that power AI models. Silicon Data's index, supported by trading firm DRW Holdings, creates a benchmark price that traders can reference. Intercontinental Exchange, another major derivatives platform, is also preparing to launch similar products.

This pattern is not entirely new. In the late 1990s, when internet companies were building out massive telecommunications networks, bandwidth — the capacity to transmit data — became something traders could buy and sell as a financial product. AI computing power is following a similar arc, driven by the same underlying pressure: demand for these resources fluctuates wildly, which makes the costs unpredictable.

China Develops Its Own AI Market

China's Shanghai Futures Exchange is developing a separate futures market of its own, designed around what are called AI tokens — a technical term for the smallest units of work that an AI model performs when answering a question or generating text. The Shanghai exchange's research is still in early stages but reflects China's broader competition with the United States in artificial intelligence.

The parallel development in both countries shows how the tools for pricing and trading AI resources have become part of the strategic competition around AI itself — just as important as controlling the chips and software beneath them. China's focus on AI tokens is more granular than the U.S. approach: rather than pricing the hardware (GPUs), it prices the actual computational work the AI performs.

These futures contracts address a real business problem. When companies deploy AI systems at scale, the computing costs can swing dramatically, especially when many organizations need resources at the same time. A standardized futures market would let traders, cloud providers, and companies lock in prices in advance — much like farmers lock in crop prices using agricultural futures.

How Regulators Are Stepping In

The U.S. Commodity Futures Trading Commission, the agency that oversees derivatives trading, has created a task force focused on how to regulate new markets tied to cryptocurrency, blockchain, AI, autonomous systems, and prediction markets. The task force is coordinating with other federal agencies, including the Securities and Exchange Commission.

The CFTC has already published guidance on artificial intelligence in financial markets, signaling that regulators see AI-specific risks that go beyond how they oversee traditional futures contracts. As AI-related financial products move from experimental into mainstream use, this regulatory foundation matters.

These developments sit alongside a broader growth in digital asset trading. Coinbase Global launched a derivatives exchange for cryptocurrency in May 2023, and CME Group introduced futures on the XRP cryptocurrency in May 2025, showing that financial institutions increasingly see a business case for allowing traders to manage risk across digital and computing-based assets.

What This Means in Practice

Rather than individual companies negotiating expensive long-term contracts with cloud providers or scrambling to buy scarce GPU resources day by day, standardized futures would offer a central marketplace where prices are transparent and discoverable. This is especially valuable for smaller AI companies and research institutions that lack the bargaining power of tech giants.

A startup building an AI model could use these futures to lock in the cost of computing power months in advance. A cloud provider like Amazon or Microsoft could hedge against sudden drops in demand. The approach mirrors how energy companies, airlines, and farmers have used futures markets for decades to manage volatility.

The transition of AI infrastructure into a financial commodity follows a well-established historical pattern. When a technology becomes scarce and expensive enough, sophisticated markets eventually develop to manage that scarcity. What stands out here is the speed: AI computing is becoming tradeable while the technology itself is still changing rapidly, which poses challenges.

One key challenge is standardization. Traditional commodities like oil or wheat have agreed-upon quality grades. GPU computing power is messier — the cost of running a particular AI task varies depending on what model you are running, the precision of calculations required, and how the code is optimized. Defining what exactly is being bought and sold in these contracts will require careful technical work.

The fact that both U.S. and Chinese exchanges are launching these products suggests institutional confidence that the underlying technology is stable enough to trade. The regulatory oversight also helps — it signals these will not be wild-west markets. But the existence of two separate systems, in two countries with strategic tensions, could create complications for global price discovery and may eventually force some kind of coordination between regulators.

For technology professionals and companies building or deploying AI systems, the practical upshot is this: within a few years, the cost of computing power may become more predictable and hedgeable, much like other operational expenses. That would be a genuine shift in how AI infrastructure gets managed and funded.