The future of AI requires GPU futures

Futures markets offer transparent pricing and risk management capabilities. AI will need these to flourish

The infrastructure behind artificial intelligence is expanding at a blistering pace. Hyperscalers and specialized compute firms are spending hundreds of billions of dollars building out data centers, much of it on borrowed capital. But the cost of that capital remains stubbornly high, largely because the single most important input to AI, GPU compute, remains volatile, opaque, and unhedgeable.

That may be about to change.

Buried in the Trump administration’s AI Action Plan is an explicit call for the creation of a compute forward market, laying the groundwork for what could become a full-fledged, regulated futures market in GPU capacity. Such a market would allow both buyers and sellers of compute to lock in prices years in advance, just as energy producers and airlines hedge oil.

It’s not just policy theorizing, either. Don Wilson, the billionaire founder of DRW, one of the most sophisticated proprietary trading firms in the world, has quietly seeded a company called Compute.exchange, which is building an auction platform for spot GPU trading. The goal is to produce a standardized, transparent price index, which is a prerequisite for any liquid futures market.

The potential implications are massive. With a forward curve in place, infrastructure developers could securitize long-dated GPU revenue just like power producers do with electricity. That, in turn, could lower the cost of capital for new builds by several hundred basis points, unlocking enormous latent supply and catalyzing a new financial layer atop the AI stack.

But designing a futures market for compute isn’t trivial, and it won’t look exactly like oil, gas, or electricity. The devil is in the structure.

In this post, I explain why GPU futures make sense, how they would transform infrastructure finance, and what can still go wrong, from obsolescence risk to jurisdictional fights to market capture. If you're serious about understanding the financial architecture of the AI era, read on.