Ex-Citadel Quant Builds AI Compute Futures to Hedge GPU Risk
As the AI trade shows cracks, a former Citadel quant is launching physical AI futures to give traders a new hedge.
The AI trade is wobbling, and Wall Street is getting nervous. Smart money poured into everything GPU-adjacent over the past two years, but cracks are forming — and hedging pure AI exposure isn't easy when there's no clean instrument to do it. That gap might finally be closing.
Jasper Zhang, a former Citadel quant, is the man with a plan. He runs Hyperbolic, a GPU exchange that's developing software for what the firm calls "physical" AI futures. Think of it like commodity futures, but instead of barrels of oil or bushels of corn, you're trading actual compute capacity. That's a fundamentally different animal from betting on Nvidia stock.
Read more Warren Buffett's AmEx Bet Is Up 440% — and Still Running →
The logic is tight. If you're long AI infrastructure plays but worried about demand destruction or a slowdown in model training spend, a physical compute futures contract gives you a direct, correlated hedge. It strips out the equity noise — management drama, buybacks, analyst rewrites — and gets you closer to the raw underlying: GPU hours. That's the thing that actually moves when AI spending moves.
Wall Street has been hunting for exactly this kind of instrument. Equity proxies like semiconductor stocks are blunt tools. They're influenced by macro, geopolitics, and supply chain dynamics that have nothing to do with whether enterprises are actually spinning up AI workloads. A futures market built on real compute supply and demand changes the hedging calculus entirely.
Whether Hyperbolic can execute and attract enough liquidity to make the market functional is the real question — but the concept is sound, and the timing is sharp. When a trade starts cracking, the traders who already have the hedge in place win. Continue reading at US Top News and Analysis.
