Ex-Citadel Quant Pitches GPU Trading as AI Hedge

As CoreWeave shares tumbled and dragged much of the AI sector down with them, investors are asking a straightforward question: what is the best way to hedge AI risk? Jasper Zhang, a former quantitative researcher at Citadel Securities, believes he has an answer.

Zhang now leads Hyperbolic, a 50-person financial technology startup building a spot marketplace for tradable GPU computing power - a tool that would let traders hedge against falling AI demand. The platform lets investors buy and sell GPU clusters, the units that power AI compute, through Hyperbolic's Forge software. Forge logs virtually into clusters worldwide and makes technical adjustments so mismatched clusters work better together, allowing traders to buy and sell a standardized unit of compute and also control the underlying GPUs for their own purposes.

Zhang says it is as good as being in the data center itself, and sees the technology as the first building block of "physically" settled GPU futures, which on settlement would give contract owners complete control over a cluster with known characteristics. GPU architecture varies widely depending on assembly, producing different output levels, and financing agreements often stop builders from sharing prices - a problem known in futures as "basis risk."

The 31-year-old Zhang came to America with several Chinese math medals and finished a math doctorate at U.C. Berkeley in two years before joining Citadel Securities to build trading models. Hyperbolic's head of finance, Hong Chan Kim, is a hedge fund veteran of Point72, Balyasny Asset Management and Citadel LLC. Hyperbolic has raised $20 million since launching in 2022, is in the midst of another funding round, and expects a $150 million annualized run rate by year-end. It says it has about 250,000 users and an estimated $200 million worth of annual GPU compute connected to its platform.

Others see a similar opportunity. CME Group plans a cash-settled futures product tied to GPU pricing maintained by Silicon Data, an index provider backed by DRW. "There's a ton of demand and capital flowing into this ecosystem so there's a lot of risk," said DRW founder and CEO Don Wilson. The CME product is under CFTC review, with the commission recently extending that period to November 9th, likely also tying up another contract proposed by Intercontinental Exchange and data provider Ornn.

Hyperbolic's Kim doubts cash-settled futures will reflect reality: "Physical usage is what will tether that connection between the reality and the derivatives." Carmen Li, CEO of Silicon Data and the DRW-backed Compute Exchange, said indices are a normalized global number: "You get financially hedged, but if you're looking for procurement, that's not going to help you." Zhang argues the AI complex - especially smaller startups lacking capital or credit for big hardware purchases - needs a virtual pipeline to physical hardware. "CoreWeave's debt situation would be less concerning if providers had more ways to monetize available GPU capacity during a slowdown, especially when long-term contracts carry thin margins," he said. "Cash-settled futures can help falling prices, but they do not by themselves put idle GPUs to work."

Source: CNBC