GPU rental prices can change over the life of a cluster. Today, Hyperbolic, Architect, and Luxor completed the first over-the-counter (OTC) option on GPU rental prices, hedging approximately $5 million of Hyperbolic’s future NVIDIA B200 rental revenue.

When we announced our Series A1 last month, we said we were exploring ways to help customers and suppliers manage compute price risk. This trade is the first result.

What the trade is

Hyperbolic worked with Architect to design a put option on B200 rental prices. Architect structured the hedge through its CFTC-registered Introducing Broker affiliate, and Luxor, acting as swap counterparty, priced and executed it.

The option settles against the Silicon Data B200 Rental Price Index (SDB200RT), a daily benchmark of on-demand B200 rental rates, with monthly cash settlements over a one-year term. If B200 rental prices fall below the agreed level, the option pays out, protecting part of our future revenue.

Architect and Luxor tailored the terms, from the index to the start date and settlement schedule, to the hardware we operate and the period we needed covered.

Why OTC, and why now

GPU operators and AI teams face price risk on different sides of the market. Operators earn less when rental prices fall, and AI teams pay more when they rise. But a standardized contract may not match the hardware or rental period an operator needs to hedge.

Exchange-listed compute futures are still on hold. As we wrote recently, regulators are examining how the underlying indices work before listed GPU futures can launch. OTC contracts can be shaped around a specific fleet, term, and revenue profile, which makes them useful for operators who want protection now.

“Our mission is to make the world’s AI compute unified, productive, and liquid, and a liquid market needs a way to manage price risk, not just a way to buy and sell capacity,” said Jasper Zhang, CEO and Founder of Hyperbolic. “With Architect and Luxor, we turned a year of B200 rental revenue into something we can plan around, and that’s what lets compute be run like the commodity it is becoming.”

How this fits what we’re building

Hyperbolic is building a global compute liquidity and orchestration platform. Today, GPU capacity is spread across more than 300 providers with different hardware, networking, software, and operating standards. Hyperbolic brings it together. Forge, our unified control plane, standardizes heterogeneous bare-metal clusters into validated infrastructure, and our managed Kubernetes and inference services help teams run workloads across providers through one platform. The result is more useful work per GPU hour for AI teams, and faster time to market, higher utilization, and better yields for GPU providers.

This trade is one of the first steps toward building financial infrastructure for compute, where participants have more tools to manage price risk. It also supports our longer-term vision for the Global Compute Grid: making GPU capacity easier to access, allocate, and scale. Over time, instruments like forwards, swaps, and options could give compute market participants more ways to plan around price changes.

Thank you

Thank you to Brett Harrison and the Architect team, and to Matt Williams and the Luxor team, for designing and executing this trade with us.

Building AI or bringing GPU capacity to market? Explore Hyperbolic.

This post is for informational purposes only and does not constitute an offer, solicitation, or recommendation to enter into any derivatives transaction.