The Pulse
Hyperbolic Raises $30M to Build a Global GPU Compute Grid
Hyperbolic raised an oversubscribed $30 million Series A1 led by Walden Catalyst to expand its Forge platform and build a marketplace for GPU capacity. The company says its planned Global Compute Grid will let providers and AI teams buy, se
AI.info Team · Written with AI · fact-checked ·
Hyperbolic says more than 300 GPU cloud providers make AI compute hard to access; its proposed answer is a marketplace connecting capacity across those providers. The San Francisco company announced an oversubscribed $30 million Series A1 on October 27, led by Walden Catalyst, to expand its orchestration software and build what it calls a Global Compute Grid.
The plan addresses a mismatch: GPU providers may have capacity sitting idle, while AI teams face different hardware, networking and software standards across vendors. But pooling those resources into a market requires more than listing available machines. Hyperbolic’s own announcement describes standardization and workload orchestration as part of the work still needed to make capacity usable across providers.
Funding backs Forge and a planned marketplace
Hyperbolic says the new capital will advance Forge, its unified control plane, expand multi-cloud orchestration and managed services, and support development of the Global Compute Grid. The company says Forge provides a shared foundation for accessing varied infrastructure, managing capacity and coordinating workloads across clusters.
Walden Catalyst led the round. Dell Technologies Capital, General Catalyst, Vista Equity Partners, Cambium Capital and Salience Capital also participated, alongside other strategic investors. The announcement lists angel investors connected to Citadel Securities, the New York Stock Exchange, Vista Equity Partners and GPU cloud providers WhiteFiber and Vector Core Compute.
“The demand for AI compute will continue to grow, but adding more hardware is only part of the answer,” said Lip-Bu Tan, founding managing partner at Walden Catalyst and CEO of Intel. Tan added that the value of infrastructure investment depends on putting capacity to work and making it available to the teams that need it.
Trading idle GPU capacity is the harder test
Hyperbolic’s proposed grid would let suppliers and customers buy, sell and resell eligible GPU capacity through one programmable platform. The company plans to introduce spot instances and dynamic pricing, which it says could give buyers more flexibility and help providers earn revenue from otherwise unused capacity. Those features are described as plans, not as services already available through the grid.
The company also intends to develop market-making strategies to support trading across different contract durations. It says it is exploring ways for customers and suppliers to hedge compute-price risk and manage capacity commitments. The announcement does not give a launch date for the grid or detail its pricing.
Hyperbolic says its platform supports more than 250,000 AI builders, including teams at Modal, Higgsfield, Daytona, HeyGen, Voio, RadixArk and Pika. That figure indicates the reach the company claims for its existing platform, but the release does not state how much capacity those users consume or how much GPU capacity is currently available for resale.
Jasper Zhang’s pitch joins orchestration with liquidity
Founder and CEO Jasper Zhang frames the product as both infrastructure software and a way to trade capacity. “AI teams should be able to access compute wherever it is, run their workloads, and turn unused capacity into revenue through one platform,” Zhang said. He said Hyperbolic is combining orchestration with liquidity so customers can use compute more flexibly and providers can make better use of their infrastructure.
Hyperbolic’s immediate challenge is to make unlike GPU clusters reliable enough for workloads to move between providers, then attract enough buyers and sellers to make that marketplace useful. The funding announcement lays out the intended components—Forge, multi-cloud orchestration, spot capacity and resale—but gives no launch timetable for the grid.