Hyperbolic Raises $30 Million to Broker GPU Capacity Across 30 Neoclouds, With Intel CEO Lip-Bu Tan's Fund Leading

Hyperbolic, a startup that aggregates AI compute from dozens of small cloud providers, has raised a $30 million Series A-1 led by Walden Catalyst, the venture firm co-founded by Intel chief executive Lip-Bu Tan. Business Insider reported the round on November 1, listing Dell Technologies Capital, General Catalyst, Vista Equity Partners, Cambium Capital and Salience Capital among the participants, and putting Hyperbolic's total funding at about $50 million.

The company's product is a brokerage and control layer for a market that has grown faster than its plumbing. Specialized GPU clouds — the neoclouds — sprang up to rent capacity that the hyperscalers could not or would not provide, but each runs its own console, billing and hardware stack. Hyperbolic's Forge control plane standardizes heterogeneous bare-metal GPU clusters behind one interface, and the company says it is building what it calls the first Global Compute Grid, with plans to let customers buy, sell and resell capacity so that GPUs can be consumed roughly like electricity.

The company's own numbers, which have not been independently audited: Hyperbolic says roughly $200 million of annual compute is connected to its platform, that it serves more than 250,000 AI builders, and that it is targeting $150 million in annual recurring revenue by the end of the year. Those are company-reported figures, and the gap between $200 million of connected compute and $150 million of revenue says a great deal about how thin broker margins in this layer are expected to be.

The strategic signal in the round is the lead investor. Walden Catalyst is the fund of Lip-Bu Tan, who runs Intel — a company trying to sell its own accelerators into an AI market dominated by Nvidia. An aggregation layer that makes demand portable across 30 suppliers is exactly the kind of infrastructure a second-tier chip vendor wants to flourish, and the round suggests the chip establishment sees compute intermediaries as strategically useful rather than disruptive.

The funding lands in a grim week for the neocloud business model itself. CoreWeave's bond yields have climbed to around 15 percent as investors reprice its roughly $35 billion debt load, Elon Musk predicted over the weekend that more than 200 neoclouds face guppy-level death and consolidation because debt markets have shut off companies without revenue independent of AI data centers, and Microsoft recently cancelled a deal with the British operator Nscale over power issues. A broker that can redirect demand from a distressed provider to a healthy one is, in theory, a hedge against exactly that fragility.

It is also a bet on utilization. Most neoclouds run meaningfully below full occupancy, and idle H100s and B200s earn nothing while their financing costs compound. Hyperbolic's pitch — unify the supply, standardize the interface, and let capacity flow to whoever pays — is the standard playbook of every wholesale market that ever formed, from airline seats to shipping containers. Whether AI compute behaves like those markets, with storable, fungible units, is the open question; a GPU cluster is not an empty seat, and buyers care about interconnect, locality and latency far more than seat maps.

For now, the company has a plausible wedge, a notable backer and a market that keeps getting bigger and more nervous at the same time. If the neocloud shakeout arrives on Musk's schedule, the survivors will be the ones with demand they can fill from anywhere. That is precisely the position Hyperbolic is trying to occupy from the software layer up.