We are excited to announce that PROOF IV invested in Hydra Host's recent $100M Series A, led by Kindred Ventures with participation from NVIDIA, ARK Invest, and other strategic investors!
Hydra Host runs the operating system for AI compute. Rather than buying GPUs and building data centers itself, Hydra gives independent data centers, telcos, and sovereigns the software, supply access, and customers they need to put GPU capacity to work. We've been tracking Hydra closely, and voted to invest in their latest round.
The AI economy is bottlenecked on compute. Demand for GPUs far outstrips supply, and the companies racing to meet it, the vertically integrated "neoclouds" like CoreWeave and Lambda, do so by raising billions to buy chips and build facilities they own and operate. That model works while compute is scarce, but it is capital-hungry and hard to scale across borders. Meanwhile, thousands of mid-sized data centers around the world have power and space but no easy way to deploy modern GPUs or find customers for them.
Hydra closes that gap with an asset-light approach it calls the franchise model for compute. The closest everyday comparison is a hotel chain. Think of GPUs as the rooms and data centers as the people who own the buildings: plenty of owners have space and power, but keeping a hotel booked and running well is a business of its own.
Hydra is the Marriott for AI compute. A data center owns the GPUs and the power; Hydra's software (Brokkr) runs the operations and brings the customers, and Hydra takes a small percentage of everything that flows through. The data center gets a profitable, fully-booked "hotel" without having to build the reservation system or the sales team itself, and Hydra gets a fast-growing network without having to buy a single GPU.
Under the hood, Brokkr is what makes that possible: it smooths over the differences between hundreds of GPU and hardware configurations so a data center can come online quickly, then routes demand from AI labs, inference platforms, and enterprises to wherever capacity sits idle. CEO Aaron Ginn describes the job in four words: converting megawatts into tokens.
What we like about Hydra Host:
Asset-light in a capital-hungry market. Hydra captures the economics of the GPU boom through software and a take rate on usage, without the balance-sheet exposure of owning and operating the underlying infrastructure. That keeps the model scalable and resilient even if GPU pricing normalizes.
Deep alignment with NVIDIA. NVIDIA named Hydra an official NVIDIA Cloud Partner and invested in the round, giving Hydra early access to scarce, high-end GPUs and preferential supply that independent operators cannot secure on their own.
Network-effect economics. Every data center Hydra onboards widens its supply footprint, and every customer using its API deepens demand. More of each makes the network more efficient and harder to leave, with switching costs that grow as operators run their entire stack through Brokkr.
A large and expanding market. Hydra serves the fast-growing tail end of the inference market and the global wave of sovereign AI programs, where governments want domestic compute they don't have to source from vertically integrated U.S. or Chinese providers. The company already operates across more than 50 data centers spanning the Americas, Europe, the Middle East, and Asia.
