GPU infrastructure management startup Hydra Host raises $100M

Data center infrastructure startup Hydra Host Inc. said today it has bagged $100 million in a Series A round of funding led by Kindred Ventures.

Investors piled into the round, with Nvidia Corp., Ark Invest, SPLY Capital, Era Funds, Comcast Ventures, Magnetar, Peak6, Founders Fund, 10x Founders, Sterling Road and Flume Ventures also participating.

The startup has carved out a novel role in the artificial intelligence compute industry, creating a marketplace that helps connect developers and enterprises to high-performance graphics processing units that are pooled from data center operators across the globe. It’s trying to solve a growing global compute capacity crunch that has made GPU resources extremely expensive.

Hydra Host has developed an operating system for data centers that supports the procurement, provisioning and orchestration of GPUs, as well as an intelligent “offtake” network that’s used by the world’s leading AI inference platform providers, frontier labs and enterprises. The Boulder, Colorado-based company says the Brokkr AI Factory Operating System has been deployed at more than 50 data centers across the world, including in the Americas, Asia Pacific and Europe, Middle East and Africa regions.

The round comes amid a backdrop of intense debate around the staggering capital cost of AI infrastructure. Last week, Oracle Corp.’s stock fell sharply following an earnings report in which it revealed a capital spending plan to build out its AI data centers that exceeded Wall Street’s forecasts. The company said it plans to raise billions of dollars in new debt, unsettling investors worried about the sustainability of such spending.

The reaction to Oracle’s plan revealed the growing tensions in the AI industry – the surging demand that makes distributed compute marketplaces like Hydra Host so attractive also requires staggering upfront investment from anyone trying to compete at large scale.

Hydra Host says that its distributed compute model is a direct result of this dynamic. Instead of trying to raise billions to build out its own data centers, it aggregates GPU supply from global providers to deliver capacity in a more flexible way that increases utilization beyond what hyperscalers can compete with.

Co-founder and Chief Executive Aaron Ginn said the company will use the funds from today’s round to expand its GPU-as-a-service platform to meet the rising demand for AI infrastructure. “We have spent years building the go-to operating system for global GPU deployments for data centers that want to become an AI factory,” he explained. “This funding will expand our data center footprint, scale our platform and support our global customer base, who are looking to rent GPUs easily anywhere and everywhere. It will also help grow our software team, worldwide operations, and data center support to make converting megawatts into tokens easier.”

Image: Hydra Host