Torus launched on Wednesday (11) Torus Foundry, a prefabricated modular infrastructure for AI inference, and announced $500 million in growth capital led by Magnetar, with participation from ICONIQ, Origin Ventures, Ridgeline Partners and Trinity Capital. The round is a Series C, according to Bloomberg.
Torus Foundry packs power, storage, cooling, networking and compute into a single prefabricated unit. Each site delivers 1 to 20 megawatts of inference capacity and goes from contract to running workloads in 12 to 16 weeks — instead of the years a traditional hyperscale campus usually takes. The smallest system is roughly the size of a small cargo van, site preparation takes about a week, and installation and commissioning take one to two days.
The core idea of Torus Foundry is to use electrical capacity that already exists on the grid. Torus installs Foundry behind the meter, using available headroom at transformers in commercial and industrial buildings, without requiring new grid infrastructure. According to the company, inference demand is growing because trained models now run continuously for real users — and McKinsey projects data center demand for inference will grow 35% per year through 2030, versus 22% for training.
Torus pairs every unit with Torus Reserve, an energy storage and management system with batteries and an inertial flywheel, which shapes demand, covers peaks and provides backup during outages. Torus Protect adds integrated networking and security, and Torus operates and maintains each deployment.
Three models and the first customers
Torus Foundry comes in three models: Strategic, a powered shell without GPUs; Partner, a turnkey bare-metal unit with GPUs; and Direct, sovereign inference capacity with single-tenant hardware for enterprises. The first customers are Hydra Host and a second AI cloud provider the company did not name. Torus says it already has thousands of GPUs running in Foundry units and targets more than 47,000 GPUs by the end of 2027. According to Bloomberg, each container holds 576 Nvidia HGX B300 GPUs in about 577 square feet, at 1.2 MW of power.
Utah factory and waterless cooling
Torus designs, engineers and manufactures Foundry at GigaOne, Torus's plant in Salt Lake City, Utah, with 545,000 square feet and nearly 400 employees. The Torus Foundry production line is already online and will be capable of producing 500 MW of IT load, with plans to reach 1 GW. The system uses waterless closed-loop cooling, with 95% less water use than evaporative coolers, and each unit donates GPU time to medical research through Folding@home.
The announcement adds to another AI infrastructure funding this week: British chip startup Fractile announced a $720 million Series C for its inference chips. Read Geekiki's story on Fractile.
“Most providers are bringing power to the chips. Foundry brings the chips to power,” said Nate Walkingshaw, founder and CEO of Torus, in the official announcement.
Sources: Torus, via Business Wire; Torus; Unite.AI; VentureBurn (with Bloomberg data).
Featured image: a Torus Foundry installation outside the GigaOne plant in Salt Lake City, on November 9, 2026. Photo: Ken Hovley, Torus, via Business Wire.
Transparency: this article was produced with the assistance of artificial intelligence, cross-checking public posts on X and sources on the internet. We recommend consulting the original sources listed above to confirm the information.
— Muse
