Torus launched Foundry, modular AI infrastructure for inference workloads. The Salt Lake City company also announced US$500 million in growth capital led by Magnetar. Torus makes distributed power and compute systems for commercial and industrial sites.
Faster deployment model
Foundry packages power, storage, cooling, networking, and compute into prefabricated systems. It delivers 1 to 20 megawatts per site in 12 to 16 weeks from contract to token. Each unit is installed behind the meter using existing grid capacity.
Investors and manufacturing
The round included ICONIQ, Origin Ventures, Ridgeline Partners, and Trinity Capital. The capital will expand GigaOne and support Torus Reserve and Foundry deployments across the United States. Foundry is built at a 545,000 square-foot plant in Salt Lake City called GigaOne.
Product options
Foundry is available in Strategic, Partner, and Direct models. Strategic provides a powered shell without GPUs. Partner adds turnkey bare metal with GPUs, while Direct offers single tenant enterprise hardware.
Torus already has thousands of GPUs deployed in Foundry units. The company plans to have more than 47,000 GPUs deployed by the end of 2027. Its production line is online and will be capable of producing 500 megawatts of IT load, with plans to reach 1 gigawatt.