Torus raises $500 million to expand modular AI data centers using grid power
Also covered by 2 sources
What happened
Torus, a Utah-based company, secured $500 million in funding to expand its modular AI data center product called Foundry. The financing, led by Magnetar, will be used to grow Torus’s Salt Lake City manufacturing facility and to deploy Foundry units across the United States. Foundry units integrate batteries, computing hardware, and cooling systems in prefabricated enclosures that use existing electrical grid capacity at business sites. Initial deployments will serve customers such as Hydra Host and an unnamed AI cloud provider.
Key details
- Foundry units can deliver 1–20 megawatts of AI inference capacity per site.
- Deployment from contract to operational status takes 12–16 weeks.
- The Salt Lake City factory, GigaOne, has 545,000 square feet and nearly 400 employees.
- Foundry uses a waterless cooling system that reduces water use by 95% compared to traditional methods.
- Torus aims to deploy over 47,000 GPUs in Foundry units by the end of 2027.
Why it matters
Businesses and AI infrastructure providers can now install AI computing capacity more quickly at existing sites without building new data centers or upgrading grid connections. This reduces the time and resources needed to scale AI workloads and allows companies to use available power more efficiently. Customers benefit from faster deployment, lower water usage, and flexible options for integrating AI hardware into their operations.