Torus, a Salt Lake City-based manufacturer of distributed power and compute infrastructure, launched a modular AI infrastructure product called Foundry on Wednesday, Nov. 11, 2026, and announced it has secured $500 million in growth capital. The new system is designed to meet rising demand for AI inference by utilizing existing power capacity in the electrical grid, aiming to deliver compute capacity in as little as 12 weeks.
Modular Infrastructure Design
The Foundry system is a prefabricated, modular solution that packages power, storage, cooling, networking, and compute into a single unit. Ranging from 1 to 20 megawatts per site, the units are roughly the size of a small cargo van and require no new building construction. According to the company, the technology allows for deployment at individual transformers by using available headroom or electricity behind the meter, avoiding the infrastructure upgrades that often lead to utility rate increases.
Addressing Power Bottlenecks
Nate Walkingshaw, founder and CEO of Torus, stated that the product represents a shift in how AI infrastructure is deployed. While most providers focus on bringing power to the chips, Foundry is designed to bring the chips to the power that is already available. Walkingshaw noted that the computing containers can be installed at businesses or factories where grids have excess power capacity, helping to avoid expensive infrastructure upgrades that put pressure on utility bills.
Rapid AI Deployment
The launch arrives as data center developers in the United States face increasing challenges in securing enough power for energy-intensive AI applications. While traditional hyperscale campuses can take years to build, Torus claims Foundry can go from contract to running AI workloads in 12 to 16 weeks. The company currently has thousands of GPUs deployed and aims to have more than 47,000 GPUs deployed in Foundry units by the end of 2027.
Expansion and Funding
The $500 million Series C funding round was led by Magnetar, with participation from ICONIQ Capital, Origin Ventures, Ridgeline Partners, and Trinity Capital. Torus intends to use the capital to fund the continued expansion of GigaOne, its Salt Lake City-based manufacturing facility, and to support Torus Reserve and Torus Foundry deployments across the United States.
Utah Manufacturing Growth
The company's Salt Lake City manufacturing facility, GigaOne, spans 545,000 square feet. GigaOne opened its first production line in early 2026 and has been opening a new production line every 21 days throughout 2026. The Foundry production line is currently online with a capacity to produce 500 MW of IT load, though the company has plans to expand that capacity to 1GW in the near future.
Inference and Grid Stability
Torus's technology is designed to address the specific needs of AI inference, which is projected by McKinsey to grow 35 percent annually through 2030, compared to 22 percent for AI training. Unlike training, inference compute runs continuously and serves users in real time, which benefits from being distributed closer to demand sites. To support this, every Foundry unit is paired with Torus Reserve, a product providing energy storage, management, and grid-enhancing technologies such as inertial-based firm power and load-shaping recipes.
Sustainability and Customers
The company's design includes a closed-loop cooling system that is waterless, using 95 percent less water than evaporative coolers. Additionally, Torus stated that each system donates GPU time to medical research through Folding@home. The company's first customers for the Foundry product include Hydra Host and another unnamed AI cloud provider.
