Torus Raises $500 Million To Run AI Data Centers On Modular Battery Systems

Torus, a Utah-based startup, has secured $500 million in a Series C round to scale its modular AI data center system, Foundry, led by Magnetar through its investment arm Magnetar. Other investors include ICONIQ, Origin Ventures, Ridgeline Partners, and Trinity Capital. The company, which didn't disclose its valuation, was founded by CEO and founder Nate Walkingshaw, who initially focused on modular battery systems for factories and businesses to optimize energy usage and reduce power costs. In 2025, Magnetar invested $200 million into Torus.

Foundry units are prefabricated containers that provide power, storage, cooling, networking, and computing capabilities. A single unit houses 576 Nvidia HGX B300 GPUs in approximately 577 square feet and operates at 1.2 megawatts. Companies can easily expand capacity by adding more Foundry units, with a range of sizes from 1 to 20 megawatts. Torus's unique approach allows it to bypass the usual delays in obtaining land and grid connections by bringing the chips to the power, rather than the other way around.

The company's Reserve product, a flywheel and battery system, manages demand spikes and maintains operations during outages. Cooling is accomplished through a closed-loop system that uses no water. Torus claims that the time from contract to live workloads is 12 to 16 weeks, with site preparation taking about a week and installation occurring within one or two days.

The company targets inference over training, as the demand for AI data center inference is expected to grow by 35% annually until 2030, compared to a 22% increase for training during the same period. Torus currently has thousands of GPUs deployed in its units and aims to have over 47,000 by the end of 2027.