Torus raises $500m Series C, launches modular AI data centers

Energy storage developer Torus has closed a $500 million Series C round led by Magnetar, and paired the raise with the launch of Foundry, a prefabricated modular AI data center product it says already runs thousands of deployed GPUs.

The Sandy, Utah-based company will use the capital to scale Foundry, a modular solution rated at 1–20MW per site and delivered in shipping-container-sized units. Each unit ships with Torus Reserve, the company's flywheel-and-battery storage system, which Torus claims smooths AI inference loads (the phase where a trained model answers queries), letting more capacity run on an existing grid connection while doubling as backup power during outages.

How does deployment actually work?

The units install behind the meter, meaning on the customer side of the utility connection, using headroom at existing transformers. That approach, the company argues, avoids building new power and generation capacity — the slowest and most expensive part of most AI infrastructure projects.

Site preparation takes roughly one week. Installation and commissioning take one to two days. The full path from signed contract to operations runs 12 to 16 weeks — a fraction of typical timelines for net-new data center construction.

Torus prefabricates every unit at its 545,000 sq ft (50,630 sqm) factory in Salt Lake City. The units are vertically integrated, combining power, storage, cooling, networking and compute in one box. Cooling is closed-loop, and because the system runs off existing grid capacity, no new generation needs to be built alongside it.

What can customers buy?

Foundry comes in three configurations:

• A powered shell without GPUs
• A turnkey bare-metal unit with GPUs installed
• Dedicated single-tenant inference capacity

The first announced customers are GPU cloud company Hydra Host and a second, unnamed AI cloud provider.

Torus CEO and founder Nate Walkingshaw framed the product as a challenge to how the industry has approached AI's power problem. "The industry has been trying to solve AI's power problem by building net-new infrastructure, which causes costs to potentially rise and takes years to bring compute online," he said. "Inference doesn't have to wait. We've spent years learning how to make the grid more affordable, flexible, resilient, and intelligent, and Foundry applies that to compute: capacity in weeks, with power that's already available. Most providers are bringing power to the chips. Foundry brings the chips to power."

What's the scale of the bet?

Thousands of GPUs already run in Foundry units, the company said. Torus plans to push that past 47,000 GPUs deployed by the end of 2027.

The Salt Lake City production facility will be capable of producing 500MW of IT load (the power delivered to computing hardware), with plans to expand to 1GW in the near future. That would make the plant one of the larger modular data center production lines in the US if the expansion lands.

The $500m Series C, led by Magnetar, gives Torus the balance sheet to chase that ramp. The bet is straightforward: inference demand grows faster than utilities can build new capacity, and the winning providers will be those who squeeze compute onto power that already exists.