Deploy sovereign inference compute

Your questions, answered.

• Is Torus Foundry a data center?

  Torus Foundry is modular AI compute that runs 1 to 20 MW per site. It’s a compact, self-contained system, built and prefabricated in Utah, that installs on already-developed land and uses available grid headroom. Every Foundry is paired with Torus Reserve, so it adds steady demand while helping the local grid stay reliable. Instead of one massive campus, inference capacity can be added one site at a time.

• How is Foundry different from other modular data centers?

  Foundry pairs compute with energy storage and deploys on existing grid interconnections. Our energy storage peak shaves and load shifts to unlock additional headroom. This translates to more IT load on every site when compared to other MDC products. Each unit arrives pre-tested and fully assembled, and installation and commissioning take 1-2 days.

• Does it require a grid upgrade or new construction?

  No. Torus builds and assembles Foundry in our manufacturing facilities and ships it ready to run. Because it uses existing land and infrastructure, you skip the multi-year wait for a utility upgrade.

• How does Foundry protect my data?

  Each Foundry is built for one customer. No other tenants share your GPUs, network, or power, so your workloads run on hardware that only your team uses. Foundry’s small footprint also means fewer people and fewer systems touch your site. Security and monitoring come built into every unit.

• What type of security is included with Foundry?

  Torus Protect is the cybersecurity layer built into every Torus system. It includes a Layer 7 stateful firewall with Zero Trust, so every connection is verified before it gets access to your equipment. Data is encrypted in transit with TLS 1.3 and at rest with AES-256, so it stays protected as it moves and while it’s stored.