Solving today's energy challenges

Energy demand is growing exponentially — and today's grid wasn’t built for what's coming. Torus was founded in 2021 in Springville, Utah to close that gap with the world’s first mesh energy infrastructure to power more resilient businesses, utilities, and communities.

The point of load is wherever power gets used: a factory floor, an office building, a neighborhood. The same problems show up over and over:

1. New load wants to connect. Every new building or business needs a connection to the grid, and each one has to be studied and built before it can draw power. For large users, utility service upgrades take 18 to 36 months and can cost millions.
2. Power has a long way to travel. Typically, power is generated hundreds of miles from where it's used. Getting it to the point of load means moving it over transmission lines and through substations that are often under stress.
3. Peak demand strains the system. On a hot afternoon, when everyone runs their air conditioning at once, utilities turn to the highest-cost resources available to keep up. That usually means firing up peaker plants that are high in cost and CO2 emissions. 
4. Loads ramp faster than the grid can serve. GPUs and advanced manufacturing equipment can swing from low power to high power in moments. Those rapid swings can make the local grid unstable.

Utilities are already using this technology at scale. Our systems are installed at commercial and industrial customer sites across Rocky Mountain Power's service territory, and the utility has dispatched them more than 450 times this year. Each dispatch is the utility signaling our systems to respond at a specific moment to support the grid. Median ramp time in PacifiCorp demand response events is under 20 seconds, with a 99.9% critical path uptime across assets. 

Foundry is modular inference compute that installs on existing grid interconnections. It sits where the grid has pockets of excess capacity, so chips land where the power already is.

Torus Reserve battery storage load shifts to ensure the lowest-cost spare capacity reaches Foundry, and it shaves peaks so Foundry doesn't add stress to the grid. It also acts as a shock absorber when loads ramp, so the edge of the grid becomes more reliable.

Because Foundry runs on capacity that already exists, it deploys in 12 to 16 weeks.

We're heavily committed to American manufacturing. Our full product stack is built at GigaOne, our 545,000-square-foot production facility. We're BABA-compliant and prioritize local Utah partners and domestic supply chain vendors. We employ over 200 manufacturing techs, and that number is growing.