Dispatchable Compute Resource

Turning compute into a grid asset.

Project Ohm builds and runs distributed AI compute, moving the work in time and place to follow surplus, low-cost power.

Step 1 of 6. 12:30, Midday. Rooftop solar floods the street. The substation is under pressure.

  • Own node
  • Partner node
  • Third-party compute

Why now

Power grids are hitting a wall with AI. We build compute that balances the grid and solves bottlenecks, instead of creating them.

00.2How it works

How it works

  1. 01

    At midday, roofs across a suburb send power back through the same wires that deliver it, and a wind farm on the edge of town adds more when it gusts.

  2. 02

    Local substations were built for one-way traffic. When too much flows back, the network has to turn away clean power or pay for upgrades.

  3. 03

    A customer sends a job. The platform weighs every place it could run on price, spare power and deadline. Best for the customer is best for the grid.

  4. 04

    The job runs on power the network would turn away. The substation gets relief, the customer gets low-cost compute, and soon a carbon receipt.

  5. 05

    Like a battery, the fleet soaks up surplus power and steps back when supply is short. To the grid, it is one Dispatchable Compute Resource.

  6. 06

    Through the day, the work follows the cheapest power across the country: midday solar, overnight wind. The same platform works on any grid.

00.3Proof

What we claim, and what backs it.

00.4Team

Dedicated experts building grid-aware AI compute.

00.5Next steps

  1. Scale the network with us.

    Qualified investors can request the investor briefing.

    Request the investor briefing
  2. Bring flexible demand to your network.

    Network operators, utilities, councils and policy teams: see how schedulable compute uses power that would otherwise be curtailed, and eases pressure on substations.

    For grid stakeholders