Particle.news

Sunrun Launches Pilot to Turn Homes Into a Distributed AI Inference Network

The company will pair small servers with home batteries to run resilient, behind‑the‑meter AI workloads during a months‑long test.

Overview

  • Sunrun announced the pilot Wednesday to install AI compute nodes in participating customers' homes and pay hosts to run inference workloads on the devices.
  • The program builds on Sunrun's roughly 1.1 million residential solar and battery systems, which the company says will form the backbone of the distributed compute network.
  • Compute nodes are designed to connect to each home's battery so workloads can continue during certain power outages and reduce strain on the grid in specific locations.
  • Sunrun says it is already talking with enterprise AI buyers, homebuilders, and utilities about commercial deals while it runs the pilot and expects to evaluate results in the coming months before deciding on scale.
  • Scaling beyond the pilot depends on participant enrollment, signed offtake contracts and utility and regulatory approvals, and the effort follows a related June proposal to aggregate residential batteries for large‑scale capacity.