Sunrun has expanded a proof of concept into a US pilot that installs AI compute nodes in homes with its solar-and-battery systems, aiming to sell inference capacity while testing performance under different power-rate structures.
The company says the model could use existing household energy infrastructure to ease grid strain while systems still provide backup power and participate in grid-services markets; it did not disclose the pilot’s scale.
VivoPower, in a parallel Atlantic-side push, said a BESS at its 41.5MW AI-ready data-centre campus in Mo I Rana, Norway, could generate up to $4 million in EBITDA from Nordic reserve markets.
That Norway site benefits from roughly $0.009/kWh local power prices and hydro supply, though ancillary-services revenues can fall as more battery projects enter the market.
Both efforts reflect a broader search for battery-linked AI infrastructure models that can deliver faster power access, manage volatile AI loads and add revenue beyond traditional renewable-energy use.
Will renting your home battery to AI companies actually lower your electricity bills?
As AI's energy demand soars, who will truly control the power grid of the future?
Is battery storage too slow to protect the power-hungry GPUs driving the AI boom?
Powering the AI Boom: How Distributed Solar Homes and BESS Are Reshaping Data Center Energy for 150 GW U.S. Demand by 2028
Overview
In July 2026, Sunrun launched a bold pilot to turn its 1.1 million solar-powered homes into a distributed AI data center by integrating AI compute nodes directly into residential settings. This approach uses the homes’ existing solar power generation and storage to create a decentralized computing network. After a successful proof of concept showed real demand and revenue potential, Sunrun is now running a pilot to assess performance and homeowner experience. The company is also working with enterprise partners and utilities to build commercial frameworks, aiming to expand this innovative model if the pilot meets its goals.