Sunrun plans home-based AI compute pilot for solar customers
Sunrun says it will pay some solar and battery customers to host small compute nodes that can be sold as AI processing capacity.
By Hana Yoshida · Markets Reporter
3 min read
Sunrun is testing a plan to put AI computing hardware inside customers’ homes instead of concentrating it in a traditional data center. The company says the pilot will use homes that already have Sunrun solar panels and battery storage, a model that could offer AI companies another way to get computing capacity as data center projects draw local opposition.
The home energy company described the effort as a distributed AI compute program. Sunrun says it will place many small compute nodes in participating homes and pay customers who take part in the pilot.
Sunrun says it plans to package the processing power from those home-based nodes and sell it to enterprise customers, including companies that need AI compute. The company is calling the broader idea a nationwide compute network, according to its announcement.
A data center spread across homes
The plan differs from the standard approach used by large AI infrastructure projects, which often group servers in purpose-built facilities. Sunrun’s version would spread the hardware across many residential locations connected to solar and battery systems.
Sunrun says it has already completed a proof-of-concept test and described that work as successful. The company has not yet shown how the system would perform at larger scale, and The Verge reported that it remains unclear how well the approach will work.
The pilot is open to Sunrun customers who are willing to host a compute node. Sunrun says its customer base totals 1.1 million, and the company has opened a waitlist for those interested in the program.
The company says it expects to finish the pilot over the coming months. After that, Sunrun says it will review the results before deciding whether to expand the program more broadly.
Data center pressure shapes the pitch
The proposal arrives as AI data centers face public scrutiny over their effects on surrounding communities and local infrastructure. The Verge reported that a May survey found more than 70 percent of Americans oppose new data center construction near where they live.
The concerns cited in that reporting included pollution, noise, and the use of water and electricity. Those issues have become a recurring problem for companies trying to add computing capacity for AI systems.
Sunrun’s pilot does not eliminate the need for hardware or power, but it changes where the equipment would sit. Instead of asking one community to accept a large facility, the company wants to distribute smaller devices among customers who already have home energy systems.
The effort also puts Sunrun in a business it has not traditionally operated in. The company is best known for residential solar and home battery storage, while AI compute services are a different market with different technical and commercial demands.
For participating customers, the immediate offer is compensation for hosting equipment. Sunrun has described the pilot structure and the intended buyers for the compute capacity, but the company has not detailed in the reported materials how much customers would be paid.
This story draws on original reporting from The Verge.