Sunrun will offer customers payments to host AI compute nodes in their homes

On July 10, 2026, Sunrun announced a pilot program for distributed AI computing: the company will place compute nodes in the homes of customers with solar panels and energy storage systems, and will pay participants for the use of the equipment. Sunrun intends to sell the resulting capacity to corporate buyers of compute, including AI companies.

How the program will work

Sunrun's pilot involves placing many compute nodes in homes that already have the company's solar energy systems and batteries. Instead of building a standalone data center, Sunrun wants to spread compute infrastructure across residential sites.

• The program is called distributed AI compute
• Nodes will be placed in Sunrun customers' homes
• Pilot participants will receive compensation
• The compute resources are planned to be sold to corporate buyers
• Potential customers include companies that develop AI

The program description does not say what hardware will be installed in homes, how much compute each node will have, or how much participants will be paid. The scale of the pilot and the exact timeline for expanding it have not been disclosed either.

Why Sunrun needs homes

Sunrun sees residential homes as an additional site for AI infrastructure. The approach could make use of power generation and storage equipment that is already installed, but the announcement does not disclose how the compute nodes would actually be connected.

The company wants to sell the combined capacity of the distributed network to "corporate buyers of compute." That category includes AI companies that need resources to run models and services. Sunrun is effectively trying to merge two infrastructures: home energy systems and compute capacity spread across many addresses.

"Distributed compute" is meant to place numerous nodes in homes equipped with

Sunrun solar panels and battery systems.

For the market, this is an unusual alternative to traditional data centers. AI compute is usually concentrated in large purpose-built facilities, where power, cooling, security and network connectivity are easier to manage. In Sunrun's case, some of those tasks shift to the level of individual homes, but the operating conditions and requirements for the homes have not yet been described.

What questions remain

The main question is whether a distributed network in residential homes can give corporate customers predictable performance. AI workloads depend on stable power, high network bandwidth, equipment cooling and constant node availability. Sunrun's description contains no data on how the company will handle these problems.

It is also unclear what restrictions homeowners will face. Compute hardware can draw a significant amount of power, give off heat and add load to home infrastructure. Sunrun says only that pilot participants will be compensated, but gives no financial terms.

For now, the program remains an experiment. The company has announced its intention to test a model in which compute resources are assembled from many small nodes and then sold as a single service to corporate customers.

What this means

Sunrun is trying to turn homes with solar panels and batteries into building blocks of distributed AI infrastructure. If the pilot confirms the technical and economic viability of the approach, energy companies could become compute providers, not just electricity suppliers.