Sunrun (Nasdaq: RUN) announced on July 8, 2026, the launch of a distributed AI compute pilot that repurposes its vast network of home solar and battery systems into a decentralized inference compute platform. The initiative enables Sunrun to sell AI inference capacity to enterprise customers while testing operational performance across varied conditions and rate structures. Participating homeowners will be compensated for hosting compute nodes, creating a novel revenue stream for Sunrun and a scalable edge compute infrastructure for AI workloads.
This move addresses the growing demand for AI inference, which McKinsey projects will surpass training workloads by 2030, accounting for over half of all AI compute. Unlike tightly coupled training clusters, inference workloads are modular and latency-sensitive—making them ideal for edge deployment. Sunrun’s existing distributed footprint offers a structural advantage over traditional data centers, enabling rapid deployment of inference capacity without the delays of permitting and construction.
Paul Dickson, Sunrun’s President and Chief Revenue Officer, emphasized the company’s two-decade expertise in scaling distributed assets and its readiness to bring compute closer to energy sources and inference workloads. This pilot marks a strategic expansion of Sunrun’s business model, blending energy infrastructure with AI compute services.