The conversation about AI’s power demand has mostly focused on one question: how do we build enough generation and transmission to feed massive centralized data centers? A Lightning Stage session at DERVOS asks a different one: what if the compute went to the energy instead?
We’re excited to announce “Bringing Distributed Compute Benefits Home,” a session exploring how residential solar and storage could become the foundation of a decentralized edge computing network.
Sunrun launched its distributed AI compute pilot in 2026, with the goal of turning residential and multi-family homes equipped with solar and battery storage into a decentralized edge computing network. In September 2026, Sunrun and SPAN expanded their partnership to pursue a gigawatt-scale, behind-the-meter distributed AI compute solution using customer-sited solar-plus-battery systems.
The model starts with new construction. Sunrun and SPAN are working with residential communities built from the ground up, beginning in Texas and other select markets. In these homes, the solar and battery systems power both the household and the AI compute nodes locally.
The pitch is that this approach avoids the transmission overloads and generation bottlenecks typical of traditional centralized data centers, and sidesteps the multi-year timelines and high capital costs of building large-scale facilities.
We can’t wait to hear more.
Henry Ford made a lot of money selling Model T's. Nobody would drive a model T today. Sunrun is a finance and sales engine. They are years behind technologically. Chasing money during a revolution is what Torries did back in the day. This moment deserves better solutions. But I'm grateful that even bumbling behemoths realize the vast potential of giving someone access to power and compute given they are likely the only predictable things to retain value into the future. Universal energy and universal compute is a way better solution than universal basic income.