U.S. residential solar provider Sunrun (RUN) recently announced that it will collaborate with Tesla (TSLA) and Renew Home to supply power support to data centers and utility companies across the United States. Boosted by this news, Sunrun’s shares surged as much as 31% intraday on Wednesday before closing up 12.57% at $14.42.
Under the agreement, the three parties will jointly establish a power supply platform that integrates hundreds of thousands of home batteries operated by Sunrun and Tesla, along with 8 million thermostats managed by Renew Home, forming a total power capacity of 16 gigawatts (GW). This scale is sufficient to meet the electricity needs of approximately 12 million U.S. households. The partners stated that the power will be supplied on a “first-come, first-served” basis, with priority given to artificial intelligence (AI) data center developers, while also being made available to PJM Interconnection, the largest grid operator in the United States.
The rapid development of the AI industry is driving a sharp increase in electricity demand, posing a severe test to the existing U.S. power grid. The collaborating companies pointed out that current grid infrastructure is primarily designed to handle peak loads that occur only during a few periods throughout the year, resulting in a large number of power generation assets sitting idle for extended off-peak periods. A recent report from the Brattle Group consulting firm indicates that if the potential of the existing grid can be more fully tapped, U.S. electricity bills could be reduced by $110 billion to $170 billion over the next decade, while also significantly shortening the interconnection waiting time for data centers.
This year, as the lucrative tax credit policy for solar panels has expired, residential solar providers have generally faced downward pressure on sales. In response, home energy companies have begun to adjust their strategies and actively seek business opportunities from the ever-growing electricity demand of data centers. At present, power supply has become a key bottleneck constraining the pace of AI expansion in the United States, and the ability to secure stable power sources ahead of competitors is emerging as a critical factor for technology companies to gain an edge in the AI race.
According to statistics from Wood Mackenzie, there were 1,940 active virtual power plant (VPP) deployments in North America in 2025, a year-on-year increase of 33%, with total capacity reaching 37.5 GW. Google has signed a three-year 100 MW agreement with VPP operator Voltus, adopting a “bring-your-own-capacity” model to aggregate load resources within PJM grid coverage areas such as Northern Virginia, where data centers are densely concentrated. At this juncture, electricity has replaced computing chips as the primary limiting factor determining the speed of AI expansion in the United States.
