Residential solar and battery systems across America are being repurposed as a distributed power grid for AI data centers, allowing tech companies to bypass traditional utility infrastructure and accelerate their buildouts by years. Sunrun, the nation's largest residential solar provider, has partnered with Voltus, a distributed energy platform, to supply capacity from thousands of home battery systems in the PJM and MISO grid regions to support AI data center developers through Voltus's Bring Your Own Capacity (BYOC) program.

This partnership represents a fundamentally different approach to solving AI's insatiable appetite for electricity. Rather than waiting for utilities to build new transmission lines and power plants, hyperscalers like Google are now aggregating existing distributed energy resources (DERs) from homes and businesses to create their own capacity stacks. Google has already signed a 100-megawatt BYOC agreement with Voltus to support operations across the PJM grid.

Why Are Data Centers Turning to Home Batteries?

The explosion of AI hyperscale data centers has created an unprecedented energy crisis. Data centers across the United States consumed about 176 terawatt-hours of electricity in 2023, roughly equivalent to all of Ohio's annual electricity usage and about 4 percent of all electricity consumed in the country. That share is expected to nearly double to 9 percent by 2030. The Pennsylvania-based PJM Interconnection, which manages electricity transmission across 13 states and the District of Columbia, recently identified data centers as "the primary driver" of growth in electricity demand, which has driven up prices for consumers.

Traditional interconnection processes can take years, requiring utilities to upgrade substations, transmission lines, and generation capacity. The BYOC model sidesteps this bottleneck by allowing data center developers to bring their own pre-aggregated capacity to utilities, enabling approvals and buildouts years ahead of schedule. Voltus claims this approach maximizes value for end users while turning distributed resources into capacity the grid can reliably count on.

How Does the Distributed Energy Model Work for Data Centers?

• Aggregation at Scale: Sunrun and Voltus are drawing on more than 16 gigawatts of residential energy capacity across the United States, including home battery systems, smart thermostats, and vehicle-to-grid devices already installed in homes.
• Market Integration: Voltus operates across all nine wholesale electricity markets in North America, with more than 7.5 gigawatts of distributed energy resources already in operation, allowing it to deliver market-accredited capacity that utilities recognize as reliable.
• Hyperscaler Funding: Tech giants provide funding to support residential battery installations, creating a financial incentive for homeowners to participate while simultaneously building the power infrastructure data centers need.

The Sunrun-Voltus partnership is the second major deal between the companies. In June, they joined with Tesla to aggregate the 16-gigawatt capacity portfolio for sale to hyperscalers and utilities. This represents a shift in how energy infrastructure is being financed and deployed in the United States.

What Are the Broader Implications for Energy Policy?

The rise of distributed energy models for data centers is reshaping political debates around AI infrastructure. In Illinois, Governor JB Pritzker initially celebrated data center tax incentives in 2019 as economic drivers, but has since paused those incentives as hyperscale facilities stress energy and water resources. Polling conducted on behalf of the Illinois Clean Jobs Coalition found that 68 percent of respondents would support regulations on data centers "to minimize their impact on our utility bills, climate and water while still allowing them to be built".

"Meeting growing energy demand requires us to maximize every single electron available across the country. In collaboration with Voltus, we are providing critical capacity from home batteries supported by funding from hyperscalers. This is just the beginning of what distributed energy assets can achieve," said Mary Powell, CEO of Sunrun.

Mary Powell, CEO of Sunrun

Pritzker's shift reflects a broader recognition that AI hyperscale data centers represent a different category of infrastructure than traditional data centers. While smaller facilities have existed for decades, the new hyperscale facilities required for artificial intelligence demand significantly more energy, water, and space. Some of the largest proposals would consume half the energy of the entire city of Chicago and occupy a space as large as Central Park.

The environmental and health impacts of this rapid expansion remain poorly understood. Ted Smith, an environmental health researcher at the University of Louisville, noted that much of a data center's environmental impact depends on where it is sited, the cooling system it uses, where it gets power from, and other complex factors. "I don't know what the health effects are of living next to a hyperscale data center. Unfortunately, I don't know anyone who knows the effects of living next to a hyperscale data center," Smith explained.

"When we think about data centers, we need to try to find a middle road, and increasingly put more effort into using the capacity of the AI enabled by the data centers to help address environmental and human health problems," said Jeff Bielicki, Associate Research Director at Ohio State University Sustainability Institute.

Jeff Bielicki, Associate Research Director, Ohio State University Sustainability Institute

The distributed energy approach championed by Sunrun and Voltus may offer a partial solution to the grid stress problem, but it does not eliminate the underlying challenge: data centers powered by fossil fuel plants still generate emissions that contribute to respiratory illness and climate change. However, by leveraging existing residential infrastructure and renewable energy sources, the BYOC model could reduce the need for new fossil fuel generation capacity while accelerating the deployment of AI infrastructure that policymakers and the public increasingly view as essential to economic competitiveness.