Lunar Energy delivers complex distribution network capacity services

Pacific Gas and Electric Company (PG&E), a utility serving more than 16 million people in California, wanted proof that a virtual power plant (VPP) of home batteries could relieve stress at particular points in its local distribution grid. Working with our partner Sunrun, Lunar proved that we could do so reliably and in multiple different locations at once.

The challenge

California's grid faces strain at the local level throughout the year, intensifying on hot summer evenings. Markets were already expecting a significant increase in demand peaks due to heating and transport electrification; the prospect of new data centres has only turbocharged the issue. The speed (and cost) of getting interconnections for new large loads is front of mind for many, as is the potential increase in network costs borne by all electricity consumers that could result from those new connections. 

PG&E wanted to establish that distributed energy resources such as home batteries could relieve temporally and geographically discrete distribution network congestion, setting the stage for a more dynamically managed distribution grid that achieves speed-to-power while keeping costs down.

The solution

Lunar and Sunrun participated in the SAVE program, a first-of-its-kind seasonal aggregation pilot program for PG&E, running from May to October 2025 across 500+ devices and 25 substation groups. 

Lunar's Gridshare Platform generated daily forecasts of available flexible capacity per home, and then used these to create a forecast of flexible capacity per substation on a 30-minute basis. It provided these forecasts to PG&E, who then provided Gridshare dynamic, 30-minute interval dispatch profiles per substation. Gridshare then delivered against those unique dispatch profiles automatically, tracking real-time performance against target, and adjusting as conditions changed (all without manual intervention).

The results

Over the course of the pilot, the fleet consistently held close to its dynamic target power profile. That's a harder problem than it sounds: Lunar’s Gridshare Platform was not dispatching Lunar hardware, but 3rd party OEM hardware. Another program participant controlled its own devices end to end, both hardware and software. In an independent assessment, Lunar's performance was compelling:

• Energy delivery percentage: Lunar under-delivered on target energy by 9.2%, compared to 15.5% under-delivery for the other participant. Given the complexity of the use case, and the fact that Lunar was dispatching 3rd party hardware, 9.2% under-delivery is a very strong performance. 
• Power target tracking: Lunar was measured at 0.30 RRMSE for the 3rd party devices it was dispatching, essentially the same as the other participant’s 0.27.

What's next

The SAVE program gave PG&E real evidence that a fleet of home batteries can reliably relieve grid stress based on locationally discrete needs; the same precision is now needed to connect the grid's largest new loads faster.

PG&E is already building its next commercial model for exactly this problem. Flex Connect is designed for large, flexible loads like data centres and EV fast-charging sites: instead of waiting years for a capacity upgrade, a site connects sooner and adjusts its consumption against a dynamic power limit PG&E sends a day ahead. 

Lunar's technology is already built for that kind of precision. If your grid is under pressure from hyperscalers, new developments, or anything else showing up faster than your infrastructure can handle, we're ready to talk.

{{testimonial-1}}