Virtual Power Plants Reach Utility-Scale Dispatch Numbers

An RMI analysis published on 14 May 2026 gathers the dispatch figures that decide whether aggregated distributed energy counts as a grid resource or as a pilot. California's Demand Side Grid Support programme reached 1,145 MW as of October 2025, of which 768 MW is the market-aware storage component; the older Emergency Load Reduction Program stood at roughly 800 MW as of 2023.

The single-event numbers are the more persuasive ones. During the heat dome of 24 June 2025, Sunrun deployed 340 MW while EnergyHub shed 900 MW of peak load and shifted 3.5 GWh. A CAISO test event on 29 July delivered more than 500 MW of demand relief during the afternoon net-load peak. In Massachusetts, National Grid's ConnectedSolutions programme holds 227 MW, with 375 MW dispatched region-wide during the June 2024 heat wave; that programme launched in 2016.

Commitments elsewhere point the same way. In Texas the NRG and Renew Home partnership reached 150 MW as of 2025 against a 1 GW target by 2035, and CPS Energy's pilot exceeded 250 MW across 175,000 customers as of 2024. Dominion Energy has been directed to develop a 450 MW pilot in Virginia, a 125 MW Xcel proposal was approved in Colorado, Xcel is procuring 200 MW of distributed storage in Minnesota, and Georgia Power has a 100 MW solar-and-storage procurement.

For anyone building the device and gateway layer underneath, these figures reframe the engineering problem. At a few megawatts the question is whether you can reach the device. At several hundred, dispatched inside an event window and settled afterwards, the questions are fan-out, latency and telemetry integrity: can you reach a hundred thousand endpoints inside the window, confirm that they responded, and prove it later to a settlement process. That is a different system from a monitoring platform, and it is the one the programme numbers now require.