AI Tech Drives Virtual Power Plant Scale-Up, Top Firms' Trading Revenue Near 60%, 49 Projects Won Jan-Jul 2026

The deep integration of AI technology, coupled with policy support, has driven a qualitative transformation in the virtual power plant industry. The year 2026 is regarded as the first year of large-scale deployment for virtual power plants. Almost all currently operational large-scale virtual power plants have embedded AI technology. Mainstream new-build and upgrade projects have generally integrated AI, with AI penetration exceeding 80% in large-scale operation platforms. National and provincial demonstration projects have essentially achieved closed-loop AI dispatch. In terms of core technical indicators, load forecasting accuracy averages between 90% and 94%, with advanced scenarios at leading companies reaching 95% to 96%, and response times at the millisecond level.

Profit models are transitioning from being driven by policy subsidies to being driven by AI trading profitability. The profit structure of leading companies and pilot projects has diversified, with trading revenue accounting for nearly 60% of total revenue. The industry has developed three main profit streams: electricity trading, ancillary services, and demand response. Electricity trading revenue is generated by arbitraging peak-valley price spreads through participation in medium- and long-term and spot markets. Ancillary service revenue comes from providing frequency regulation, peak shaving, and other services. Demand response revenue is obtained through competitive bidding clearing or time-of-use pricing mechanisms to participate in load shifting and peak shaving. With deep application of AI technology, arbitrage profits in the electricity spot market have increased by an average of 10% to 15%, response speed has improved by over 60%, and dispatch costs have been reduced by approximately 35%.

The National Development and Reform Commission and the National Energy Administration have set clear targets: by 2027, the national virtual power plant regulation capacity should be no less than 20 million kilowatts, and by 2030, no less than 50 million kilowatts. Since 2026, the scaling of virtual power plants has accelerated, with a significant year-on-year increase in the number of procurement and bidding projects. From January to July 2026, the national public resource trading platform announced 49 winning bids for virtual power plant-related procurement and bidding projects, with an additional 8 projects still in the bidding process, both showing notable year-on-year growth. Project scales have jumped from the million-yuan level to the billion-yuan level, with the share of large-value projects rising by over 60% year-on-year.

Multiple companies have indicated that the pace of large-scale virtual power plant deployment is accelerating. Some companies saw year-on-year growth of over 200% in related orders in 2025, and the trend continued into the first quarter of 2026. Some companies have shifted from project delivery to a sustainable model of 'software delivery plus operational revenue sharing', enhancing cash flow stability. Starting in 2026, a large number of private enterprises have entered the queue for virtual power plant qualification applications, and the industry is transitioning from scale expansion to quality competition.