Two phases
The development of virtual power plants in China can broadly be split into two phases. The first is subsidy-driven via demand response, where the core of the business is aggregation — whoever signs up more flexible load earns more subsidy. The barriers in this phase are mainly business development and hardware integration.
The second phase arrives as spot and ancillary-service markets open up and revenue becomes market-based. What then determines return is no longer the size of the contracted portfolio but forecast accuracy and trading strategy: with the same 100MW of resources, an operator who forecasts well pays materially less in deviation penalties and captures more arbitrage in price windows.
Where the weight of capability moves
This shift implies three things for operators:
- Data infrastructure must be built ahead of business scale — without clean historical load and output data, there is nothing for algorithms to work with.
- The algorithm team moves from cost center to profit center.
- Revenue management has to be visualized, so resource owners can see what they contributed and the relationship holds.
An often-ignored problem
Discussions of the business model tend to overlook retention on the resource side. Behind every flexible load is a real production plan; if dispatch requests disrupt production too often, the resource owner will not renew next year. A good dispatch strategy therefore optimizes for more than single-transaction yield — response frequency and production constraints belong in the objective function.
This column is industry commentary and technical opinion, and does not constitute investment advice.