On May 12, Securities Times republished a Securities Daily article describing how compute-power coordination is reshaping China’s green-power industry. The article linked the shift to the May 8 action plan on artificial intelligence and energy issued by the NDRC, NEA, MIIT and National Data Administration. It also cited the early-May operation of what it described as China’s first large-scale compute-power coordination green-power direct-supply project: Datang’s 500 MW photovoltaic station serving the Zhongwei cloud base in Ningxia, using both physical direct supply and bilateral power trading.
The event matters because it moves green electricity from a commodity story to a service story. A wind or solar project used to be assessed mainly by capacity, tariff and utilization. For AI infrastructure, the buyer needs something more complex: stable supply, credible low-carbon attributes, location compatibility, storage support, backup arrangements and possibly cooling or waste-heat solutions. The product is no longer simply electricity. It is a managed energy environment for high-load digital infrastructure.
The policy language creates three business-model changes. First, supply and demand matching becomes smarter. The article says compute-power coordination can link wind, solar, storage and computing facilities to reduce the time-and-location mismatch that creates renewable curtailment. Second, geography changes. Qinghai, Gansu, Inner Mongolia and other clean-energy-rich regions can move from being power export bases to hosting compute-energy clusters. Third, revenue changes. Green-power companies may earn not only from kilowatt-hours, but also from price arbitrage, green certificates, direct-supply contracts, grid services and integrated data-center energy management.
For ESG investors, that is constructive but not automatically clean. A compute-power campus can improve renewable absorption if computing tasks are flexible and power delivery is real. It can also become a large new electricity load that absorbs scarce clean power and pushes other users toward dirtier marginal supply. The difference depends on hourly matching, grid constraints, storage configuration and transparency. A ‘green computing’ label should not be accepted without evidence of physical electricity use and carbon accounting.
The corporate examples in the article illustrate the direction. Industrial companies described distributed solar, external green-power procurement, green certificates, waste-to-energy, data-center cooperation and cooling integration. These examples are useful because they show how energy, environmental services and digital infrastructure are converging. But they also raise verification questions. Investors should ask whether green-power procurement is incremental, whether certificates correspond to actual consumption, and whether waste-heat or cooling claims are measured.
The strongest companies will be those that can bundle generation, storage, trading, carbon accounting and operational flexibility into bankable contracts. The weakest will simply attach AI language to ordinary power assets. Compute-power coordination is therefore a quality filter. It rewards operators that understand both electricity markets and digital-load needs, and it exposes companies that only own capacity without system capability.
The broader China ESG signal is that the green-power sector is entering a more sophisticated phase. Scale remains important, but value is shifting toward delivery certainty and verifiable attributes. That will affect financing. Banks and bond investors should not only ask how many megawatts a project has; they should ask who consumes the power, how the low-carbon attribute is documented, whether the load can respond to system needs, and what happens when renewable output is low.
This is also a regional-development test. If western clean-energy regions can host flexible computing loads, they may capture more value locally rather than exporting raw electricity. If projects are built without water, grid and market discipline, they may repeat old industrial-park problems under a digital label. The opportunity is real, but so is the risk of overbuilding. Compute-power coordination deserves attention because it is where China’s AI ambitions meet the physical limits of the energy system.
From Issue 005 · 11–17 May 2026.
Questions or corrections? Contact the editor.