China's AI Data Centers Face Hurdles in Green Power Adoption, 80% Target by 2030 Challenged by Load Issues
Alina Collins
China wants renewables to cover 80% of data-center power by 2030, up from just 11% in 2023; but inflexible server loads and resistant grid operators are creating structural obstacles to that seven-fold leap.
How wide is the gap to 80%?
Renewables supplied just 11% of China's data-center power in 2023. The 2030 target is 80% — a more-than-sevenfold jump in seven years.
The 2026 Government Work Report explicitly called for integrating computing infrastructure with the power grid. This means → Beijing now treats "AI eats electricity" as a national-scale infrastructure priority.
Pei Shanpeng, a director at State Power Investment Corp (国家电投), estimates data centers will add 300–500 billion kWh of demand between 2026 and 2030 — 18% of all new national power demand.
In plain terms = the low end of that range alone is roughly equal to the UK's entire annual electricity consumption.
Why don't green-power suppliers want data centers?
Compared with aluminium smelters and other traditional heavy users, data-center loads are hard to flex up or down.
Pei explained: GPUs are so expensive that operators run them "as soon and as hard as possible" once purchased — they won't throttle back to match green-power swings.
This means → renewables are inherently intermittent (less wind, less sun = less power), and they need demand-side users who can shift load; data centers are precisely the users who cannot.
Why are grid operators pushing back too?
Routing green power directly into data centers — bypassing grid resale — cuts into grid companies' electricity revenue.
Grid operators also worry that if AI demand slows, the transmission and distribution infrastructure they built upfront becomes harder to pay off.
In plain terms = the grid's fear is: "We build the road during rush hour, you stop driving at night, and we're stuck with the toll-road debt."
Where is the real tension?
The core goal of pushing data centers toward green power is cutting carbon emissions, not cutting electricity costs. This reflects a policy motivation rooted in climate commitments, not cost optimization.
Yet rapid data-center expansion is already straining regional grids — pushing up both average and peak loads, forcing operators to balance meeting demand growth against maintaining grid reliability.
This means → even with clear policy intent, neither the economics nor the engineering on the grid and renewables side fully add up yet — the path to 80% remains unclear.
Content is for reference only, not financial advice.