China AI for Energy and Climate: An Immersion Guide
Assess AI energy and climate projects using measured savings, compute demand, rebound effects, grid constraints and transparent system boundaries.
An AI energy claim needs a defined system boundary. Separate the electricity used by the model and supporting infrastructure from the operational savings it is intended to produce. Ask whether the comparison controls for output, weather and operating hours. A more efficient prediction does not establish a net climate benefit by itself.
What to evaluate
System boundary
Define the asset, grid, process or emissions outcome being optimised.
Data and control
Inspect sensing, forecasting, optimisation, automation and human authority.
Net impact
Measure efficiency, reliability, emissions, compute energy and rebound effects.
Deployment fit
Test local infrastructure, weather, regulation, markets and maintenance.
Worked evaluation exercise
The following is an illustrative assessment exercise, not a reported customer result.
For a building-control pilot, compare energy use under similar occupancy and weather conditions, while recording comfort complaints and manual overrides. Add the compute and support costs that sit outside the building meter. If lower operating cost encourages longer running hours, include that change in the assessment. Report estimated benefits separately from measured ones. The IEA's Energy and AI report provides wider context; the pilot still requires local measurements.
Questions for the operating team
- What is the measured baseline and system boundary?
- Does AI change a decision or only report it?
- How is compute energy included in net impact?
Warning signs
- Percentage improvements without baseline.
- Avoided emissions with no methodology.
- Pilot results extrapolated across unlike assets.
Planning the visit
Use these questions to scope an industry-focused China AI program. Agree which records and operating workflows can be examined before confirming meetings; host participation and access require confirmation. Our research method explains how we structure the brief and follow-up.
References and scope
These references provide policy or risk-management context. They do not independently verify a host's performance or the illustrative exercise above.