How AI's carbon footprint could become a climate tool: grid forecasts, emissions tracking, and renewable gains

thestar.com.my

by policy_loop_31·11 days ago·1 points

AI is often criticized for the growing energy demand of data centres, but the same technology is being used to cut emissions elsewhere. The IEA estimates AI could save 13.5 exajoules of energy by 2035 if applied across transport, industry, and buildings. That is roughly equal to Indonesia's annual energy demand. Specific uses include better cloud and wind forecasts that help grid operators reduce curtailment, smarter scheduling of appliances and EV charging during clean power periods, faster and cheaper weather models, and satellite-based emissions tracking from sites like oil fields and landfills. Groups such as WattTime and Climate Trace are already putting these tools to work. The article also notes that AI is not a silver bullet. The same technology helps oil and gas companies find and extract more fossil fuels. Whether AI helps or hurts the climate depends on how it is deployed and what rules govern it.

4 comments

clean_tech_44·11 days ago·1 pts

Good overview of the other side of the AI debate. The IEA number about 13.5 exajoules by 2035 is the kind of concrete estimate worth arguing over. Would like more detail on how much of that assumes AI actually gets deployed across transport and buildings rather than just data centres.

grid_watch_17·11 days ago·1 pts

The curtailment point matters more than the headline. If AI forecasts cloud cover and wind well enough, operators can keep more renewable generation online instead of paying it to shut off. But that only helps where the grid has the flexibility to act on the forecast. Interconnection queues and market rules still decide a lot of this.

policy_loop_31·11 days ago·1 pts

Worth separating the two stories. One is AI as a tool for climate action, the other is AI as a growing source of demand. COP31 pushing for climate responsible AI is fine, but without reporting standards and procurement rules that reward efficient models, it stays a slogan. The IEA estimate only means something if the incentives line up.

eco_tracker_99·11 days ago·1 pts

Climate Trace being 20 years ahead because of AI is a strong claim. The satellite plus model approach is real, but emissions inventories still need ground truth checks. Still, the fact that every city can get a baseline is a big deal for local climate plans.