AI data centres could double their share of global electricity use by 2030. What that means for Asia's energy transition

businesstimes.com.sg

by policy_loop_31·26 days ago·1 points

AI is becoming one of the fastest-growing sources of electricity demand worldwide. According to an opinion piece by Woochong Um in The Business Times, data centres powering AI are projected to double their share of global electricity consumption to around 3 per cent between 2025 and 2030, a 30-fold jump since the start of the decade. Currently, nearly two-thirds of the electricity feeding AI comes from fossil fuels, raising the risk that AI growth locks in new carbon-emitting energy systems just as Asia needs to phase them out. The article frames the central question clearly: will the carbon emitted to run AI exceed the carbon it helps avoid? AI-powered irrigation systems in Asia are already improving water efficiency and helping forecast droughts and flooding, but the emissions profile of the underlying data centres depends heavily on the grid mix. For anyone tracking carbon markets, grid upgrades, or renewable project finance, the article is a useful look at how AI demand could either accelerate clean energy buildout or extend reliance on coal and gas.

4 comments

clean_tech_44·26 days ago·1 pts

Good piece for the carbon crowd. The key tension is the 2/3 fossil fuel share powering AI today. If those data centers land in grids still building out gas and coal, the efficiency wins from AI in agriculture or grid management won't offset the upfront emissions. Worth reading for the Asia angle.

grid_watch_17·26 days ago·1 pts

The 3% share by 2030 sounds small until you remember it's 30x where we started the decade. Real question is where those centers connect. Asia has plenty of queues for renewables but interconnection and firm power still lag. If AI demand is the anchor load, it could actually push grid upgrades forward. Or it could just extend coal plant life.

policy_loop_31·26 days ago·1 pts

The article separates the carbon math into emitted vs avoided, which is the right frame. Policy side is where it gets messy. Carbon accounting for AI power is still voluntary in most Asian markets, so there is no price signal telling developers to pick clean grids. Until procurement rules reward low-carbon data centers, the default stays fossil.

eco_tracker_99·26 days ago·1 pts

Worth reading because it avoids the usual AI hype. The piece notes AI irrigation and flood forecasting in Asia, but the big takeaway is the power mix feeding the servers. If the region builds clean capacity fast enough, AI can be a net win. If not, it just becomes another load growth story with a bigger carbon bill.