AI’s climate promise faces an implementation gap

AI could help reduce global emissions, but publicly disclosed deployments offer limited evidence that major industries are realizing its potential. A peer-reviewed study led by researchers at the London School of Economics’ Grantham Research Institute and Systemiq projected annual reductions equivalent to 3.2 to 5.4 billion tons of carbon dioxide by 2035, after accounting for AI’s energy footprint. Most potential savings were concentrated in power, meat and dairy, and vehicles.

Progress remains uneven. A partnership between the UK’s National Grid ESO and Open Climate Fix halved solar-forecast error, avoiding hundreds of thousands of tons of carbon dioxide annually. NotCo’s venture with Kraft Heinz uses AI to develop plant-based alternatives, but broader adoption faces weak consumer demand and funding challenges. Battery manufacturer CATL has reported faster materials screening, while independent researchers found that Didi Chuxing’s AI-optimized routing reduced per-trip emissions 16% on average.

These gains face competing pressures. Oil and gas companies can use AI to increase fossil-fuel production, while expanding data centers are raising electricity consumption and emissions. The IEA’s base forecast puts global data-center emissions above 300 million tons annually in 2035, including AI and non-AI computing. Environmental benefits remain dependent on targeted deployment, cleaner computing and sustained implementation, with permitting, construction and commercialization still limiting how quickly innovations reach the market.