Introduction
As delegates gather in Antalya, Turkey for the latest United Nations climate negotiations, artificial intelligence has become a frequent talking point. The host country has urged governments to ensure AI is built, powered, and rolled out in ways that support climate action. Despite bold promises from AI executives, there is scant evidence that the technology is currently lowering the greenhouse gas emissions responsible for global warming.
What Happened
Industry figures have set high expectations. OpenAI's Sam Altman said in 2024 that AI would produce remarkable breakthroughs, including solving the climate crisis. A 2023 opinion article co-authored by Google's sustainability chief projected that AI could reduce worldwide emissions by 5% to 10% by 2030. A peer-reviewed study from the London School of Economics' Grantham Research Institute and Systemiq estimated that AI might prevent 3.2 to 5.4 billion tons of carbon dioxide each year by 2035, with most of the benefit concentrated in electricity generation, meat and dairy, and transportation.
A review of public disclosures from leading firms in those sectors found that few are pursuing the specific AI applications underlying those estimates, and fewer still report meaningful climate results.
In electricity, AI-based forecasting of supply and demand could increase renewable output from already-built solar and wind facilities by up to 20%, avoiding roughly 1.8 billion tons of emissions annually. Yet among grid operators in China, Europe, and the United States, only one collaboration stands out: the United Kingdom's National Grid ESO working with Open Climate Fix. That project reduced solar forecast errors by half and prevented hundreds of thousands of tons of carbon dioxide per year. Other operators in the EU, the US, and China have little public evidence of similar AI integration.
In meat and dairy, achieving the projected cuts would require AI to help alternative proteins match conventional products on taste and cost. NotCo and Kraft Heinz have partnered, using NotCo's Giuseppe platform to analyze plant molecules and recreate the flavor, texture, and aroma of animal-based foods. Still, most alternative-protein companies do not show similarly deep AI use, and the sector struggles with slow consumer adoption and shrinking investment. Venture capital for cultivated meat fell from a $989 million high in 2021 to about $83 million in the latest year.
Transportation offers a few encouraging cases. CATL, which holds roughly 40% of the global electric-vehicle battery market, created an AI system trained on more than 100,000 battery designs and 600 terabytes of test data, cutting candidate material screening time by two orders of magnitude. Ride-hailing service Didi Chuxing reported a 16% average drop in per-trip emissions through AI-optimized routing that reduces detours.
Why This Matters
The climate case for AI lacks solid proof. Oil and gas companies are also interested in these tools, partly to expand fossil fuel production. Tech firms that have made sustainability pledges should clarify whether their products will advance clean energy or enable more extraction of dirty energy.
AI's own energy demand is another concern. The rapid expansion of AI models and data centers has driven up electricity use, helping push Microsoft, Google, and other major tech companies off their climate targets. The International Energy Agency forecasts that data center emissions could surpass 300 million tons annually by 2035.
Finally, deploying AI is only the first step. The technology does not accelerate permitting, certification, construction, packaging, marketing, or sales, so even promising tools face slow real-world rollout.
Key Takeaways
- Projections of major AI-driven emissions cuts by 2035 are not yet matched by widespread industry adoption.
- The UK's National Grid ESO and Open Climate Fix provide the clearest example of meaningful AI use in power-grid renewable forecasting.
- NotCo and Kraft Heinz show early progress in alternative proteins, but cultivated meat funding has collapsed and mainstream consumer acceptance remains limited.
- CATL and Didi Chuxing demonstrate that AI can deliver measurable emissions reductions in batteries and ride-hailing.
- Rising data center electricity consumption and possible AI use by fossil fuel producers could erase any climate gains without deliberate policy and corporate action.
Conclusion
The current record does not support the assertion that AI will solve the climate crisis. Stronger, better-targeted adoption in power, food, and transport might eventually bring the 2035 estimates within reach, but only if companies and governments act with purpose. Corporate customers should favor low-impact models and developers with strong climate track records, while policymakers should create incentives or mandates for responsible AI use. Without those measures, AI's environmental costs could outweigh its benefits.

Discussion
Join the conversation
Thoughtful reactions, questions, and follow-up ideas help shape the next story.