AI data centers move to the center of Climate Week as leaders weigh energy costs and climate uses

Artificial intelligence has become one of the most contested topics at Climate Week because the same technology that promises to improve forecasting, grid management and scientific discovery is also increasing demand for electricity, water and computing infrastructure. AP reported that discussions around AI and data centers moved into the center of the climate agenda, with speakers warning that the industry’s rapid expansion could make emissions targets harder to meet if power systems and construction plans fail to keep up.

The tension is easy to understand. AI models require vast amounts of computation to train and serve. That computation lives in data centers that need electricity for chips, cooling and backup systems. When those facilities are built faster than clean power can be added, they can increase fossil fuel use or keep older power plants online for longer. Local communities may also face pressure on water supplies, land use and electricity prices. For climate advocates, the fear is that AI’s growth could quietly consume a share of the emissions reductions other sectors are working to achieve.

At the same time, many climate researchers and companies argue that AI can help solve difficult problems. Better models can support weather forecasting, wildfire risk analysis, building efficiency, battery research and grid balancing. Utilities can use machine learning to anticipate demand and integrate renewable energy more smoothly. Scientists can use AI to search large datasets for new materials or to understand changing ecosystems. The question at Climate Week was not whether AI belongs in climate work, but whether its infrastructure can be built responsibly enough for the benefits to outweigh the costs.

AP noted that climate leaders framed the issue as a fast-moving governance challenge. The pace of data center expansion is difficult for regulators, grid operators and local planners to absorb. Developers often announce enormous projects before communities know how power will be supplied or whether new transmission can be built on time. If clean energy procurement is only an accounting exercise, critics say, the climate benefit may be weaker than advertised. If it directly supports new renewable generation, storage and grid upgrades, the industry’s footprint could look very different.

The discussion also reflects a broader shift in the AI debate. For the past two years, many public arguments focused on model safety, copyright, jobs and misinformation. Energy use was discussed, but often as a side issue. The data center boom has changed that. AI now has a physical geography: substations, cooling systems, land deals, power contracts and local political fights. That makes its climate impact easier to see and harder to dismiss.

For policymakers, the likely path is not a simple choice between embracing or rejecting AI. The more practical challenge is to require transparency about electricity and water use, connect data center approvals to clean energy and grid planning, and push companies to disclose whether their climate claims match real-world power additions. Climate Week made clear that AI will be judged not only by what its models can predict, but by how its infrastructure is built.