Report: AI's climate benefits are overblown and amount to greenwashing

AI's climate benefits are overblown, report finds

Energy analyst Ketan Joshi has published a report concluding that claims of AI’s climate benefits are largely unsupported and amount to greenwashing. The report, funded by climate action groups, analyzed 154 claims from eight sources, including the International Energy Agency, Microsoft, and Google.

Joshi found that the vast majority of benefits cited are tied to older, "traditional" AI, not the energy-hungry generative AI systems driving today's data center boom. This conflation, he argues, makes the tech industry's defense of its growing emissions "utterly implausible."

Generative AI's climate role is minimal

The report reveals a stark disconnect. Of the 154 climate benefit claims examined, only four were related to generative AI tools like ChatGPT or Copilot. The other 150 pointed to traditional AI models for tasks like prediction or computer vision.

"At no point did this analysis uncover examples where consumer generative systems... led to a verifiable and substantial level of emissions reductions," Joshi wrote. He notes that most projected AI energy consumption will come specifically from generative AI, not traditional machine learning.

Evidence for any AI benefit is weak

Joshi's analysis found the evidence supporting even traditional AI's climate benefits to be thin. Only 26 percent of the 154 claims cited published academic papers, while 36 percent provided no citations at all.

Corporate publications made up 29 percent of citations, and Joshi found these mostly lacked "any primary assessable evidence or peer-reviewed... work." The report concludes that benefits are exaggerated across the board.

• 26% of claims cited academic papers
• 36% of claims had no citations
• 29% of citations were corporate publications

Google claim sparked the investigation

Joshi told The Register he was motivated to research the issue after hearing Google persistently claim its AI would reduce emissions by 5-10%. He characterized the evidence for this as "mind-bogglingly weak" and "one of the worst greenwashing claims I've seen in a while."

When contacted for comment, Google defended its position. "We stand by our methodology, which is grounded in the best available science," a spokesperson said. Microsoft declined to comment, and the IEA did not respond.

Public skepticism is a powerful tool

Joshi believes climate advocates should leverage growing public distrust. "There is widespread public skepticism for what all of this is actually for," he said, noting bipartisan hostility toward destructive data center developments.

He suggests the current role is to prevent new fossil fuel infrastructure from being locked in for decades. This aligns with real-world action, as several U.S. states consider moratoriums on new data centers.

The fight is moving to the states

Activist and regulatory pressure is mounting. Elon Musk's xAI is facing scrutiny in Tennessee over gas turbines powering a Memphis data center, including a lawsuit and state regulator investigation.

Joshi echoes arguments that AI is a bubble built on hype. "The entire software class is leveraged on hype rather than effectiveness," he said, suggesting the narrative may collapse under its own weight as public opinion turns.