Decentralized AI needs more evidence on energy use

Cassidy Buhler and colleagues urge researchers to shift away from using mega data centres and run open-weight artificial-intelligence models on local institutional servers (see C. K. Buhler et al. Nature 656, 296–298; 2026). The case for having greater autonomy over research tools is strong, but the authors’ predictions related to energy savings currently rest on weak evidence.

Nature 658, 834 (2026)

doi: https://doi.org/10.1038/d41586-026-03245-2

Related Articles

• Data centres will use twice as much energy by 2030 — driven by AI

• Why open-source generative AI models are an ethical way forward for science

• See more letters to the editor

Subjects

• Information technology
• Sustainability
• Environmental sciences

Latest on:

• Information technology
• Sustainability
• Environmental sciences

• 

  Private, offline and yours: how to search the literature with local AI systems

  Technology Feature 12 OCT 26

• 

  Retrofitting language models to operate over bytes

  Article 07 OCT 26

• 

  How digital breakdowns affect supply chains — and endanger the global economy

  Comment 06 OCT 26

• When microbes degrade plastic, where does it go?

  Correspondence 13 OCT 26

• 

  Chemists struggle to ditch hazardous solvents — even after decades of ‘green’ efforts

  News 25 SEP 26

• 

  Cities and countryside are intimately linked — green initiatives must recognize that

  Comment 15 SEP 26

• When microbes degrade plastic, where does it go?

  Correspondence 13 OCT 26

• 

  Woody ‘blankets’ are being deployed on Arctic permafrost to reduce carbon emissions

  News 08 OCT 26

• 

  Australia’s current wildfire crisis linked to colonial land-use change

  Article 07 OCT 26