AI and Sustainability in Practice: Challenges and Solutions
Is AI part of the climate problem or the sustainability solution? In this hands-on workshop, participants will explore its carbon footprint, discover practical tools, and co-design new pathways — including a visit to EPFL’s EcoCloud facility.
Artificial Intelligence is becoming increasingly energy-intensive, with datacenters projected to consume up to 10% of global electricity by 2030. Racks alone may soon reach 1MW per unit, raising urgent questions about the sustainability of AI infrastructure. The shift from performance-driven to sustainability-driven datacenters presents multiple challenges — and opportunities.
This workshop explores these challenges and highlights pathways toward sustainable AI. We will focus particularly on the carbon footprint of datacenters, while also examining how AI itself can serve as an enabler of sustainability.
The session will alternate between short expert inputs and dynamic small-group discussions on key questions. Participants will reflect on the dual role of AI — both as part of the sustainability problem and as part of the solution.
The workshop will introduce a new carbon assessment methodology developed at EPFL and include a guided tour of the EcoCloud experimental facility, where cutting-edge technologies are being tested to reduce AI’s environmental footprint.
We will conclude by inviting participants to co-create new approaches and propose innovative paths for building sustainable AI systems.
Target audience:
The workshop is designed for anyone interested in AI sustainability, including datacenter providers, AI practitioners, policymakers, researchers, and students who seek a holistic perspective on sustainability.
Organisers:
Xavier Ouvrard et Julia Paolini
