AI Services and Energy: Driving Datacenter Sustainability
As AI demands grow, RISE explores how datacenters can become more sustainable through energy efficiency, heat recovery, and innovative cooling technologies at the ICE Datacenter in Lulea.
AI and advanced digital services raise fundamental questions: are they environmental villains or climate heroes? While they can significantly help increase sustainability, the datacenters that power them require substantial amounts of energy and water. RISE is at the forefront of making AI infrastructure more sustainable.
The Scale of the Challenge
The numbers are significant:
- EU datacenters consume approximately 3% of total electricity
- This equals the energy needs of roughly 4 million households
- Up to 99.97% of datacenter energy converts to waste heat
According to Tor Bjorn Minde, Head of Unit at ICE Datacenter: “A lot is happening in the industry now, with investments in everything from energy efficiency to heat recovery.”
Eight Sustainability Solutions
RISE identifies key approaches to sustainable AI infrastructure:
- Software and AI model optimization - Making algorithms more efficient
- Enhanced server virtualization - Better resource utilization
- Energy-efficient hardware - New component designs
- Predictive analytics - Smart resource management
- Renewable electricity - Clean power sourcing
- Improved cooling systems - Advanced thermal management
- Heat recovery applications - Turning waste into value
- Component recycling - Circular economy approaches
Innovative Heat Recovery
Heat recovery projects in Sweden demonstrate creative applications of datacenter waste heat:
- Greenhouse agriculture using recovered heat
- Mealworm breeding for sustainable animal feed production
- District heating integration
The ICE Datacenter
The ICE (Infrastructure and Cloud research and test Environment) facility in Lulea is approximately 900 square meters, built in 2016 to leverage northern Sweden’s renewable electricity availability. It’s one of the largest open facilities in the world where companies can experiment with sustainable datacenter technologies.
Looking Forward
RISE research focuses on liquid cooling for generative AI workloads, recognizing that as AI models grow, traditional air cooling becomes insufficient. The combination of sustainable power, innovative cooling, and heat recovery positions Sweden to lead in green AI infrastructure.
