Green AI in Cloud Native Ecosystems: Strategies for Sustainability and Efficiency
Talkat KubeCon + CloudNativeCon Europe 2025 · on Friday, 2025-04-04· at 15:15· Level 1 | Hall Entrance S10 | Room A·
summary
Cloud-native AI platforms focus on cutting inference costs because 65% of operational carbon often comes from inference; data distillation can cut training work by up to 70% and embodied carbon is ~50% from hardware manufacturing. The team ships an open-source GPU-slicing tool InstaSlice and combines it with right-sizing to share GPUs; platform optimizations claim up to 800× improvement in energy use. Work uses Kepler for energy observability and explores cache sharing, routing, and autoscaling strategies that interoperate.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 1:07Why Green AI matters now
- 7:58Efficient inference in AI
- 11:10Efficient AI lifecycle techniques
- 18:21AI optimization platform challenges
- 26:33CNCF’s sustainable AI platforms
- 29:04Call to action
related talks
32mOptimizing AI/ML workloads for sustainability
45mGreen AI: Making Machine Learning Environmentally Sustainable
04mA Cloud Native Stack From Bare Metal To Tokens for Large-Scale AI Inference
20mOperating Frontier Intelligence at Scale
20mBuilding GenAI That Doesn't Cost the Earth
44mThe Invisible Monster: The Carbon Footprint of AI
01h 02mJoint Keynote Address: Scaling AI Sustainably: An Uncharted Territory
06mAI Inference at the Edge: Beyond the Hype