Carbon Connect
U.S. NSF grant to rethink the future of sustainable computing through research on efficient algorithms and datacenter operations.
Pillar 01
Sustainability
Reducing the energy, carbon, and resource costs of artificial intelligence.
Our approach
Work that turns ambition into measurable progress.
We support work across algorithms, computing systems, and infrastructure that makes AI more efficient and less resource-intensive at every layer of the stack.
Portfolio
Projects in Sustainability
5 projects
Parallex
A democratized, incentivized, community-driven scientific computing framework powered by idle compute.
MLCommons Algorithmic Efficiency
Infrastructure for measuring improvements in neural-network training algorithms in collaboration with MLCommons.
JouleLedger
A request-aware measurement and modeling system for understanding energy use in shared large-language-model serving.
TechRefresh
An open framework for deciding when datacenter hardware should be refreshed under financial, energy, and lifecycle-carbon objectives.
Keep exploring
The other pillars
Pillar
Safety
Advancing dependable AI systems that are developed and used responsibly.
Pillar
Science
Using AI to accelerate rigorous discovery in fields with meaningful public impact.
Pillar
Research Infrastructure
Building open systems and shared tools that accelerate discovery in high-impact fields.