NVIDIA spotlights AI projects for cleaner power

NVIDIA used New York Climate Week to showcase five clean energy efforts applying its AI platforms to bottlenecks that have slowed large-scale adoption, including aging infrastructure, long development timelines and upfront costs. ThinkLabs AI is using NVIDIA CUDA-based digital twins and agents to speed grid interconnection studies, while Southern California Edison used its software to cut application evaluations from 30-45 days to just two minutes.

Atomic Canyon is deploying NVIDIA accelerated compute through Neutron and NIVA, platforms designed to turn nuclear procedures, regulatory guidance, operating records and other plant data into an AI-enabled knowledge layer for reactor operators. TerraPower is using NVIDIA Omniverse-powered digital twin software to support faster siting and delivery of its emission-free Natrium reactor, which is cooled with liquid metal sodium and designed to remain stable without offsite water or electricity in an emergency.

Redwood Materials is building off-grid power systems for AI factories using 100%-recycled electric vehicle batteries, Blackwell-based intelligence and simplified power electronics. Commonwealth Fusion Systems is using NVIDIA Omniverse libraries and OpenUSD to accelerate work on SPARC, with high-temperature superconductors enabling a design that is 40x smaller and aimed at grid-connected ARC power plants in the 2030s.