NVIDIA has showcased five companies using its artificial intelligence technologies to tackle different challenges in the energy sector, from managing power grids and operating nuclear plants to storage built from recycled batteries and the development of fusion reactors. The projects combine AI accelerators, digital twins, and simulation tools to cut the time needed to design, operate, or integrate new facilities.
NVIDIA’s AI for clean energy in 30 seconds
- ThinkLabs AI uses digital twins and agents to speed up connecting new energy sources to the grid.
- Atomic Canyon applies AI to information management and operational support at nuclear plants.
- Redwood Materials reuses electric-vehicle batteries to power AI facilities with local storage.
- TerraPower uses digital twins to accelerate development of its future Natrium reactors.
- Commonwealth Fusion Systems uses Omniverse and OpenUSD to shorten experimentation cycles for its SPARC fusion reactor.
The five cases are part of a selection NVIDIA published during New York Climate Week. The company presents them as examples of how accelerated computing and AI can play a role at different stages of energy projects, from planning a grid to simulating nuclear or fusion facilities.
The applications don’t follow a single technological model. Some use AI to analyze large amounts of information, others rely on digital twins to simulate physical systems, and in some projects accelerators are used to control or adapt power supply to variable loads.
Digital twins speed up power grids
ThinkLabs AI works with digital twins and AI agents on NVIDIA’s CUDA platform to analyze the power grid and study the integration of new energy sources.
One problem its technology addresses is connecting new facilities to the grid. According to NVIDIA, Southern California Edison used ThinkLabs’ software to cut the time needed to evaluate each interconnection request from 30-45 days down to roughly two minutes.
The system uses an agent capable of running grid simulations and identifying possible solutions to interconnection bottlenecks. ThinkLabs also proposes a model in which utilities’ decisions incorporate probability and risk scenarios, rather than working only from a static picture of the grid.
The application is especially relevant for power infrastructure facing growing variability from certain renewable sources, alongside the emergence of new large-scale loads.
AI for nuclear plants and new energy sources
Atomic Canyon applies NVIDIA’s accelerated computing to information management at nuclear facilities through its Neutron and NIVA platforms.
Neutron works as an AI environment for professionals in the nuclear sector. The platform brings together procedures, regulatory documentation, operating experience, licensing records, design calculations, and other structured and unstructured data to build a knowledge layer to work from.
NIVA, short for Nuclear Industry Virtual Assistant, was developed together with the Institute of Nuclear Power Operations, the Electric Power Research Institute, and the Nuclear Energy Institute as an AI-based assistance platform for the US nuclear industry.
Separately, TerraPower works with NVIDIA Omniverse to build digital twins for developing its facilities. The company is developing the Natrium reactor, a design that uses liquid sodium as a coolant instead of water.
According to NVIDIA’s description, TerraPower’s system aims to use digital simulations to speed up certain phases related to siting and delivering future plants. The company itself frames this as an effort to compress processes that can currently span years into much shorter periods, though that’s a development goal rather than an outcome already generalized to new plants.
The Natrium reactor also includes a thermal energy storage system tied to its design, though NVIDIA’s publication focuses AI’s role on digital twins and the planning of future facilities.
Recycled batteries to power data centers
Redwood Materials addresses another problem created by AI’s expansion: data centers’ electricity demand can grow faster than new grid infrastructure can be built.
The company uses batteries from electric vehicles, which NVIDIA says are 100% recycled, to build energy storage systems for AI facilities. The approach provides local energy capacity without depending exclusively on immediate grid expansion.
The system combines the batteries with power electronics developed by Redwood and an AI layer that runs on the NVIDIA Blackwell platform. The goal is to match supply to a data center’s shifting needs, including the swings tied to AI model training loads.
Redwood also says its architecture needs fewer transformers and inverters and can use both reused and new batteries. These claims come from the company itself and are part of the project description NVIDIA published.
Fusion aims to cut years of experimentation
Commonwealth Fusion Systems (CFS) uses NVIDIA Omniverse and OpenUSD-based libraries to work on its SPARC demonstration reactor.
The company aims to use simulations to speed up part of the experimentation needed to develop fusion systems. CFS says these tools let it compress work cycles that could take years into periods of weeks — an estimate that comes from the company.
SPARC uses high-temperature superconductors to generate stronger magnetic fields. CFS says this technology lets it considerably shrink its design compared with earlier fusion concepts.
The company’s next planned step is ARC, a future fusion plant CFS intends to build in Chesterfield County, Virginia, with grid connection during the 2030s. That timeline comes from the company’s own plans and doesn’t mean the plant is currently operating.
The five projects show very different applications of AI within energy. NVIDIA contributes through accelerated computing, simulation, and software technologies, while the selected companies develop the actual energy systems. In several cases, final results still depend on projects that remain under development.
The push echoes an earlier collaboration in which Microsoft and NVIDIA teamed up to apply AI to nuclear power, part of a broader trend of AI’s rising electricity demand pulling nuclear energy back onto Big Tech’s radar.
Frequently asked questions
Which companies does NVIDIA showcase as examples of AI applied to clean energy?
NVIDIA highlights ThinkLabs AI, Atomic Canyon, Redwood Materials, TerraPower, and Commonwealth Fusion Systems as five companies using its technologies in energy projects.
What does ThinkLabs AI use artificial intelligence for?
ThinkLabs AI uses digital twins and AI agents to simulate power grids and help evaluate interconnection requests for new facilities.
How does Redwood Materials use electric-vehicle batteries?
Redwood Materials uses recycled electric-vehicle batteries to build energy storage systems that, among other uses, supply electricity to artificial intelligence facilities.
What fusion project is Commonwealth Fusion Systems developing?
Commonwealth Fusion Systems is developing SPARC, a demonstration tokamak, and uses NVIDIA Omniverse and OpenUSD tools to speed up part of its design’s simulation and experimentation.
via: blogs.nvidia
