As AI technology spreads, its use cases in changing power grids are increasing. Nvidia CEO Jensen Huang presents AI vision for the power grid, emphasizing energy efficiency and productivity improvements.

NVIDIA founder and CEO Jensen Huang announced on the 20th that he attended the annual meeting of the Edison Electric Institute (EEI), an American and international utility association, on Tuesday the 18th.

The power grid and the utilities that manage it have a critical role to play in the next industrial revolution driven by AI and accelerated computing, Jensen Huang told the Edison Electric Association's annual meeting.

“The future of digital intelligence is very bright, and the future of the energy sector is equally bright,” said Jensen Huang. “The greatest impact and return will come from applying AI to power grids to supply energy.”

Jensen Huang gave one example of how the power grid could use AI-powered smart meters to let customers sell excess power to their neighbors.

“By connecting resources and users like Google, the power grid will become a smart network with a digital layer like an energy app store,” he said. “AI, like previous industrial revolutions, will drive productivity to levels we have never experienced before.”

■ AI that will illuminate the power grid

Today’s power grid is primarily a one-way system connecting a few large power plants to many users. But in the future, it is expected to become a two-way, flexible, distributed network, with solar and wind power plants connecting homes and buildings with solar panels, batteries, and electric car chargers.

This requires an autonomous control system that can process and analyze massive amounts of data in real time. It’s a massive task that’s well-suited for AI and accelerated computing. Thanks to the vast ecosystem of companies using NVIDIA’s technology, AI is being applied to use cases across the power grid.

In a recent GTC session, utility provider Hubbell and NVIDIA Inception program member startup Utilidata demonstrated next-generation smart meters using the NVIDIA Jetson platform.

NVIDIA Jetson is the platform that utilities are deploying to process and analyze real-time power grid data using AI models at the edge. Recently, Deloitte announced its support for this initiative.

In a recent GTC session, Siemens Energy presented a use case using AI and NVIDIA Omniverse to build digital twins of substation transformers to improve predictive maintenance and make the grid more resilient, and Siemens Gamesa presented a video showing how Omniverse and accelerated computing are helping optimize turbine placement for large-scale wind farms.

“Deploying the AI and advanced computing technologies developed by NVIDIA will allow us to modernize our electric grid faster and better, so we can better serve our customers,” said Maria Pope, CEO of Portland General Electric.

■ NVIDIA provides 45,000 times more energy efficiency

In his recent COMPUTEX keynote, Jensen Huang said that over the past eight years, NVIDIA has improved the energy efficiency of running AI inference on state-of-the-art large language models (LLMs) by a whopping 45,000x.

NVIDIA Blackwell architecture GPUs deliver approximately 20x more energy efficiency than CPUs for AI and high-performance computing. If all CPU servers were converted to GPUs for these tasks, users would save 37 terawatt-hours (TWh) per year, equivalent to 25 million metric tons of carbon dioxide and the electricity used by 5 million homes.

For this reason, NVIDIA-based systems occupy seven of the top 10 spots in the latest Green500 ranking of the world's most energy-efficient supercomputers. The top six positions are all NVIDIA-based systems.

A recent report urges governments to accelerate the adoption of AI as a critical tool for energy efficiency across a range of industries. It’s time for utilities to explore how they can use AI and accelerated computing to improve operations and cut costs and energy.