The energy industry, responsible for ensuring the electrical supply to populations and industries around the world, is facing one of its biggest challenges: adapting to a constantly changing landscape marked by the growth of distributed energy resources and a significant increase in demand. To address these challenges, EPRI (Electric Power Research Institute), along with NVIDIA and other collaborators, has announced the creation of the Open Power AI Consortium, a consortium aimed at promoting the adoption of artificial intelligence (AI) in the energy sector.
The official presentation took place during the NVIDIA GTC 2025 conference, held in San Jose, California. This consortium brings together energy companies, technology firms, and research centers to develop open AI models trained with industry-specific data from the electric sector, with the goal of optimizing grid management, improving efficiency, and strengthening the resilience of the electric system.
Next-Generation AI Models for the Energy Sector
One of the consortium’s main projects is the creation of large multimodal language models, trained with vast technical and regulatory databases provided by EPRI. These models, developed with the power of hundreds of NVIDIA H100 GPUs, will enable electric companies to automate complex processes, expedite grid connection studies, and prepare licenses and permits in markedly reduced timeframes.
According to Arshad Mansoor, President and CEO of EPRI:
“AI has the potential to revolutionize the electric sector in the next decade, enhancing the reliability of grids and optimizing asset performance. The Open Power AI Consortium positions us at the forefront of achieving a more resilient and affordable energy future.”
The first open model in the sector, resulting from this collaboration, will soon be available for early access through a NVIDIA NIM microservice.
A Global Backing with Major Industry Names
The executive board of the consortium already includes participation from over 20 major energy companies, including Duke Energy, Exelon, Pacific Gas & Electric, and Portland General Electric, as well as tech giants like AWS, Oracle, and Microsoft. Additionally, Articul8, a member of the NVIDIA Inception program for disruptive startups, is actively participating in the development of these advanced models.
Vincent Sorgi, CEO of PPL Corporation and Chair of EPRI’s Board, added:
“Collaborating with EPRI and NVIDIA will enable us to tackle the major challenges facing the industry and keep us at the forefront of innovation and operational excellence.”
AI for a More Efficient Energy Future
With an estimated 4% annual growth in global energy consumption until 2027, electric companies are seeking new ways to strengthen infrastructure, balance diverse energy sources, and expand grid capacity. AI agents trained with industry-specific documentation will be able to accelerate processes that currently can take up to four years, such as interconnection studies for new generation facilities. The consortium aims to reduce these timeframes by at least 80%.
The AI model will also facilitate the preparation of environmental documentation, regulatory studies, and rate-setting processes, reducing administrative burdens and increasing agility in decision-making.
Finally, the consortium will develop a standard framework of benchmarks and reliability metrics so that companies, researchers, and regulatory bodies can assess the performance of these new technologies.
For more information about the Open Power AI Consortium and its developments, EPRI has scheduled several highlighted sessions as part of the NVIDIA GTC 2025, available on the event’s official website.