The Open Power AI Consortium is a global collaboration focused on developing open AI models for the energy sector.
The new Open Power AI Consortium brings together energy companies, technology partners, and research institutions to accelerate the development and application of domain-specific generative AI models. Chip giant Nvidia is also part of the consortium. Through this collaboration, they aim to simplify the deployment of AI in the energy sector.
Open Models
The Open Power AI Consortium focuses on three main objectives. First, it aims to develop open-source AI models and datasets specifically tailored to the challenges within the energy sector. Second, a testing environment, a so-called ‘sandbox’, will be created where companies, research institutions, and startups can develop and validate AI applications. Finally, the consortium wants to implement AI models using global knowledge and resources to accelerate innovation and reduce implementation risks.
read also
‘AI can greatly advance sustainability and completely undermine it’
For this purpose, EPRI collaborated with Articul8 and Nvidia on the first set of domain-specific GenAI models for electricity and energy systems. These models will be available through Nvidia’s NIM microservices in an early access phase.
Energy Sector
The consortium includes members from various corners of the energy sector. Companies such as Duke Energy, Exelon, Southern Company, Saudi Electricity Company, and KEPCO are participating. Technology companies like AWS, Microsoft, Nvidia, and Oracle are also involved.
According to EPRI president Arshad Mansoor, AI can play a crucial role in improving grid reliability and energy management in the coming decade. Partners within the consortium emphasize the importance of collaboration to develop practical AI applications that address evolving energy issues, such as grid modernization and the transition to low-carbon energy.
The demand for energy is only increasing, partly due to the growing number of data centers. These AI data centers require an enormous amount of computing power, so it sounds somewhat ironic to address this problem with AI.