NVIDIA and Emerald AI are partnering with U.S. power companies AES, Constellation Energy, Invenergy, NextEra Energy, Nscale Energy & Power, and Vistra to jointly promote the development of next-generation AI factories. These facilities will not only connect to the grid more quickly but will also function as flexible energy assets, providing support to the power grid. According to the partners, the new AI factories will adopt NVIDIA’s Vera Rubin DSX AI factory reference design, which includes the DSX Flex software library to enable efficient integration between AI factories and grid services. To facilitate rapid deployment, the AI factories can initially utilize nearby power generation and energy storage facilities as transitional power sources, forming hybrid AI factories. Later, these resources can be flexibly dispatched to supply power to the grid, accelerating the grid connection process for AI factories while supporting the broader power system. The collaborators stated that this model helps speed up the deployment of AI computing capacity, creating greater value for customers and communities. NVIDIA founder and CEO Jensen Huang remarked, "NVIDIA and Emerald AI are working together to build the future of AI, enabling performance, efficiency, and grid responsiveness to work in instant synergy." The companies also noted that Emerald AI’s Conductor platform will coordinate computing flexibility, integrating on-site power generation, energy storage, and other customer-side resources to achieve precise power dispatch that responds to grid demands. AES, Constellation, Invenergy, NextEra Energy, Nscale Energy & Power, and Vistra have committed to actively developing power generation capacity to ensure supply meets growing electricity needs. The parties will jointly evaluate optimized power generation solutions for AI factories. These facilities will be constructed based on an architecture jointly developed by NVIDIA and Emerald AI and will explore hybrid project models using localized power supply to shorten energization timelines and deliver greater value to the grid.