Nvidia joins energy firms on 'AI factories'
Nvidia and Emerald AI are partnering with major U.S. energy companies to treat data centers as flexible grid assets that balance power, stabilize the grid, and run large AI workloads on a variable schedule. The move frames compute capacity as critical infrastructure for embodied AI and robot fleets, tying robotics scale directly to energy strategy and grid flexibility. (nvidianews.nvidia.com)
NVIDIA and Emerald AI named AES, Constellation, Invenergy, NextEra Energy, Nscale Energy & Power and Vistra as partners in the AI‑factory effort announced at CERAWeek 2026. (nvidianews.nvidia.com) The program centers on NVIDIA’s Vera Rubin DSX AI Factory reference design and its DSX Flex software library, while Emerald AI will use its Conductor platform to orchestrate compute flexibility with energy resources. (nvidianews.nvidia.com) Emerald AI’s first live implementation is the 96 MW “Aurora AI Factory” under construction by Digital Realty in Manassas, Virginia, slated to open in the first half of 2026 as a reference deployment. (prnewswire.com) Emerald’s earlier announcement estimated that the power‑flexible reference design, if widely adopted, could unlock roughly 100 GW of existing grid capacity—about 20% of annual U.S. electricity use. (prnewswire.com) Siemens Smart Infrastructure disclosed a strategic investment in Emerald AI and plans to integrate Fluence battery storage and PhysicsX modeling into an ecosystem intended to accelerate grid interconnections for AI data centers. (press.siemens.com) The reference architecture supports “hybrid” projects that use co‑located generation and storage as bridge power to speed time‑to‑power and later operate flexibly for grid services, with Emerald’s GridLink/Conductor tying into NVIDIA NIM microservices and Mission Control for workload scheduling and power management. (nvidianews.nvidia.com)
