AI Factories Become Grid Assets in New Energy Partnership

As artificial intelligence demands unprecedented electricity consumption, a new approach is emerging that transforms AI data centers from passive power drains into active grid assets. NVIDIA and Emerald AI have announced a collaboration with six major energy companies—AES, Constellation, Invenergy, NextEra Energy, Nscale Energy Power, and Vistra—to develop AI factories that connect to power grids faster while providing flexibility back to the electrical system. This partnership represents a cross-industry effort to support AI innovation in the United States while building a more reliable power infrastructure for the growing demands of computational workloads.

The core innovation lies in the NVIDIA Vera Rubin DSX AI Factory reference design, which includes the DSX Flex software library for connecting AI factories to power-grid services. These next-generation facilities can use co-located energy generation and storage as bridge power during initial deployment, then later harness these same resources to flexibly supply electricity back to the grid. This approach accelerates AI capacity deployment while creating broader value for energy customers and local communities, addressing what Constellation CEO Joe Dominguez describes as "a peak problem, not a supply problem" in current electrical systems.

Emerald AI's Conductor platform orchestrates computational flexibility alongside onsite generation, batteries, and other behind-the-meter resources to deliver precise, grid-responsive power management. This coordination helps operators meet power targets, protect priority AI workloads, shorten time on bridge power, and support larger and faster grid interconnections. According to Emerald AI CEO Varun Sivaram, AI factories produce "tremendously valuable AI tokens and knowledge" and with DSX Flex can provide "measurable relief back to the grid" rather than functioning as permanent energy islands.

The energy companies involved will evaluate optimized generation applications designed specifically for AI factories built with this architecture, including hybrid projects that use co-located power to speed time to electricity access. AES CEO Andrés Gluski notes that "grid flexibility will be key to addressing AI's unprecedented demand while supporting system reliability," while NextEra Energy's John Ketchum emphasizes the need for technologies that allow new demand and related generation to integrate into grids quickly and at the lowest possible cost.

Power-flexible AI factories could help unlock up to 100 gigawatts of capacity across the U.S. power system by combining optimized infrastructure design with efficient use of existing assets. These facilities convert electricity into what the announcement describes as "among the highest-value outputs modern infrastructure can produce"—AI tokens, models, and intelligence. The approach addresses that many gigawatt-scale AI projects face with conventional interconnection timelines being too slow for the pace of AI investment.

Over the past year, Emerald AI and NVIDIA have trialed AI power flexibility demonstrations at five commercial data centers worldwide. DSX Flex is expected to deploy at commercial scale later this year at the NVIDIA AI Factory Research Center in Virginia, planned as one of the world's first power-flexible AI factories with NVIDIA Vera Rubin infrastructure. The companies intend to identify and advance project opportunities using this reference design to accelerate large-scale AI infrastructure deployment and support faster grid interconnections.

The announcement includes forward-looking statements noting that many products and features remain in various stages and will be offered on a when-and-if-available basis. These statements are subject to risks including global economic conditions, technological development and competition, market acceptance of products, and changes in industry standards and regulations. The development, release, and timing of features described remain at NVIDIA's sole discretion.

This collaborative approach represents a significant shift in how energy-intensive computing infrastructure interacts with electrical grids. By treating AI factories as flexible assets rather than permanent loads, the partnership aims to strengthen U.S. energy leadership while enabling broader AI deployment over time, creating what Vistra CEO Jim Burke describes as "better utilization of current grid infrastructure" while continuing to build additional capacity for long-term needs.