NVIDIA and Emerald AI Collaborate with Major Energy Companies to Pioneer Flexible AI Factories Interconnected with the Power Grid

CERAWeek 2026 — NVIDIA and Emerald AI announced a collaboration with AES, Constellation, Invenergy, NextEra Energy, Nscale Energy & Power, and Vistra to advance a new class of AI factories. These factories will function as flexible energy assets capable of accelerating grid connections, generating valuable AI tokens and intelligence, and supporting the power grid.

By bringing together leaders in technology, energy, and infrastructure, this collaborative effort demonstrates how cross-industry partnerships can fuel AI innovation in the U.S. while simultaneously building a more reliable power system for the American people.

These next-generation AI factories will utilize the new NVIDIA Vera Rubin DSX AI factory reference design. This reference design includes the DSX Flex software library for connecting AI factories to grid services.

To accelerate deployment, these AI factories can use co-located power generation and energy storage facilities as bridge power for hybrid AI factories. Later, they can leverage these resources to flexibly supply power to the grid, expediting the AI factory's grid interconnection and supporting the broader power system. This approach helps bring AI online faster while creating extensive value for customers and local communities.

The DSX reference architecture can also support flexible AI factories without the need for co-located energy resources, enabling larger and faster grid connections.

Emerald AI's Conductor platform coordinates computing flexibility in conjunction with onsite generation, battery storage, and other behind-the-meter resources to provide precise, grid-responsive power flexibility while ensuring service quality for AI computing tenants. This coordination helps operators meet power goals, protect priority workloads, reduce reliance on bridge power, and support larger, faster grid interconnections. It also lessens the need to design infrastructure for peak demand, thereby reducing the burden of future system costs.

"AI factories are the engines of the intelligence era, and like any great engine, all systems—energy, computing, networking, and cooling—must be co-designed as a single architecture," said Jensen Huang, founder and CEO of NVIDIA. "NVIDIA and Emerald AI are working together to realize an AI future where performance, efficiency, and grid responsiveness are available right out of the box."

Varun Sivaram, founder and CEO of Emerald AI, stated: "AI factories are too valuable to be treated as passive loads or isolated facilities. Beyond generating highly valuable AI tokens and knowledge, AI factories can significantly contribute to easing grid load through DSX Flex. Emerald Conductor coordinates computing flexibility with onsite energy resources to support the grid. As a result, it can expedite connections, maintain service quality for AI tenants, and ultimately fortify the surrounding power system."

Building AI Factories that Strengthen the Grid Today's power systems are built to handle peak demand but are underutilized for most of the day. Highly demand-responsive AI factories, combining optimized infrastructure design, efficient use of existing assets, and newly built generation facilities as needed, can help unlock up to 100 gigawatts of capacity across the U.S. power system. Furthermore, by acting flexibly during limited periods of high grid load, they reduce the need to expand the broader grid to maintain reliability.

AI factories convert electricity into AI tokens, models, and intelligence—some of the most valuable outputs that modern infrastructure can produce. Addressing this opportunity requires innovation not just in computing, but also in how corporate energy infrastructure is planned, built, and operated.

Many gigawatt-scale AI projects are moving toward co-locating power generation and energy storage because traditional grid interconnection timelines may not keep pace with AI investments. However, completely isolating generation and storage from the grid has drawbacks. Isolation can lead to underutilized assets, increase the long-term cost per AI token, and prevent energy resources from supporting grid reliability.

AES, Constellation, Invenergy, NextEra Energy, Nscale Energy & Power, and Vistra are working to build the generation capacity required to secure the power supply needed to meet surging demand.

Together, these companies will evaluate optimized power generation applications for supplying electricity to AI factories built on the architecture developed by NVIDIA and Emerald AI. This includes hybrid projects that use co-located facility power to accelerate time-to-power while bringing value to the broader grid. This approach helps improve grid reliability by combining large-capacity AI loads with flexible operations, new energy generation capabilities, and intelligent control.

Additionally, co-located energy resources, if available, can be used to support flexible, grid-connected AI factories from the very beginning.

Andrés Gluski, CEO of AES, stated: "Grid flexibility will be key to supporting system reliability while meeting the unprecedented demand for AI. AES aims to enable next-generation AI infrastructure to shorten the time to power delivery for our customers. DSX Flex is built with flexibility from the start, allowing AI infrastructure to operate as a grid asset that supports faster and more efficient growth."

Joe Dominguez, President and CEO of Constellation, stated: "As the nation's largest producer of clean energy, we recognize that data centers hold immense potential to drive investment in energy infrastructure, create jobs, and benefit local communities. Through demand response, we can also address the need for additional capacity. The challenge isn't a supply issue, but a peak demand issue. By effectively utilizing existing assets, such as highly demand-responsive AI factories that also enable AI-driven demand response, we can more efficiently manage new load growth."

Michael Polsky, founder and CEO of Invenergy, noted: "AI is changing the way we think about energy. Our customers need rapid power delivery as well as the ability to gradually scale capacity. Combining short-term generation solutions with a pathway to full grid connectivity and flexible operation is an innovative and efficient way to meet customer energy demands faster while maintaining system reliability."

John Ketchum, Chairman, President, and CEO of NextEra Energy, said: "Meeting unprecedented new power demand while maintaining a reliable and resilient grid requires more generation resource augmentation than ever before. We also need technology that can integrate new demand and associated generation into the grid as quickly and affordably as possible. NextEra looks forward to collaborating with NVIDIA and Emerald AI to help design efficient energy campuses and flexible AI factories that economically support growing demand while further strengthening U.S. energy infrastructure."

Daniel Shapiro, Chief Power and Energy Officer at Nscale Energy & Power, stated: "We are committed to stabilizing the grid and ensuring that homes and businesses in West Virginia always have access to the power they need. Interconnecting with the grid allows us to receive power on days when demand is at its highest. That is the significance of onsite generation scalable from 2 gigawatts to 8 gigawatts. Nscale's Monarch Campus is not a burden on the grid, but a power asset for West Virginia."

Jim Burke, President and CEO of Vistra, noted: "The U.S. grid is designed to meet peak demand that occurs only for a very small fraction of the year. AI factories that can flexibly adjust power usage according to grid conditions offer a faster solution, and when combined with co-located generation facilities, they can make much better use of current grid infrastructure. This helps accelerate deployment while long-term infrastructure development is underway."

Throughout the past year, Emerald AI and NVIDIA conducted proof-of-concept experiments on AI demand responsiveness at five commercial data centers globally. DSX Flex is scheduled for commercial-scale deployment later this year at the NVIDIA AI Factory Research Center in Virginia. The center is planned to be one of the world's first highly demand-responsive AI factories equipped with NVIDIA Vera Rubin infrastructure.

By identifying and driving construction project opportunities using the Vera Rubin DSX reference design and DSX Flex, both companies aim to accelerate the deployment of large-scale AI infrastructure, support larger and faster grid interconnections, pioneer technical pathways for building a new generation of facilities, expand the economic benefits that AI and energy investments bring to local communities, strengthen U.S. leadership in the energy sector, and achieve the phased expansion of the AI adoption landscape.

*This announcement is a summary translation of a press release issued on March 23, 2026, U.S. time.