Emerald AI Teams with NVIDIA and Partners to Develop Power-Flexible AI Factory and Reference Design to Unlock 100 GW of Grid Capacity and Supercharge the AI Revolution
Emerald AI Teams with NVIDIA and Partners to Develop Power-Flexible AI Factory and Reference Design to Unlock 100 GW of Grid Capacity and Supercharge the AI Revolution
WASHINGTON, October 29, 2025 /PRNewswire/ — A coalition of leaders from the artificial intelligence and energy industries, including NVIDIA, Emerald AI, EPRI, Digital Realty and PJM, today announced the world’s first flexible power AI factory. This facility, the 96 MW Aurora AI Factory in Virginia, will serve to implement a new benchmark design and certification standard for flexible power AI infrastructure, enabling utilities to unlock existing network capacity and quickly connect new AI data centers.
The new flexible power reference design, if adopted nationally, could pave the way for a capacity estimated at 100 GW on the existing electricity system to fuel AI’s growth, equivalent to 20% of total U.S. electricity consumption in one year. Flexible AI factories could connect to power grids more quickly, accelerating the construction of AI infrastructure and AI innovation while improving the reliability and affordability of the power grid.
“Virginia has long been the leader in data center investment and is committed to promoting economic development and AI competitiveness while protecting the affordability and reliability of the power grid,” said Virginia Governor Glenn Youngkin. “This pioneering flexible AI factory in Manassas represents a critical innovation to boost America’s competitiveness in AI while strengthening energy security and affordability in our communities.
The Aurora AI Factory, under construction by Digital Realty in Manassas, Virginia, and scheduled to open in the first half of 2026, will be the first facility built to this new standard and will serve as a live innovation center. It aims to demonstrate how AI computing can align with grid needs to improve the utilization of existing electricity networks, reduce grid stress during peak demand, and ultimately contribute to a more stable and affordable energy future.
As the demand for generative AI drives unprecedented growth in computing, Emerald AI is working with NVIDIA’s development of an AI factory infrastructure, which not only advances AI, but also actively participates in network stabilization. At the heart of this architecture is a tight integration between the technologies of Emerald AIan NVIDIA NVentures portfolio company and Member of NVIDIA Inceptionand the NVIDIA AI software stack. Emerald AI’s GridLink and Conductor products leverage NVIDIA AI Enterprise components, including NVIDIA NIM Microservicesin coordination with the NVIDIA Mission Controlproviding a crucial link between real-time network needs and data center operational controls, enabling intelligent workload planning and power management to meet both network needs and stringent AI workload performance standards.
To validate the effectiveness of the architecture, the Aurora installation will be subject to demonstration tests in collaboration with EPRI DCFlex Initiative. DCFlex explores how data centers can provide network services, benefiting utilities, operators and consumers. The effort will test accurate, real-time responses to simulated grid stress events. Scenarios will include simulating peaks in demand during summer heatwaves or sudden drops in renewable energy production. The overall goal of the demonstration is to see if the data center can intelligently reduce power consumption to support the network while using sophisticated workload orchestration to ensure acceptable quality of service and performance for critical AI training and inference workloads.
“We are excited to work with NVIDIA and this distinguished group of industry leaders to demonstrate what is possible when data centers become intelligent, flexible network partners,” said Varun Sivaram, founder and CEO of Emerald AI. “Through our tight integration with NVIDIA’s AI factory stack, we transform network insight into actionable data center control. Aurora is the first step in proving that AI can scale sustainably, accelerating AI innovation, leveraging massive reserve network capacity, and preserving network reliability and affordability.
“AI factories are essential to AI leadership and essential to economic development. They must be built in harmony with the power grid and the communities they serve,” said Marc Spieler, senior general manager of energy at NVIDIA. “By making Project Aurora the world’s first flexible power AI factory, Emerald AI establishes the reference architecture that enables NVIDIA’s AI platform to operate as a massive, flexible power asset. This model will support communities, unlock underutilized grid capacity and help make energy more affordable for everyone.
“As part of DCFlex, we are excited to have the opportunity to test the viability of the Emerald solution as collaborators with the Aurora facility,” said David Porter, EPRI vice president of electrification and sustainable energy strategy. “Computing flexibility allows data centers to dynamically adjust workloads in response to network conditions, and could play a role in ensuring reliable and affordable power for everyone. »
“As global demand for AI computing grows, infrastructure innovation must scale with it,” said Chris Sharp, chief technology officer at Digital Realty. “Digital Realty is committed to advancing solutions that integrate compute performance, sustainability and network resiliency. The Aurora initiative and Emerald AI’s innovative solutions provide a valuable opportunity to explore how flexible energy management can support both our customers’ growth and the broader energy ecosystem.
The power-flexible AI Factory aims to redefine how data centers are built and operated, ensuring that the exponential growth of artificial intelligence coincides with the sustainable growth of the world’s power grids.
About Emerald AI
Emerald AI is the pioneer of AI-driven IT orchestration, transforming energy-intensive data centers into smart grid assets. Its AI-powered Conductor platform enables data centers to dynamically adjust power consumption by orchestrating AI workloads, such as training, fine-tuning and inference, across data center networks in real time. For more information, visit www.emeraldai.co.
