US energy firm Invenergy has partnered with Nvidia and Emerald AI to develop and deploy what it calls a new class of “flexible AI factories” across the US.

According to the partners, the collaboration will see them develop a portfolio of AI factory-ready powered sites designed to meet a range of workloads, ranging from Edge and distributed inference deployments up to multi-gigawatt campuses. Each site will be paired with either colocated rapidly deployable operational or new-build generation and storage infrastructure.

“I’ve seen firsthand innovation happen when ambition meets execution,” said Michael Polsky, founder and executive chairman of Invenergy. “We know that speed matters. But in this market, experience and execution capabilities matter just as much. As the largest privately held power plant developer in the US, we’re ready to leverage our decades-long track record of building and operating to solidify the ecosystem required to support this supercharged phase of the AI infrastructure buildout.”

All of the data centers built under the partnership will be combined with Nvidia DSX reference design and Emerald AI’s Conductor platform. Nvidia’s DSX reference design, released in March of this year, is a framework for building co-designed, gigawatt-scale AI infrastructure.

EmeraldAI’s Emerald Conductor program is an orchestration platform that serves as a mediator between the grid and data centers, orchestrating AI workloads in real time. This enables data centers to dynamically adjust their energy consumption, supporting grid stability while ensuring acceptable AI compute performance. The company has so far completed three demonstration projects in Phoenix, Chicago, and the UK. It recently announced its latest pilot project with California municipal utility Silicon Valley Power (SVP) in Santa Clara, California.

“With Nvidia’s DSX AI factory architecture, combined with Invenergy’s scalable power assets and Emerald AI’s real-time orchestration, we can dynamically align AI workloads with available energy, improving intelligence per watt while accelerating time to deployment. This approach allows AI factories to scale more efficiently and operate more flexibly—turning them from static loads into adaptive infrastructure that can better align with and support the grid,” said Vladimir Troy, VP of AI infrastructure at Nvidia.

“We’re proud to work with Invenergy and Nvidia to advance flexible AI factories that combine scalable power infrastructure, next-generation AI factory architecture, and Emerald AI’s Conductor platform. By coordinating compute flexibility with onsite energy resources, this approach can help projects come online faster, preserve performance for AI workloads, and support a more reliable and affordable power system,” added Dr. Varun Sivaram, founder and CEO of Emerald AI.

The companies claim the model will support a faster interconnection timeline while also improving grid flexibility through Emerald AI’s orchestration software and its on-site generation and storage assets.

The deal is Inverenergy’s first foray into the direct development of data centers. It has long been one of the largest power providers to the US data center sector, signing an 800MW power supply deal with Meta in June of last year.

Invenergy is a Chicago-based energy developer. It has a diversified portfolio of energy assets, including 17.6GW of wind, 6GW of solar, 5.9GW of natural gas, and more than 300MW of battery storage. It has a presence across the US, Europe, and APAC.