Emerald AI Raises $150M at $1.05B Valuation to Turn Data Centers Into Grid Allies

Emerald AI has raised 150 million dollars in an oversubscribed Series A financing round at a 1.05 billion dollar valuation to scale software that turns AI data centers into flexible, grid‑responsive assets rather than constant, inflexible electricity loads.

The round was co‑led by Energize Capital and DCVC, joined by a broad group of strategic and financial investors spanning the AI, energy, and industrial sectors. Participants included Nvidia, Samsung Ventures, Siemens, Aramco Ventures, Salesforce Ventures, GE Vernova, RWE, and JERA Ventures, among others. Emerald AI says 12 Fortune Global 500 companies now count among its investors, a roster that spans chipmakers, cloud and enterprise software providers, and major global utilities and energy firms simultaneously.

The Washington, D.C. based company was founded in 2024 by chief executive Varun Sivaram, who previously held a senior climate policy role in the Biden administration. That background shapes the company's core premise directly: the constraint holding back AI infrastructure growth today is not chips or capital, but power, and specifically the years‑long permitting and construction timelines required to build new grid transmission capacity. Energize Capital managing partner John Tough put it bluntly: "The binding constraint on AI is no longer chips or capital; it is power."

Emerald AI's answer is a software platform called Emerald Conductor, which schedules AI computing workloads against the batteries and onsite power generation available at a given facility. When the local grid comes under stress, the system reduces the facility's power draw automatically, while keeping critical training and inference jobs running uninterrupted. Sivaram described the underlying bet behind the company's founding directly: "We founded Emerald AI on the conviction that the intelligence driving the AI revolution could solve its own greatest bottleneck: power. Our demonstrations around the world proved that data centers can adjust their power use precisely when the grid needs relief, without compromising critical computing workloads. Today that technology runs commercially at full data center scale, and this financing lets us take it everywhere AI is built, so the AI era can accelerate while the grid becomes more reliable and more affordable for the communities it serves."

In effect, Emerald AI's software gives utilities a data center they can dispatch somewhat like a generator, temporarily dialing capacity up or down as grid conditions shift, rather than treating the facility as a fixed load that must be supplied continuously regardless of overall system stress. That flexibility matters because it changes what utilities are willing to approve. Rather than requiring every megawatt of a new data center's capacity to be available around the clock from day one, load flexibility gives grid operators room to interconnect facilities faster and at larger scale, since the utility retains the option to temporarily curtail flexible portions of that load during genuine periods of peak demand.

Having completed five commercial demonstrations globally, Emerald AI has now moved past pilot testing into full commercial deployment, with its software running at multi‑megawatt, full data center scale for customers spanning leading AI companies, data center operators, and electric power utilities. The company has partnered with Silicon Valley Power on a Flexible Load Interconnection Program designed to give data centers faster grid access in exchange for verified, dispatchable flexibility. In Manassas, Virginia, Emerald AI is working with Digital Realty and Nvidia on a nearly 100‑megawatt project known as the Vera Rubin AI Research Factory, tested in collaboration with the Electric Power Research Institute, Dominion, and PJM Interconnection, and expected to come online later in 2026.

DCVC co‑founder Zachary Bogue pointed to the durability of the underlying shift Emerald AI is betting on, framing the company's technology as a way to make power flexibility a permanent structural feature of how data centers are built and operated going forward, rather than a temporary workaround for today's grid constraints. Siemens USA chief executive Ann Fairchild similarly emphasized that better coordination between AI computing loads and grid operations will only become more important as computing demand continues to climb.

The scale of the opportunity Emerald AI is targeting is substantial. Research from Duke University's Nicholas Institute has estimated that the existing US electric grid could accommodate nearly 100 gigawatts of additional large electricity loads if those loads were able to temporarily reduce consumption during periods of system stress, though that research examined flexible large loads generally rather than Emerald AI's specific technology. Separately, the International Energy Agency projects that data centers will drive nearly half of all US electricity demand growth through 2030, a trajectory that is putting sustained pressure on utilities and grid operators to find faster paths to new capacity than conventional transmission buildout allows.

With the new capital, Emerald AI plans to scale commercial deployments worldwide across its existing customer base of AI companies, data center operators, and utilities. The company's positioning at the intersection of AI infrastructure and energy markets, backed by an investor base that includes both leading AI hardware providers and major global utilities, suggests a bet that software‑orchestrated power flexibility is likely to become a standard feature of how AI data centers are financed, permitted, and operated going forward, rather than a niche add‑on layered onto conventional facility design. Whether that becomes true at the pace the AI industry's power needs demand will depend heavily on how quickly utilities and regulators are willing to formalize flexible‑load agreements as a standard part of new data center interconnection processes.