Emerald AI Raises $150M to Turn Data Centers Into Grid-Flexible Assets

Emerald AI has raised $150 million in an oversubscribed Series A round at a $1.05 billion valuation, co-led by Energize Capital and DCVC, with participation from a group of investors including 12 Fortune Global 500 companies. The company's software dynamically adjusts data center power consumption in response to grid conditions, an approach it says could unlock more than 100 gigawatts of untapped capacity on the existing US power grid. The company has completed five commercial demonstrations across Arizona, Illinois, Virginia, Oregon and London, and has moved into commercial deployment at full data center scale.

 

Why Demand Flexibility Addresses a Genuine Timeline Mismatch

 

Building new grid infrastructure, transmission lines, substations and generation capacity, can take a decade or more to plan, permit and construct, while data centers are projected to account for nearly half of US electricity demand growth through 2030, according to the International Energy Agency figures cited in the release. That mismatch between how quickly physical grid infrastructure can be built and how quickly AI-driven electricity demand is growing creates a genuine bottleneck: even if utilities and grid operators fully commit to expanding capacity, the physical construction timeline cannot keep pace with the speed at which data center developers want to bring new AI infrastructure online.

Emerald AI's Conductor platform addresses that mismatch through a different mechanism entirely, rather than requiring new physical infrastructure, the software dynamically reduces a data center's power draw during periods of grid stress by adjusting AI computational workloads and onsite energy resources, while protecting the performance of critical workloads. Founder and chief executive Varun Sivaram framed this as founded on "the conviction that the intelligence driving the AI revolution could solve its own greatest bottleneck: power," arguing that demonstrations proved data centers "can adjust their power use precisely when the grid needs relief, without compromising critical computing workloads." That capability effectively lets data centers connect to and draw from the grid at larger scale than a fixed, inflexible demand profile would otherwise permit, since a facility capable of temporarily reducing its draw during genuine grid stress periods requires less permanently reserved grid capacity than one that draws a constant maximum load regardless of grid conditions.

 

Why the Silicon Valley Power Programme Represents a Genuinely Novel Mechanism

 

Emerald AI's partnership with Silicon Valley Power to launch what the release describes as the first-in-the-nation Flexible Load Interconnection Program creates a specific regulatory trade: data centers receive expanded grid access in exchange for verified, dispatchable flexibility, meaning the utility grants a facility greater connection capacity than it might otherwise approve, conditioned on that facility demonstrably reducing its draw when called upon during grid stress events.

That structure represents a genuinely different interconnection model than the standard approach, where a utility grants a fixed connection capacity based on a facility's maximum anticipated draw and treats that draw as static regardless of broader grid conditions. By instead tying expanded grid access directly to verified flexibility, the programme gives data centers a concrete incentive to invest in and maintain genuine demand-response capability, since greater flexibility translates directly into greater permitted grid access, a mechanism that could meaningfully accelerate how quickly new data centers can secure grid connections if adopted more broadly by other utilities facing similar capacity constraints.

 

UPCOMING WEBINAR

AI academy: Two real ESG workflows saving days of manual work

Online, 03 September 2026 · Turning days of emissions data work into minutes with audit-ready AI workflows

 

Why the Vera Rubin AI Research Factory Serves as a Concrete Scale Test

 

Emerald AI's work with Digital Realty and NVIDIA on the nearly 100-megawatt Vera Rubin AI Research Factory in Manassas, Virginia, described as the world's first power-flexible AI factory, gives the company's technology a genuine large-scale proving ground. The project is being tested in collaboration with EPRI (the Electric Power Research Institute), Dominion Energy, and the PJM Interconnection, the regional grid operator covering much of the mid-Atlantic and parts of the Midwest, and is slated to come online later this year.

That combination of a nearly 100-megawatt commercial facility tested alongside both a major regional utility and the grid operator responsible for managing electricity across a multi-state region gives this specific deployment considerably more weight as a scale validation than the company's earlier, presumably smaller demonstration projects, since PJM and Dominion's direct involvement suggests the flexibility Emerald AI's software delivers will be assessed against the genuine operational requirements of a large, complex regional grid rather than a more controlled or limited pilot environment.

 

Why the Investor Roster Signals a Coordinated Bet Across the Value Chain

 

The round's participant list spans companies representing distinct segments of the AI and energy value chain simultaneously: chip maker NVIDIA, cloud and AI infrastructure companies, energy companies including Aramco Ventures, GE Vernova and RWE, and specialist climate and deep tech investors including Energy Impact Partners and Lowercarbon Capital. DCVC co-founder Zachary Bogue framed the technology's significance as turning "data centers into grid-responsive assets instead of energy-hogging liabilities—increasing America's strength in AI, decreasing rises in electrical bills for communities, and protecting the environment."

That breadth of investor participation, spanning chip manufacturers, cloud providers, traditional and renewable energy companies, and specialist venture funds, suggests the round attracted capital from companies with direct commercial stakes in solving the AI power constraint from multiple different angles simultaneously, rather than representing a typical financial venture round backed primarily by investors without direct operational exposure to the underlying problem the technology addresses. Energize Capital managing partner John Tough described the binding constraint on AI as "no longer chips or capital" but "power," framing software as "the fastest way through it."

 

What Comes Next

 

Emerald AI was recently named to the 2026 TIME 100 Most Influential Companies list and recognised as a 2026 Technology Pioneer by the World Economic Forum, external validation markers that extend beyond the company's own investor and customer relationships. Whether Emerald AI's technology performs reliably at the scale the Vera Rubin AI Research Factory and other planned large-scale deployments require once fully operational, and whether the Silicon Valley Power interconnection model gets adopted by additional utilities facing similar AI-driven grid capacity constraints, will determine how significantly this funding round advances the broader goal of unlocking the more than 100 gigawatts of existing grid capacity the company says its approach could make available to AI infrastructure development.

 

Source: Emerald AI