Pomegra
Emerald AI raises $150M at $1.05B as Nvidia, Siemens, and GE Vernova back software that lets data centers flex their power draw to stabilize the grid.
- Emerald AI's $150M Series A is oversubscribed; Energize Capital and DCVC co-led at a $1.05B valuation, its first unicorn round.
- Emerald Conductor software dynamically throttles data center power without disrupting AI workloads, targeting 100+ GW of untapped U.S. grid capacity.
- Nvidia's Vera Rubin AI Research Factory in Manassas, Virginia - nearly 100 MW - will be the world's first power-flexible AI facility.
Lead
Emerald AI closed a $150 million oversubscribed Series A on August 25, 2026, reaching a $1.05 billion valuation less than two years after the company's founding. Energize Capital and DCVC co-led the round, which drew a wide list of strategic investors spanning the semiconductor, industrial, and utility sectors. The raise brings Emerald AI's total funding above $220 million, after an initial $24.5 million seed in July 2025 and a $22.7 million seed extension completed in early 2026.
What Does Emerald AI Actually Do?
The company's software platform, Emerald Conductor, sits between an AI data center's workload scheduler and the electrical grid, allowing the facility to cut or shift its power draw within seconds when the grid is under stress - without interrupting critical AI training or inference runs.
Historically, data centers are fixed electrical loads. Grid operators cannot rely on them to shed demand during peak stress events the way industrial customers can. Emerald AI's pitch is that a software layer changes that, converting passive consumers into dispatchable demand-response assets. The company projects that its approach can unlock more than 100 gigawatts of untapped U.S. grid capacity by enabling expanded interconnection agreements tied to verified, auditable flexibility.
Why Did This Round Attract Strategic Heavyweights?
The investor list describes the company's commercial strategy more clearly than any press release. Nvidia, Siemens, GE Vernova, Samsung Ventures, Aramco Ventures, Salesforce Ventures, RWE, JERA Ventures, Sabanci Climate Ventures, and more than a dozen others joined financial backers including Radical Ventures, Energy Impact Partners, Lowercarbon Capital, In-Q-Tel, Emerson Collective, and General Catalyst's scout fund.
For Nvidia, grid interconnection delays are a direct cap on how fast AI infrastructure can be built. Shorter queue times mean faster deployments of its hardware. For GE Vernova and Siemens - both major power equipment and grid services suppliers - taking a position in Emerald AI provides early visibility into whether software-based flexibility becomes a durable market or a bridge technology. The breadth of the cap table is also a form of customer development: every strategic investor is a potential deployment partner.
The Grid Math
U.S. power grids are already stretched by AI demand growth. New data center interconnection queues in major markets run to five years or longer. Emerald AI's core argument is that demonstrable, verifiable flexibility - power reduction that grid operators can dispatch on command - can move a project ahead in that queue, or unlock stranded capacity that would otherwise require years of new transmission construction.
The company has begun validating that premise with utility partners. Silicon Valley Power launched the country's first Flexible Load Interconnection Program in partnership with Emerald AI, granting data centers expanded grid access in exchange for confirmed demand reduction commitments. In Virginia, Digital Realty and Nvidia are commissioning the nearly 100-megawatt Vera Rubin AI Research Factory in Manassas - the world's first power-flexible AI facility - with PJM Interconnection, Dominion Energy, and EPRI all participating in verification protocols.
What Comes Next for Emerald AI?
Additional large-scale deployments are scheduled for later in 2026, and the Series A capital is expected to fund geographic expansion. Varun Sivaram, who founded the company in 2024 after building a research profile at the intersection of AI and energy, is now running a unicorn-valued company simultaneously backed by the semiconductor industry's largest company and the power sector's leading equipment manufacturers. That configuration is rare at Series A stage.
The next test is whether interconnection reform keeps pace with the software. Utilities and grid operators must design and operationalize the programs that reward flexibility - and that work moves on regulatory timelines, not startup timelines. The investor roster provides no shortage of advocates pushing it forward. Regulators set their own calendar.
Outlook
Emerald AI exits its seed phase with commercial deployments running, a verified technology stack, and strategic coverage across nearly every constituency it needs to scale. The 100-gigawatt figure the company cites is a long-run ceiling, not a near-term projection. Near term, the story is whether the Manassas deployment and the Silicon Valley Power pilot produce the kind of verified, repeatable results that accelerate utility adoption. If they do, the $1.05 billion valuation will look conservative. If regulatory frameworks stall, the backlog of planned deployments provides a reasonable runway either way.
