Emerald AI, a start-up that uses software to manage how much electricity data centers draw from the power grid, has raised $150 million, according to reporting by Sri Muppidi in The New York Times. The round was led by DCVC and Energize Capital and values the company at $1.05 billion.
The pitch is straightforward. Data centers running AI workloads are enormous, and largely inflexible, consumers of electricity. Emerald AI's software adjusts a facility's power consumption in response to conditions on the grid — throttling demand when the grid is strained rather than treating the data center as a fixed, always-maxed-out load.
The New York Times frames the company's ambition in unusually blunt terms: its coverage is headlined around an A.I. start-up that "aims to reverse the backlash against data centers." That backlash is the business context here. As AI companies race to build computing capacity, the facilities that house it have become a local political problem in the communities that host them, where residents and utilities weigh the strain on regional power supplies against the promised economic benefits.
Emerald AI's bet is that flexibility is the way out of that fight. If a data center can dial its consumption up and down alongside grid demand, it stops looking purely like a burden on the local power system and starts looking like something closer to a cooperative participant — one that can share capacity with everyone else during peak hours.
Investors are clearly buying that argument, at least at the valuation implied by this round, as reported by Techmeme's summary of the Times story.
Why it matters: the constraint on AI's expansion increasingly isn't chips or code but electricity and the willingness of communities to host the buildings that consume it — and Emerald AI is wagering that software, not just new power plants, can ease that squeeze.