Emerald AI Raises $150 Million at a $1.05 Billion Valuation to Flex Data Center Power
Emerald AI has raised $150 million at a $1.05 billion valuation, backed by investors like Nvidia, Siemens, RWE, and CIA-linked In-Q-Tel. Its software optimizes data center workloads to reduce electricity consumption during grid stress. The company claims this approach could unlock over 100 gigawatts of untapped capacity on the existing U.S. grid, significantly easing strain.
What the Software Does
Emerald Conductor prioritizes computing jobs based on latency tolerances, adjusting or pausing non-critical tasks when the grid is under stress. This differs from typical efficiency software, which focuses on cooling systems.
Founder and CEO, Dr. Varun Sivaram, previously held climate policy roles in the Biden administration, and Prof. Ayse Coskun, Emerald’s Chief Scientist, is a renowned computer science professor whose research pioneered this field.
The Claimed Impact
The technology could enable nearly half of U.S. electricity demand growth by 2030, according to the International Energy Agency. By revolutionizing power consumption, AI itself might find a solution to its biggest bottleneck.
Utility’s Perspective
Silicon Valley Power, serving numerous data centers in Santa Clara, faced energy capacity constraints. They collaborated with Emerald, implementing a Flexible Load Interconnection Program to balance grid demands.
Demonstrated Results
A field test in Phoenix (2025) showed a 25% reduction in power draw for an Nvidia GPU cluster during three hours of optimization. Across 33 experiments, the system successfully managed 212 jobs without compromising service quality.