Skip to content

Emerald AI

Grid-responsive power control for AI data centers

Raised to date

$147.1M

of $210.3M offered across 3 Form D filings

Still open
$63.2M
Investors
23
Last filed
Aug 3, 2026

About

Emerald AI builds software that lets AI data centers act as flexible loads on the electric grid instead of fixed ones. Its platform orchestrates AI compute workloads in response to real-time grid conditions, shifting or throttling them across time and geography and coordinating on-site batteries so facilities can cut power draw during grid stress without breaking service commitments. A demonstration in Phoenix with a 256-GPU cluster reduced consumption roughly 25% over a three-hour stress window and returned to baseline afterward. The company was founded in 2024 by CEO Varun Sivaram and is backed by Radical Ventures, NVIDIA's NVentures, CRV, and Salesforce Ventures.

Rather than adding generation or storage, it makes AI compute itself dispatchable, and pairs that with auditable measurement and verification so utilities can treat a data center as a credible, contractible grid resource.

Funding

$150M round, first sold Jul 2026$90.2M sold$59.8M still openJul 2025Jan 2026Jul
Now
  • EquityFiled Aug 3, 2026Form D
    $90.2M/ $150M
  • EquityFiled Feb 20, 2026Form D
    $22.7M/ $25M
  • EquityFiled Aug 13, 2025Form D
    $34.2M/ $35.3M

Product

Emerald Conductor

Energy and compute orchestration software

A platform the company describes as an operating system for AI and power, sitting between the grid and the data center and adjusting AI workloads in real time to match available power.

  • Temporal flexibility: pausing or slowing batchable AI workloads during grid stress periods
  • Spatial flexibility: shifting workloads between regions with more available power
  • Coordination of on-site batteries and other energy resources alongside compute

Built for

Data center developers and operators, plus the utilities, grid operators, and AI/cloud providers on either side of their interconnection

People on file