Emerald AI

Emerald AI builds orchestration software that lets AI data centers flex their power use — shifting or curtailing compute and drawing on on-site batteries and generation — so they can connect to the grid faster and act as a resource rather than a strain. Backed by NVIDIA and piloted with utilities and power companies, it targets the tens of gigawatts of latent grid capacity that flexibility could unlock.

In August 2026 its software was named in a California Energy Commission project worth 8.5 million dollars at the San Diego Supercomputer Center, pairing Emerald AI orchestration with a solid-state transformer from Alderbuck Energy and San Diego Gas and Electric as the utility partner. The two-megawatt demonstration targets roughly 25 percent energy savings and a substantially smaller power-equipment footprint, and is one of the first named US sites to test grid-responsive AI compute alongside the power hardware feeding it.

In September 2026 Emerald AI formed the AI Energy Management Alliance with Google and Nvidia, joined by Anthropic and the utilities AES, Constellation, National Grid and NRG Energy, to make demand response a standard part of how data centers are built. The alliance argues that pausing non-critical work and shifting load between sites during grid peaks could let about 100 gigawatts of additional data centers connect. That figure is the alliance's own, not a measured result, and Emerald's chief scientist, Ayse Coskun, has said the approach reduces the need for new power plants without eliminating it. The company raised a 150 million dollar Series A led by Energize Capital and DCVC.