Emerald AI, Google, and NVIDIA Launch AEMA to Flex AI Data Centers on the Grid

Emerald AI, Google, and NVIDIA announced the AI Energy Management Alliance (AEMA) on September 16, 2026, a coalition aimed at connecting more AI data centers by making their electricity demand flexible.

AEMA’s pitch is grid-responsive load, not just bigger power plants.
Per NVIDIA’s blog, founding members want data centers that can dynamically manage electricity use — shifting compute, using storage or paired generation, and responding when the system is constrained — so operators treat large AI sites as controllable resources instead of flat, inflexible loads. The alliance frames the work as technology-neutral and performance-based: response speed, duration, predictability, and emergency behavior matter more than any single vendor stack.

TechCrunch’s September 17 report widens the roster and states the 100 GW claim.
Beyond the founding trio, TechCrunch says Anthropic joined along with utilities including AES, Constellation, National Grid, and NRG Energy. AEMA argues that pausing noncritical tasks and shifting compute could allow about 100 gigawatts of additional data centers to interconnect. That figure is an alliance claim about demand-response headroom, not an audited grid study Circuitry independently verified.

Emerald AI’s software is the operational glue in secondary coverage.
TechCrunch describes Emerald AI coordinating utility requests with data centers that can pause work or move loads to sites with spare grid capacity — an alternative to leaning on diesel backup for peaks. The piece notes Emerald AI recently raised a $150 million Series A led by Energize Capital and DCVC. NVIDIA’s post emphasizes convening AI platforms, operators, power producers, utilities, and regional grid operators around shared interconnection and reliability rules.

Flexibility is meant to speed interconnection, not erase new generation needs.
Both primaries stress reliability: ride-through, curtailment, and contingency obligations should be clear before a facility connects. TechCrunch quotes Emerald AI chief scientist Ayse Coskun saying the approach can blunt — but not eliminate — the need for new generating sources.