Corporate scorecards

Who builds the AI grid — and what they disclose

Environmental data from 14 data center companies — hyperscalers, operators, and developers — side by side. The gap between what they claim and what they report is the gap your community should ask about.

Hyperscalers

The four companies that build and operate their own AI infrastructure at the largest scale.

DC TWhPUEScope 2 (location) Scope 2 (market)Water (Mgal)Renewable
Google42.41.0915.1M2.8M10,86965% CFE
Meta18.11.086.0M1K825100% renewable
Microsoft29.8*1.189.7M2.7M6,441100% renewable
Amazon (AWS)38.0*1.1411.9M04,600100% renewable

* DC-only TWh and location-based Scope 2 are estimates; these companies do not break out data-center-only electricity.

Data center operators & developers

Companies that build, own, or lease hyperscale data center campuses — the facilities where AI workloads actually run.

DC TWhPUEScope 2 (location) Scope 2 (market)Water (Mgal)Renewable
Equinix8.17*1.392.6M253K1,10496% renewable
Digital Realty1.383.3M948K93% renewable
EdgeConneX1.661.3302590% renewable
STACK Infrastructure1.35295K100% renewable (since 2021)
CyrusOne1.46402K61.6% carbon-free
Vantage Data Centers1.2649K4 of 34 campuses >99% renewable

— = not disclosed or not applicable. * = estimated figure.

Transparency gaps

These companies build or operate significant data center capacity but publish little or no environmental data. That silence is itself a data point — ask why.

ScaleDisclosure level
CoreWeave43 data centers · >850 MW active · ~3.1 GW contractednone
QTS (Blackstone)>2 GW contracted capacitypartial
Switch (DigitalBridge)Las Vegas 315 MW · Tahoe Reno up to 2 GW plannedmarketing only
Compass DatacentersTotal capacity not disclosedpartial