(MENAFN- EIN Presswire) EINPresswire/ -- Today, Epoch AI launched the Frontier Data Centers Hub, an independent database tracking the construction timelines of major US AI data centers through high-resolution satellite imagery, permits, and public documents. The analysis reveals that five facilities will likely cross the 1 gigawatt threshold in 2026.

Recent announcements of 1 GW data centers, such as Meta Prometheus and xAI Colossus 2, are just the beginning. Many people aren't aware that Amazon's recently announced Project Rainier facility will also reach 1 GW next year.

"Data centers typically come online in stages, expanding power usage over months or years. Companies rarely clarify this, which creates confusion about true capacity," said Ben Cottier, researcher at Epoch AI.

The database shows several other data centers will likely cross 1 GW of total facility power in 2026:
- Anthropic–Amazon New Carlisle: January 2026
- xAI Colossus 2: February 2026
- Microsoft Fayetteville: March 2026 (borderline 1GW)
- Meta Prometheus: May 2026
- OpenAI Stargate Abilene: July 2026

At 1 gigawatt scale, a single AI data center consumes enough electricity to serve around one million U.S. homes.

The Frontier Data Centers Hub combines high-resolution satellite imagery, construction permits, local news reports, power agreements, and company documents to track the physical infrastructure changes at each site over time. The result is the most detailed publicly accessible and independent database of how much power, land, and hardware the biggest AI labs are actually deploying-and when.

"Two years ago, we called on top AI labs to disclose their compute. They didn't. So we did it for them," said Yafah Edelman, Head of Data & Trends at Epoch AI. "Using satellite imagery, permits, and company filings, we've tracked these major facilities to reveal the actual build-out pace."

"Compute is the foundation of AI. We at Epoch AI are tracking the construction of the largest AI data centers to help everyone grasp the scale of the next wave of AI infrastructure," said Jaime Sevilla, CEO of Epoch AI.

The 13 large U.S. data centers tracked in the hub account for a substantial share of total compute stock globally: about 2.5 million (~15%) of the roughly 15 million H100-equivalents (H100e) that have been delivered to customers in the past several years as of late 2025. H100e is a measure of compute processing performance, expressed in units of NVIDIA H100 processing power.

“A key thing that makes large-scale data centers discoverable is their cooling infrastructure. It's hard to hide the massive chillers required to prevent billions of dollars in equipment from overheating. By analyzing satellite imagery and correlating it with permit documents and hardware specifications, we can estimate capacity within surprisingly narrow margins," said Ben Cottier, researcher at Epoch AI.

The hub is freely accessible at .

About Epoch AI
Epoch AI is a nonprofit, tax-exempt 501(c)(3) research organization. Our research advances evidence-based discussions on the drivers, progress, and consequences of AI development and deployment, promoting informed policy and decision-making.

For more information, visit epoch or follow us on X at @EpochAIResearch.

MENAFN04112025003118003196ID1110294910