Data Insight
Nov. 19, 2025

The largest AI data center campuses will soon be a fifth the size of Manhattan

By Ben Cottier

AI data centers have a rapidly growing compute and energy demand, and their physical footprint is expanding to match. While the buildings that house IT equipment form the core of these facilities, data center campuses also need land for power and cooling infrastructure, parking areas, and access roads.

The scale of these developments is striking. The campus for xAI’s Colossus 1, which trained Grok 4, spans several Manhattan blocks. OpenAI’s Stargate campus in Abilene will dwarf this, reaching the size of Central Park once eight buildings are completed in mid-2026. Larger still is Meta’s Holly Ridge campus, reaching almost four times the size of Central Park by 2030.

Epoch's work is free to use, distribute, and reproduce provided the source and authors are credited under the Creative Commons BY license.

Learn more about this graph

We chose three notable examples from our Frontier Data Centers database: xAI Colossus 1 as one of the largest data centers today; OpenAI Stargate in Abilene as a well-known upcoming data center; and Meta Hyperion as the largest one planned in the United States.

Using the documented locations of these three data centers, we obtained free satellite imagery of the campuses from Sentinel-2 via Copernicus Browser. We used this browser to measure land area. The land area measurements were also informed by published photos and blueprints: see examples for xAI Colossus (p.6), OpenAI Stargate (p.157), and Meta Hyperion (p.33).

Analysis

Assumptions and limitations

Explore this data