🏭 AI Factory Case Studies · Overview

AI Factories at a Glance: Nine Bets on Building Intelligence

The world's largest AI factories on one page: xAI's Colossus, OpenAI's Stargate Abilene, Microsoft's Fairwater, Amazon's Project Rainier, Meta's Prometheus and Hyperion, Google's TPU fleet, CoreWeave and the neoclouds, China's supernodes and the sovereign projects. Find each on the map, sort them in the table, compare their timelines and pick a reading track.

Explore the map ↓ Start at Part 1 →
⚠️ About the numbers. Everything here comes from the ten articles, which cite their sources and flag secondary claims. Sizes are mostly Epoch AI estimates (as of October 2026), because most owners do not publish megawatts. Dates are approximate to the month. Follow the links to each part for detail.

🗺️ Where the AI factories are

A schematic world map (equirectangular, not exact). Most of the US sites cluster in the Midwest and South, shown in the inset. Tap a dot.

World

United States (inset)

Tap a site.

📋 Compare the sites

Click a column heading to sort. Click a site to read its case study.

⏱️ Timelines side by side

Each bar runs from the first concrete step to today; the white dot is first compute (dashed if still planned) and the dashed line runs to the next planned milestone. Tap a row.

Tap a row.

🎲 The nine bets

For the scoreboard, five patterns and the risk chain, read Part 10: Nine AI Factories, One Map →

📚 Reading tracks

🔤 Glossary

The other series in "Build the Machine": 🖥️ AI Compute Stack at a glance · 🏭 Producing Intelligence at a glance · 🧑‍🔧 Who Runs the AI Factory at a glance