The Physical Stack of AI · Datacenters and gigawatt campuses
You can describe Stargate as a joint venture (not a single building), name its four named principals, and explain why a 10 GW / $500B AI compute platform is a story about integration, not just capacity.
Stargate is the largest single infrastructure commitment in computing history, and easily the most misunderstood. It is not a building, not a company OpenAI owns, and not — yet — a single coherent network. It is a joint venture announced in January 2025 between OpenAI, SoftBank, Oracle, and Abu Dhabi's MGX, with a stated goal of $500B and 10 GW of dedicated AI compute capacity within four years.
By February 2026, ~$400B+ of that is committed across roughly 7 GW of announced sites: Abilene TX (the flagship, Oracle-built), five more US sites (Shackelford County TX, Doña Ana County NM, an unnamed Midwest site, Lordstown OH, Milam County TX), plus international expansions starting with Stargate UAE.
This chapter walks Stargate end-to-end: the announcement and who's in the JV; the Abilene flagship that turned the announcement into an operational training campus; the five new sites and which are funded vs aspirational; and the strategic argument for why OpenAI is willing to take on infrastructure obligations of this size in the first place. By the end you should be able to read any Stargate-related headline in 2026 and place it on a real map.
Type: multi-choice
Prompt: > Which of these is the most accurate one-line description of Stargate as of mid-2026?
Chapter contains 4 lessons.