The US–China AI arms race: is “The Project” coming?

The thesis

A core strand of Situational Awareness is geopolitical: as AGI nears, algorithmic secrets and model weights become the decisive strategic asset, the US must "lock down the labs," and the national-security state inevitably runs a centralized AGI effort — "The Project" — by roughly 2027/28.

Where it stands

The prediction is graded Open: national-security attention has grown (export controls, security requirements, defense interest), but there's no centralized, Manhattan-Project-style government AGI effort, and frontier development remains led by private labs. The 2027/28 deadline hasn't elapsed.

Why the premise is shaky

The Project's logic depends on AI being concentrated and controllable. But capability diffused instead — open-weight Chinese models trail by only ~3–6 months. If intelligence is cheap and diffuse, compute concentration buys less geopolitical leverage than the essay assumes, and the arms-race logic that motivates a national Project weakens. So the "arms race" is real in attention, but the durable-moat premise underneath it is the part that's already looking wrong.

Frequently asked questions

In attention, yes — export controls, security requirements, and defense interest have all grown. But as of mid-2026 there's no centralized US government AGI effort ('The Project'), which Aschenbrenner predicted for 2027/28; it's graded Open.

Aschenbrenner's term for a predicted Manhattan-Project-style US government AGI effort, which he expected the national-security state to launch by roughly 2027/28.

Thinner than assumed. Open-weight Chinese models trail by only ~3–6 months, so the durable proprietary moat that the arms-race logic depends on is already looking wrong.