🌎🎇 Geopolitics + superintelligence: 8 scenarios

In the global race to AGI, second place is the first loser

My fellow pro-growth/progress/abundance Up Wingers,

I want NASA astronauts to return to the Moon before any inaugural footprints by taikonauts from China’s National Space Administration.

I want the first fusion reactors to dot the American landscape and power America’s electrical grid.

And I want, getting to the point of this essay, the United States to dominate and win the artificial general intelligence race. The case is overwhelming. A world shaped by US-aligned machine intelligence may be messy and even risky. But a world shaped by a Chinese-developed AGI is far worse — and possibly irreparable to one where human freedom is again ascendant.

Yes, skepticism about near-term arrival of human-level AI is warranted. Business use-cases for current models remain a work in progress, at best.

No wonder prediction markets have pulled back. At Metaculus, strong AGI arrival has retreated from May 2031 to January 2033, while weak AGI moved from February 2026 to April 2027. Splitting the difference suggests AGI capable of replacing remote workers might arrive around February 2030. And that’s assuming these predictions aren't off by decades — and the puzzle is even solvable. It might not be.

The race is on, ready or not

Yet policymakers need not buy the “Singularity is nigh” hype to take AGI seriously. As argued in the new RAND report “How Artificial General Intelligence Could Affect the Rise and Fall of Nations: Visions for Potential AGI Futures,” even if AGI is still many years or even decades away, the geopolitical risks and rewards are large enough to merit planning and policy action now.