Leopold Aschenbrenner

“The AGI race has begun. We are building machines that can think and reason. By 2025/26, these machines will outpace college graduates.”

Deadline runs to the end of 2026. Frontier models beat college graduates on many benchmarks and exams, but whether they 'outpace' graduates as workers is contested and not yet judgeable.

“AGI by 2027 is strikingly plausible. That doesn't require believing in sci-fi; it just requires believing in straight lines on a graph.”

Deadline is end of 2027. The essay's own operationalization is an AI that functions as a 'drop-in remote worker'; no such system exists as of late 2026.

“As the race to AGI intensifies, the national security state will get involved. The USG will wake from its slumber, and by 27/28 we'll get some form of government AGI project. No startup can handle superintelligence.”

Deadline is end of 2028; the essay says 'by late 26/27/28 it will be underway.'

“The 2028 cluster will be 10GW — that's 10 million H100 equivalents, costing hundreds of billions of dollars.”

Target is end of 2028.

“By the end of the decade, they will be smarter than you or I; we will have superintelligence, in the true sense of the word.”

Target is end of 2029.

“By 2030, you get the trillion-dollar cluster using 100 gigawatts, over 20% of US electricity production. That's 100 million H100 equivalents.”

Target is end of 2030.