In Oct ’25 Epoch AI speculatively BOTEC-ed that OpenAI had the compute to run ~7M “human-equivalent digital workers” doing hour-ish-long SWE-related tasks like debugging, with very wide confidence intervals (90% CI 0.4 to 321M, mean 62M). Also in Q4 ’25 they estimated global AI compute capacity at ~20M H100-equivalents, about 18x OpenAI’s 1.1M, so assuming 44% inference allocation like OpenAI this is (very simplistically) 135M digital worker-equivalents, nearly 3x the 47M human SWEs worldwide, or 3.6% of the 3.73 billion global human labor force last year. Switching from stock to flow, Epoch’s estimate of 13.7M H100e added in 2025 converts to over 90M newly added digital worker-equivalents at the same 44% inference allocation, 3x the 30M net increase in human labor force globally and nearly double the 50M gross youth labor force added in 2025.
This reminded me of Stross’ Accelerando describing the 2020s:
Welcome to the early twenty-first century, human.
It’s night in Milton Keynes, sunrise in Hong Kong. Moore’s Law rolls inexorably on, dragging humanity toward the uncertain future. The planets of the solar system have a combined mass of approximately 2 x 1027 kilograms. Around the world, laboring women produce forty-five thousand babies a day, representing 1023 MIPS of processing power. Also around the world, fab lines casually churn out thirty million microprocessors a day, representing 1023 MIPS. In another ten months, most of the MIPS being added to the solar system will be machine-hosted for the first time. About ten years after that, the solar system’s installed processing power will nudge the critical 1 MIPS per gram threshold – one million instructions per second per gram of matter. After that, singularity – a vanishing point beyond which extrapolating progress becomes meaningless. The time remaining before the intelligence spike is down to single-digit years …
I hadn’t realised this bar had been surpassed, albeit for jagged SWEs instead of MIPS.
In Oct ’25 Epoch AI speculatively BOTEC-ed that OpenAI had the compute to run ~7M “human-equivalent digital workers” doing hour-ish-long SWE-related tasks like debugging, with very wide confidence intervals (90% CI 0.4 to 321M, mean 62M). Also in Q4 ’25 they estimated global AI compute capacity at ~20M H100-equivalents, about 18x OpenAI’s 1.1M, so assuming 44% inference allocation like OpenAI this is (very simplistically) 135M digital worker-equivalents, nearly 3x the 47M human SWEs worldwide, or 3.6% of the 3.73 billion global human labor force last year. Switching from stock to flow, Epoch’s estimate of 13.7M H100e added in 2025 converts to over 90M newly added digital worker-equivalents at the same 44% inference allocation, 3x the 30M net increase in human labor force globally and nearly double the 50M gross youth labor force added in 2025.
This reminded me of Stross’ Accelerando describing the 2020s:
I hadn’t realised this bar had been surpassed, albeit for jagged SWEs instead of MIPS.