10x compute for one company may suffice to train a 200T-param ASI, argues blogger
Dr_Singularity · x · 2026-09-05
Blogger @DrSingularity argues that 10x more compute available to a single company — OpenAI, Anthropic, Google, Microsoft, or SpaceX — could be enough to train ASI, likely a 200T+ parameter model combining pretraining, RL, and inference-time compute.
He extends the thesis to a civilizational scale: a civilization's advancement is a function of its compute. A civilization with 1000x our compute — let alone a million or a billion times — would operate at an entirely different level, and the yottaflop era could solve disease, aging, poverty, and pollution.
Purely speculative macro commentary with no new data, but a representative take in the compute-scaling discourse.
More from AGI Musings
- The 4 layers of an agent system: failures are often architecture problems, not prompting problems — blaizedsouza · 2026-09-06
- Podcaster who lost his team: intelligence solves execution, not coordination — lennysan · 2026-09-06
- GTA 6 will be the last mega game made without AI, argues dev commenter — gabriel1 · 2026-09-06
- AI safety reporter urges frontier lab staff to whistleblow, shares his own McKinsey experience — AaronBergman18 · 2026-09-06
- Your 99% Benchmark Score Is a System Score: Why GPT-6 Astra Numbers Blur Model vs Harness — algo_diver · 2026-09-06
- Economist Alex Weyl coins 'normalcy overhang': superintelligence arrives before daily life changes — GregCook2011 · 2026-09-06