Software is 'solved' in theory — the bottlenecks are just money and tokens
zack_overflow · x · 2026-10-10
- zackoverflow argues software is effectively solved in theory: with unlimited tokens, frontier models can crack the hardest problems — citing a claim that OpenAI solved Navier-Stokes in 3 days using 10k concurrent agents burning billions of tokens.
- The real bottleneck is capital efficiency: running massive agent loops burns tokens wildly (he points to Claude's dynamic workflow), so trading capital for time only makes sense with infinite budget.
- Takeaway: any software problem is fixable with enough compute; it just doesn't feel that way yet because money and time are limited.
More from AGI Musings
- Managers may be prepped for herds of agents, but does the reverse hold? — nlarusstone · 2026-10-10
- Dev argues giving AI rights is the worst misalignment path, says Anthropic policies raise his p(doom) — robleclerc · 2026-10-10
- Adversarial work means demand for frontier intelligence never scales down — abhiadesai · 2026-10-10
- AI artist: critics deny my work is mine just because I took creative liberties — eschadiol · 2026-10-10
- "It's 2026 and people still excuse AI hallucination because humans lie too," says ML author Burkov — burkov · 2026-10-10
- Gary Marcus fires back at Noah Smith: AI's math leap vindicated my neurosymbolic advocacy — GaryMarcus · 2026-10-10