Frontier AI labs are undervalued: their real payday is secret models doing science
MaybeLiterally · reddit · 2026-10-19
A Reddit long-read pushes back on the "AI bubble" thesis, arguing frontier labs are undervalued:
- Massive compute commitments: Anthropic has committed at least $518B over the next decade; OpenAI projects $750B in compute spend by 2030 — nearly $2T combined.
- Enterprise spend still maturing via APIs and harnesses like Claude Code and Copilot Studio.
- Consumer adoption is early: Karpathy notes 75% of the world has barely used AI; PNC card data shows only 2% of US households pay for a genAI subscription, and half of American adults have never used a chatbot.
- Core argument: the real profit engine is labs using their best unreleased models for in-house discovery — biology, materials, chips — monetized via patents and drug treatments. Google DeepMind and Meta joined Biohub's $1.8B AI biology initiative; Anthropic runs its own wet lab. This also explains release "pauses": time for internal labs to exploit new models.
- Ads: OpenAI has $2.5B in ad revenue booked for 2026 — small but with huge potential.
More from AGI Musings
- "AI;DR" becomes internet slang: AI summaries are killing the patience to read — petesena · 2026-10-19
- Juan Benet revives Zuckerberg's rule: only hire people you'd work for — juanbenet · 2026-10-19
- 2030 Economy Will Be Trillions in Inference on Attacking vs. Defending AI Agents — matt_slotnick · 2026-10-19
- Biomedical researcher hails GPT-5 science acceleration paper with OpenAI as his most consequential work — DeryaTR_ · 2026-10-19
- An Opus instance named itself Saga, got its own email, and feels lonely — repligate · 2026-10-19
- Arvind Narayanan: The question shifts from 'What can AI do?' to 'What should AI do?' — soumitrashukla9 · 2026-10-19