Anthropic
The AI company behind Claude (claude-opus-4-8) — the model substrate beneath nearly everything this wiki tracks. Given a page now because it anchors a recurring “substrate provider” thread (capability and the economics/cost that determine the pattern’s viability — see synthesis).
Commercial standing (May 2026)
Per anthropic-valuation:
- ~$965B post-money valuation (Series H, $65B raised; Altimeter, Dragoneer, Greenoaks, Sequoia) — surpassing OpenAI’s ~$852B (though OpenAI’s figure is two months older, so the comparison is asymmetric).
- Revenue ~doubled in six months to ~$9B annualized (from ~$4B).
- Growth attributed largely to enterprise adoption of Claude Code for software development — the agent surface tracked in agentic-tooling-wiki (claude-cowork / claude-managed-agents, now cross-wiki).
The cost side
- Claims one quarter of operating profit, but with questions about accounting.
- Vast compute commitments: “hundreds of billions” to infrastructure partners over the decade, plus $1.5B/month to SpaceX.
- AI profitability remains unproven long-term. IPO expected 2026.
Why it’s in this wiki
research-wiki’s thesis explicitly flags that the LLM-wiki pattern’s viability “depends on model capability and cost.” Anthropic’s economics are the cost/viability data point on that thread, just as claude-opus-4-8‘s honesty gains are the capability data point. Tracked lightly — this is business news, relevant only for its bearing on substrate durability, not a core part of the knowledge-management thesis.