Anthropic Is Now Worth More Than OpenAI (Gizmodo)

A Gizmodo report (30 May 2026) on anthropic overtaking OpenAI by valuation.

Facts

• Anthropic ~$965B post-money (Series H; $65B raised, led by Altimeter, Dragoneer, Greenoaks, Sequoia) vs OpenAI ~$852B (most recent disclosed, ~2 months older — an asymmetric comparison).
• Driver: “mind-blowing growth” from enterprise adoption, “particularly around its Claude Code product for software development.”
• Revenue doubled in six months to ~$9B annualized (from ~$4B).

Critical context (the article’s caveats)

• Anthropic claims one quarter of operating profit, but “questions remain about accounting methods.”
• Enormous compute obligations: “hundreds of billions” to infrastructure partners over the next decade; $1.5B/month to SpaceX.
• AI profitability is unproven as a long-term strategy. Both firms expected to IPO in 2026, which will give clearer market valuations.

Update (2026-07-01) — corroborated by a T1 primary, and the IPO moved

A freshness pass confirms and advances the Gizmodo snapshot:

• The $965B / $65B Series H is now T1-sourced. Anthropic‘s own release (anthropic.com/news/series-h) states the $65B Series H at a $965B post-money valuation (Altimeter/Dragoneer/Greenoaks/Sequoia) — so the headline figure no longer rests on the T4 trade-press report alone. OpenAI’s ~$852B (late-March, $122B round) stands as the most-recent disclosed comparison; the “asymmetric comparison” caveat holds.
• IPO: from “expected” to filed. Anthropic has confidentially filed draft IPO paperwork (Fortune, 1 Jun 2026), with a possible listing as soon as fall 2026; OpenAI is reported to be preparing its own confidential filing on a similar fall timeline. The page’s “both expected to IPO in 2026” line is now a concrete filing, not a forecast.
• Revenue: later reporting cites a higher run rate than the ~$9B annualized here (figures vary by metric and month and are contested — recorded, not overwritten; the compute-obligation and accounting-method caveats below still apply).

Bearing on synthesis: the substrate provider’s commercial strength is now primary-sourced and heading to public markets, which will replace these private-round guesses with a market price — the clearest future test of the “does the cheap-maintenance LLM bet stay cheap” thread.

Why it’s here

Routed to research-wiki (over a park) because it lands squarely on the existing “model substrate — capability and cost” thread in synthesis: the substrate provider’s commercial strength and its unproven, compute-heavy economics directly bear on whether the cheap-maintenance LLM-wiki bet stays cheap. Filed via anthropic. Off the core knowledge-management thesis; tracked lightly.