Anthropic raises $65B in Series H funding at $965B post-money valuation

Original Article ↗ Hacker News Discussion ↗

Anthropic’s new $65B Series H round, valuing the AI lab at $965B, is prompting both awe at its explosive revenue growth and concern over whether such near‑trillion‑dollar private valuations are sustainable. Commenters probe how many funding rounds a company can take before IPO, how investors and employees actually get liquidity, and whether “run‑rate revenue” and heavy subsidies on usage mask underlying profitability and risk. The funding is also framed within the broader AI race against OpenAI and Google, amid signs that enterprises are starting to rein in soaring AI spend and question the long‑term economics of large‑scale model usage.

Funding rounds and IPO timing

  • Commenters note funding rounds can be effectively unlimited (A–Z, then AA etc.); Databricks is cited as an extreme case.
  • IPO is seen as the main liquidity event, but many late-stage companies now stay private for a long time, using secondaries and tender offers for partial liquidity.
  • Some argue Anthropic is smart to raise at peak hype to secure a long runway before any downturn or IPO.

Valuation, upside, and employee equity

  • A $965B private valuation is viewed as astonishing and possibly bubble-like; some compare it to sovereign wealth funds or national infrastructure budgets.
  • Debate on whether joining now still offers meaningful equity upside; some think VCs will still expect 3x+, others think most upside is already captured.
  • Several advise employees to sell equity early and diversify rather than overconcentrate in employer stock.

Revenue, run-rate metrics, and profitability

  • Anthropic’s reported run-rate revenue jumps rapidly ($9B → $14B → $30B → $47B within months) and is described as “unfathomable.”
  • Some see this as evidence of explosive real demand; others see “vibes,” question how run-rate is calculated, and worry a few huge customers could be skewing numbers.
  • There is disagreement over whether Anthropic is truly near profitability; linked pieces characterize the claim as both plausible and “murky.”

AI usage economics and “tokenmaxxing”

  • Enterprises reportedly track AI spend closely; some are capping per-engineer usage (~$100/week) or pulling back on experiments.
  • “Tokenmaxxing” = measuring employees by tokens burned, encouraging wasteful usage. Many expect this to be unsustainable and a risk to Anthropic’s revenue if widely practiced.
  • Others point out that subscription plans are heavily subsidized relative to API list pricing, suggesting current unit economics may not be stable.

Competitive landscape

  • Some say Anthropic has “blown past” OpenAI in revenue/valuation; others note valuations are still close and see both coexisting.
  • Arguments:
    • Pro-Anthropic: better coding experience, strong enterprise momentum, aggressive marketing/branding (“Claude is the good one”).
    • Pro-OpenAI: broader customer base, perceived reliability, strong consumer brand (ChatGPT as a generic term), strong coding product (Codex).
    • Google is often cited as the dominant consumer player via Gemini’s integration into search/Android, and as a major competitor in foundation models.
  • Some predict the market can’t support multiple frontier labs; others think several “good enough” models will coexist.

Compute, datacenters, and hardware

  • Debate over whether owning datacenters/GPUs (ascribed to OpenAI in the thread) is an advantage vs. relying on hyperscalers (Anthropic).
  • One side: dedicated capacity lowers cost and hedges shortages.
    Other side: locking into hardware is risky; flexible use of TPUs/GPUs from multiple vendors may age better.
  • Power and supply-chain constraints (HBM, optics, fabs) are called out as deeper bottlenecks than GPUs alone.

Market structure, IPOs, and index funds

  • Several see modern IPOs as a “dumping ground” where private investors extract most upside, then offload at peak multiples to retail and index funds.
  • Index inclusion (S&P 500) is viewed as forcing pensions/401(k)s to become bagholders of overvalued tech, with parallels drawn to past bubbles.
  • Others counter that past IPOs (Google, Facebook, Tesla) looked overpriced at the time but proved lucrative in hindsight.

Bubble and societal concerns

  • Some call current AI funding “deeply troubling,” likening circular money flows (e.g., vendors investing in each other) to a bubble that could “implode harder than housing.”
  • Worries that VC-subsidized AI (like past “subsidized burrito delivery”) will end, leading to cost shocks and cuts.
  • Broader anxieties: potential mass job displacement, lack of tangible public benefits vs. alternative uses of capital (e.g., infrastructure), and the risk that non-participants (ordinary workers, retirees) bear losses via pensions and index funds.