Anthropic raises $3.5B at $61.5B valuation

Original Article ↗ Hacker News Discussion ↗

Perceived product quality and use cases

  • Multiple commenters report Claude as “head and shoulders above” competitors, especially for coding and agentic workflows.
  • Claude Code is praised as a strong terminal-based coding tool, often preferred to Cursor/Windsurf/Aider by people who like staying in their own IDE (e.g., PyCharm) and using the terminal as the AI interface.
  • Some users find Claude’s code abilities significantly better than open models; others claim DeepSeek or QwQ match or surpass it for their own coding tasks.
  • Cost is a recurring theme: one user spends ~$25/day and hits Claude Code’s $100/month cap; another finds Claude Code ~4x more expensive than Aider due to broader context usage, though recent updates to respect .gitignore are noted.

Valuation, profitability, and bubble worries

  • Many see the $61.5B valuation as extremely rich or “bubble-like,” especially given:
    • Reported 2024 figures mentioned in-thread: ~$908M revenue vs. ~$5.6B training spend.
    • Heavy ongoing capex required to stay ahead of open models.
  • Skeptics question whether there’s a realistic path to ROI in a market with many free or cheap alternatives (DeepSeek, Qwen, Llama).
  • Others argue this is a classic “gold rush / greater fool” phase: investors are buying lottery tickets on the chance Anthropic (or peers) becomes the next Microsoft/Google.

Open-source and foreign competition

  • DeepSeek and Qwen are cited as strong, free or low-cost models; some claim they’re “close” to Claude on benchmarks and good enough for many tasks.
  • Counterpoint: several developers report Claude still feels “several times better” in real coding workflows, despite benchmark parity.
  • Debate over sustainability:
    • One side: closed leaders must overspend on training just to stay marginally ahead of an open-source “floor,” making profits elusive.
    • Other side: open models also require huge training budgets; “someone is paying” for that progress.

Moats, market structure, and business models

  • Comparisons made to:
    • Microsoft vs. Linux (value from distribution, lock-in, and integrated products rather than raw tech).
    • Cloud (AWS/GCP/Azure oligopoly) and telco “big 3” dynamics.
  • Proposed moats:
    • Distribution and enterprise relationships.
    • Government contracts and potential protectionism.
    • Proprietary training data/content deals (e.g., Reddit-style licensing) and eventual replacement of search with AI.
  • Some predict LLMs become commodity infrastructure; the winners will be the best companies (products, cost structure, distribution), not just best models.

Investment access and secondary markets

  • Commenters note ways small investors might get exposure:
    • Funds like the “innovation fund” mentioned.
    • Secondary marketplaces (EquityZen, Hiive), with warnings about high premiums.
  • Some see the new round as “a steal”; others think any >$70B valuation is unjustifiable.

Macro and systemic risk concerns

  • Several worry about an AI/tech bubble whose collapse could resemble the dotcom bust, with broader economic fallout and misallocation of capital.
  • Others argue capital will simply rotate to the next theme; disruption would be temporary.

xAI valuation aside

  • Brief tangent asks why xAI’s valuation is rising:
    • Some attribute it more to its owner’s political/institutional leverage and procurement influence than to technical strength.
    • Again framed as part of the same lottery-ticket dynamic for large funds.