Skip to content

🐂 ANTH — Company Profile (Teaching)

Teaching site, not investment advice. Private company does not file with SEC; figures come from official announcements + secondary reports, change frequently, so specific values are intentionally omitted — use the official announcement links below for verification.

Note: ANTH (Anthropic) is a private company, not publicly traded. Retail investors cannot directly buy Anthropic stock — indirect exposure mainly through AMZN ($8+$5+$20B invested, LEDGER F-027) or GOOGL ($40B invested, LEDGER F-028).

📌 Basics

Dimension Content Source
Core business Claude-series large models (Opus 4.7 + Sonnet 4.6 currently frontier) + Claude.ai consumer / API + Claude Code (terminal-native coding agent) anthropic.com
Founded / HQ 2021 / San Francisco, USA (the Amodei siblings + ex-OpenAI safety team spin-off) wiki
CEO Dario Amodei (co-founder, since founding); Daniela Amodei is President wiki
Primary "listing" Not public — Public-Benefit Corporation (PBC) structure, emphasizing safety / long-term view secondary
SEC filings Does not file (not public) —
Fiscal year Not disclosed (private) —

💡 Valuation / revenue: specific figures intentionally omitted — see LEDGER F-043, F-044 for anchors

📐 LEDGER anchors (frozen-in-time)

Per LEDGER 2026/05/26 snapshot:

  • Valuation: $380B post-money (Series G, 2026/02, F-043; previously $183B Series F 2025/09)
  • Revenue run-rate (2026/04 official): $30B+ (LEDGER F-044, "80x growth from previous year" per VentureBeat)
  • AMZN commitment: $8B already invested + $5B (2026/04 follow-on) + up to $20B future (F-027)
  • Anthropic-AWS commitment (2026/04/20 expanded): Anthropic pledged >$100B over 10 years in AWS infra; AWS to provide up to 5GW of Trainium + Graviton capacity (Trainium⅔/4 + tens of millions of Graviton CPUs); first 1GW coming online in 2026 (Trainium⅔)
  • GOOGL commitment: $40B (2026/04/24 announced) (F-028)
  • SpaceX/Colossus compute deal (per SpaceX S-1, 2026/05/20): Anthropic pays $1.25B/month through May 2029 for Colossus + Colossus II capacity (300MW + 220K+ Nvidia GPUs); total >$40B (either party may exit on 90 days' notice)
  • Claude current version: Opus 4.7 + Sonnet 4.6 (both officially confirmed at 1M context via API beta, F-095)

🏭 Supply Chain Coordinates (Part 1-C6 framework)

Upstream (Who ANTH Depends On)

Supplier What flows Key dependency
AMZN AWS primary cloud + Trainium in-house ASIC (Anthropic pledged >$100B over 10 years; AWS provides up to 5GW of Trainium⅔/4 + Graviton capacity) + capital $8+$5+$20B Major investor + primary cloud/training partner; Anthropic = AWS's large customer (circular chain)
SPACE (SpaceX) Colossus + Colossus II compute ~$1.25B/month through May 2029 (300MW + 220K+ Nvidia GPUs, per SpaceX S-1) Major inference compute supplier
GOOGL GCP as secondary cloud + capital $40B (2026/04) + TPU compute Strategic
NVDA Partial GPU usage (although AWS primarily promotes Trainium, Anthropic still uses some NVDA; SpaceX Colossus is also a Nvidia GPU cluster) Critical
MRVL Trainium ASIC IP partner (co-designs with AWS) Indirect

Downstream (Who Depends on ANTH)

Customer Relationship Concentration
API enterprise customers (Cursor, Glean, etc., enterprise AI startups primarily use Claude) API billing Increasing concentration (Cursor reportedly Anthropic's #1 API customer)
Claude.ai consumer + Pro subscriptions $20/month, ChatGPT-like model Diversified
Claude Code developers terminal-native coding agent Developer base
Bedrock (AMZN primarily promotes Claude) AWS customers use Claude via Bedrock Indirect

Competitors

  • Closed-source frontier: OPENAI GPT / GOOGL Gemini / XAI Grok
  • Open source: META Llama / DEEPSEEK (China) / Mistral / Qwen (BABA)
  • Enterprise AI apps: Glean / Harvey / Cursor (many use Claude as backbone)

💼 Business Model + Value Capture (Part 1-C7 view)

  • Revenue mix: API ~70%+ (developers + enterprise) / consumer Claude.ai Pro ~20% / Enterprise ~10%+
  • Gross margin: Strongly inferred loss-making (similar to OPENAI, high training + inference cost), but API GM likely turning positive after reported 80x growth
  • Pricing power: Pricing parity with OpenAI (Claude.ai Pro $20/month; API at GPT-4 levels) — medium
  • Substitution difficulty: Medium — developers using Claude for coding (Claude Code, Cursor's default) creates habit stickiness

🏰 Moat Assessment (Part 3-C2 Buffett 5-step lens)

Type Strength (0-1) Source
Brand 0.7 "Claude" has extremely high developer mindshare; weaker consumer awareness than ChatGPT
Switching cost 0.5 Anthropic Computer Use / Claude Code system integration provides some lock-in
Network effects 0.5 Developer feedback → model quality; Cursor / Claude Code tooling ecosystem
Scale + cost 0.5 Training capex still high, but AWS Trainium + multi-investor backing gives long runway

Composite: 2.2 / 4 → brand + developer niche + ample capital / compute support, but business model still evolving

⚠️ Key Reverse-Trigger Conditions (Part 3-C5 anti-thesis view)

  • Trigger 1 (valuation): Next round down-round (valuation below $380B) — private AI valuation reset signal
  • Trigger 2 (quarterly): API revenue growth stagnates (vs. current 80x growth path)
  • Trigger 3 (annual): Claude frontier quality persistently lags GPT-5 / Gemini — business model foundation shakes
  • Trigger 4 (event): AMZN or GOOGL investment strategy shifts — Anthropic capital sourcing tightens
  • Trigger 5 (low probability): Major enterprise customers (Cursor / Glean / Harvey) switch to OpenAI — Anthropic API revenue hit

🔗 Teaching Cross-References

📚 Official Announcements + Primary Sources (look it up yourself)

Indirect exposure paths: Retail investors gain indirect exposure to Anthropic via AMZN or GOOGL

Teaching use · numbers change — learn the method, not the specific figures · investment decisions are based on your own research + risk tolerance