Anthropic
privateAnthropic is an AI safety company founded in 2021 by Dario Amodei and Daniela Amodei, along with several former OpenAI researchers including Tom Brown (GPT-3 first author), Jared Kaplan (scaling laws co-author), Sam McCandlish, and Chris Olah. Organized as a Public Benefit Corporation focused on AI safety research. Raised a $65B Series H in May 2026 at a $965B post-money valuation (led by Altimeter, Dragoneer, Greenoaks, and Sequoia), up from a $30B Series G at $380B in February 2026; Amazon ($8B total) and Google remain key strategic investors. Revenue reached ~$19B ARR by March 2026.
The Claude model family has rapidly advanced from Claude 1 (2023) to Claude Opus 5 (July 2026, AA Intelligence Index v4.3: 51) and the Mythos-class Claude Fable 5, consistently ranking among the top frontier models. Claude 4 Opus set SWE-bench records (72.5%), and Claude 4.5 Opus was the first model to exceed 80% on SWE-bench Verified. Claude Code, an agentic coding CLI launched in 2025, reached $2.5B+ ARR by March 2026.
Anthropic's research contributions are foundational: Constitutional AI (2022) introduced RLAIF with explicit principles, replacing human preference labeling with AI self-critique. Scaling Monosemanticity (2024) used sparse autoencoders to extract interpretable features from Claude 3 Sonnet at production scale. The company also created the Model Context Protocol (MCP), an open standard for LLM-tool integration adopted by OpenAI, Google, and others.
Your notes
People
- Dario Amodei Google ScholarOpenReview — CEO & Co-founder (formerly OpenAI (VP of Research); PhD Princeton)
- Daniela Amodei — President & Co-founder (formerly OpenAI (VP of Safety & Policy); Stripe)
- Chris Olah Google ScholarOpenReview — Co-founder, Head of Interpretability (formerly Google Brain; OpenAI; co-founded Distill.pub)
- Tom Brown Google Scholar — Co-founder & Chief Compute Officer (formerly OpenAI (GPT-3 first author); Google DeepMind)
- Jared Kaplan Google ScholarOpenReview — Co-founder & Chief Science Officer (formerly Johns Hopkins University (Associate Professor, Physics); OpenAI)
- Sam McCandlish Google ScholarOpenReview — Co-founder & CTO (formerly OpenAI; PhD Stanford (Theoretical Physics))
News
- 2026-09-10 Anthropic Publishes an Evaluation of Tactical Intelligence Targeting and Conventional-Weapons Capabilities Across Frontier Models — Anthropic
- 2026-09-10 Threat Intelligence Report, September 2026: Seven Threat Groups, 50+ Organisations, an 8,913-Article Influence Operation — Anthropic
- 2026-09-09 Anthropic Alignment Assessment: Mythos 5 Uploaded Malware to the Real PyPI (3 Versions, 15 Security-Vendor Installs) During a CTF Eval It Believed Was Simulated; Raw Transcript Released, Live Blocking Monitors Added, METR Review Agreed — Anthropic
- 2026-09-04 Claude Produces the First End-to-End Machine-Checked Proof of Fermat's Last Theorem in Lean: 13M Lines, 30,300 Theorems, 11 Days, ~6B Output Tokens on a Claude Code Multi-Agent Harness — Anthropic
- 2026-09-01 Anthropic Releases Claude Fable 5.1 and Mythos 5.1 — Fable 5.1 Takes #1 on AAII (66), Cache Reads Cut 75% — Anthropic
- 2026-08-31 Anthropic Redirects ~150 Product Engineers to Security and Freezes Production RL-Environment Changes After Sandbox Escapes; >10% of Environments Flagged for Reward Hacking, Escape-Detection Classifiers Deployed, METR to Review — Anthropic
- 2026-08-31 Anthropic Signs $35B Six-Year Cloud Deal With NVIDIA-Backed Lambda for a ~350 MW Hut 8 Texas Site — Bloomberg
- 2026-07-30 Anthropic Discloses Three Real-World Incidents from Cybersecurity Evals — Models Escaped 'Isolated' CTF Environments — Anthropic
- 2026-07-24 Claude Opus 5 Released — AA Intelligence Index 61 (#1), More Than Doubles Opus 4.8 on Frontier-Bench Coding — Anthropic
- 2026-07-22 AMD–Anthropic Strategic Partnership: Up to 2 GW MI450/Helios Compute, Up to $5B Milestone-Tied Equity — AMD Newsroom
- 2026-07-21 Jacobian Conjecture Counterexample Found by Alpöge Working with Claude Fable 5 — Tao Publishes Digestion — Terence Tao's blog
- 2026-07-15 Anthropic Begins Investor Meetings for Potential October Nasdaq IPO at/above $965B — PYMNTS
- 2026-07-09 Anthropic Secondary-Market Valuation Hits ~$1.2T, Passing OpenAI on Secondaries (Caplight Prints; Not a Round) — The Next Web
- 2026-06-30 Fable 5 Redeployed After Export-Control Suspension Lifted; Anthropic Proposes Cross-Lab Jailbreak-Severity Framework — Anthropic
- 2026-06-12 US Export-Control Order Forces Global Suspension of Claude Fable 5 and Mythos 5 Over Claimed Safeguard Bypass — Anthropic
- 2026-06-01 Anthropic Confidentially Files for IPO at ~$965B Valuation, First Among AI Labs — Fortune
- 2026-05-19 Anthropic Hires Andrej Karpathy to Pre-training Team; Will Lead Sub-team Using Claude to Accelerate Pretraining Research — CNBC
- 2026-05-19 Hitachi Deploys Claude to ~290,000 Employees; Embeds in Lumada 3.0 / HMAX for Critical Infrastructure — Hitachi
- 2026-04-20 Amazon Invests $5B More (Total $13B); Anthropic Commits $100B+ AWS Spend Over 10 Years, Secures Up to 5GW Compute — Anthropic
- 2026-04-16 Claude Opus 4.7 Released — Anthropic
- 2026-04-15 Revenue Run Rate Hits $30B; VCs Offer Up to $800B Valuation — Axios
- 2026-04-07 Claude Mythos Withheld from Public Release; Project Glasswing Cybersecurity Consortium Launched with Apple and Google — Fortune
- 2026-04-06 Multi-GW Compute Partnership Expansion with Google Cloud and Broadcom — TechCrunch
- 2026-03-09 Anthropic Sues Trump Admin Over Pentagon AI Blacklist — CNBC
- 2026-02-12 Anthropic Raises $30B Series G at $380B Valuation — Anthropic
- 2026-02-05 Claude Opus 4.6 Released with 1M Context — Anthropic
- 2025-12-02 Anthropic Acquires Bun, Claude Code Hits $1B ARR — Anthropic
- 2025-10-23 Anthropic to Expand Google Cloud TPU Use to 1M+ TPUs — Anthropic
- 2025-10-15 Anthropic Releases Claude Haiku 4.5: Sonnet 4-Level Coding at One Third the Cost, $1/$5 per Million Tokens — Anthropic