Anthropic PBC
Company History
Anthropic History: Founding, Growth & Key Milestones
Founded 2021 in San Francisco, United States
When was Anthropic founded?
Anthropic was founded in 2021 in San Francisco, United States by Dario Amodei, Daniela Amodei, Jack Clark, Sam McCandlish, Jared Kaplan, Tom Brown, Chris Olah and Ben Mann, making it 5 years old.
- Founded
- 2021
- Founding place
- San Francisco, United States
- Founders
- Dario Amodei, Daniela Amodei, Jack Clark, Sam McCandlish, Jared Kaplan, Tom Brown, Chris Olah and Ben Mann
- Headquarters today
- San Francisco, California, United States
- Milestones documented
- 19 (2021–2026)
- Current CEO
- Dario Amodei
- Acquisitions
- 5
| Year | Event |
|---|---|
| 2021 | Anthropic founded by former OpenAI staff |
| 2022 | $580 million round and the first Claude held back |
| 2022 | Constitutional AI published |
| 2023 | Claude and Claude 2 released |
| 2023 | Amazon and Google take stakes |
| 2024 | Claude 3 family and Claude 3.5 Sonnet |
| 2024 | Amazon investment reaches $8 billion |
| 2025 | Claude 4 and Claude Code general availability |
| 2025 | $1.5 billion copyright settlement with authors |
| 2025 | Google TPU deal for up to one million chips |
| 2025 | Microsoft and NVIDIA commit up to $15 billion |
| 2025 | Bun acquired as Claude Code passes $1 billion run-rate |
| 2026 | Series G of $30 billion at a $380 billion valuation |
| 2026 | Pentagon dispute over surveillance and weapons limits |
| 2026 | Series H of $65 billion at a $965 billion valuation |
| 2026 | Confidential draft S-1 submitted to the SEC |
| 2026 | Claude Opus 5 and Opus 5.5 released |
| 2026 | IPO prospectus figures reported |
| 2026 | Public Form S-1 filed with the SEC |
- Where was Anthropic founded?
- Anthropic was founded in San Francisco, United States in 2021. wikidata.org
- Who founded Anthropic?
- Anthropic was founded by Dario Amodei, Daniela Amodei, Jack Clark, Sam McCandlish, Jared Kaplan, Tom Brown, Chris Olah and Ben Mann.
- When did Anthropic start?
- Anthropic was founded in 2021.
- How old is Anthropic?
- Anthropic is 5 years old as of 2026.
Founding Story
Dario Amodei was OpenAI's Vice President of Research and led the teams behind GPT-2 and GPT-3. As OpenAI turned from a non-profit lab into a capped-profit company partnered with Microsoft, he and his sister Daniela, then OpenAI's Vice President of Safety and Policy, concluded that commercial pressure was outpacing alignment work. They left in late 2020 with five colleagues, including Jared Kaplan, Sam McCandlish, Tom Brown, Chris Olah, Ben Mann and Jack Clark, and incorporated Anthropic on January 26, 2021 as a Delaware public benefit corporation, with an independent Long-Term Benefit Trust holding the right to appoint board members so that safety commitments would not depend on the goodwill of whoever ran the company.
Founders
Dario Amodei
Dario Amodei is co-founder and Chief Executive Officer of Anthropic PBC. He previously served as Vice President of Research at OpenAI, directing the teams that built GPT-2 and GPT-3 and the scaling law research that still shapes how frontier models are planned. He left OpenAI in late 2020 with senior colleagues and incorporated Anthropic in January 2021 as a Delaware public benefit corporation. Under him the company raised a $13 billion Series F at a $183 billion valuation in September 2025, a $30 billion Series G at $380 billion in February 2026 and a $65 billion Series H at $965 billion in May 2026, and confidentially submitted a draft S-1 for a US listing on June 1, 2026. He has also called publicly for pacing frontier development, refused Department of Defense demands in February 2026 to drop limits on mass surveillance and fully autonomous weapons, and in September 2026 agreed to give independent evaluators permanent access to Anthropic models. Dario Amodei was born in 1983 in San Francisco.
Daniela Amodei
Daniela Amodei is co-founder and President of Anthropic PBC, responsible for the commercial and operating side of the company: go-to-market, finance, people, investor relations and public benefit compliance. She previously led safety and policy at OpenAI as Vice President and held operations roles at Stripe. At Anthropic she has overseen growth to more than 2,500 employees as of early 2026 and the distribution strategy that put Claude on Amazon Web Services, Google Cloud and Microsoft Azure. Daniela Amodei was born in 1987 in San Francisco. Education: University of California, Santa Cruz.
Jack Clark
Jack Clark is a co-founder of Anthropic and its head of policy. Since March 2026 he has also led the Anthropic Institute, a think tank the company launched to study the effects of AI. He was previously Policy Director at OpenAI and a technology journalist at Bloomberg and The Register, has co-chaired the Stanford AI Index, and represents Anthropic in work with government safety bodies and in external evaluations of frontier models. (Exact birth year remains publicly undisclosed).
Sam McCandlish
Sam McCandlish is a co-founder of Anthropic and a researcher known for work on the scaling laws of neural networks. At OpenAI he co-authored research showing a predictable relationship between training compute and model performance. At Anthropic that work informs how the company sizes and plans training runs for the Claude family. (Exact birth year remains publicly undisclosed).
Jared Kaplan
Jared Kaplan is a co-founder and Chief Science Officer of Anthropic. He was a lead author of 'Scaling Laws for Neural Language Models' in 2020, which showed that model loss falls predictably as compute, data and parameter count increase. A former Johns Hopkins physics professor, he collaborated with OpenAI researchers on that work before co-founding Anthropic in January 2021, and now directs the science behind Claude training and alignment. Education: Harvard University. (Exact birth year remains publicly undisclosed).
Tom Brown
Tom Brown is a co-founder of Anthropic and was the lead author of the 2020 GPT-3 paper 'Language Models are Few-Shot Learners' at OpenAI, which showed that large language models can handle many tasks from a few examples without task-specific fine-tuning. He left OpenAI in 2021 to help found Anthropic, where he built much of the early training infrastructure behind Claude. (Exact birth year remains publicly undisclosed).
Chris Olah
Chris Olah is a co-founder of Anthropic and leads its interpretability research, which works out which internal features and circuits drive a model's behaviour. His team's dictionary learning work in 2024 extracted millions of interpretable features from Claude 3 Sonnet, including one tied to the Golden Gate Bridge. He previously worked on interpretability at Google Brain and OpenAI. Christopher Olah was born in 1991 or 1992 and is Canadian. He is known for research on neural network interpretability, in particular mechanistic interpretability.
Ben Mann
Ben Mann is a co-founder of Anthropic and one of the authors of OpenAI's 2020 GPT-3 paper 'Language Models are Few-Shot Learners'. At Anthropic he has worked on training and aligning the Claude models and on the practices that govern how they are released. (Exact birth year remains publicly undisclosed).
Key Milestones
Anthropic founded by former OpenAI staff
Dario Amodei, Daniela Amodei and five other former OpenAI employees incorporated Anthropic on January 26, 2021, and raised a $124 million Series A in May 2021 led by Jaan Tallinn.
Source: en.wikipedia.org$580 million round and the first Claude held back
Anthropic raised $580 million in April 2022, including $500 million from FTX. It finished training the first version of Claude that summer but did not release it, citing further safety testing.
Source: en.wikipedia.orgConstitutional AI published
Anthropic published Constitutional AI, a training method in which the model critiques and revises its own answers against a written set of principles instead of relying only on human ratings.
Source: en.wikipedia.orgClaude and Claude 2 released
Claude and Claude Instant were released in March 2023 to approved users, and Claude 2 opened to the public in July 2023.
Source: en.wikipedia.orgAmazon and Google take stakes
Amazon announced a strategic collaboration in September 2023 with an initial $1.25 billion investment and AWS as primary cloud provider. Google invested $500 million in October 2023 and committed $1.5 billion more over time.
Source: en.wikipedia.orgClaude 3 family and Claude 3.5 Sonnet
Anthropic released Claude 3 Opus, Sonnet and Haiku on March 4, 2024, then Claude 3.5 Sonnet in June 2024, which became the default model inside many AI coding tools.
Source: anthropic.comAmazon investment reaches $8 billion
Amazon invested a further $4 billion in November 2024, taking its total to $8 billion. Anthropic named AWS its primary training partner and agreed to train on AWS Trainium chips.
Source: aboutamazon.comClaude 4 and Claude Code general availability
Anthropic released Claude Opus 4 and Claude Sonnet 4 in May 2025 at its first developer conference, and made the Claude Code command line agent generally available the same month.
Source: anthropic.com$1.5 billion copyright settlement with authors
Anthropic agreed in September 2025 to pay $1.5 billion, about $3,000 per book, to settle the Bartz class action over pirated books used in training. A judge granted final approval on July 20, 2026.
Source: techcrunch.comGoogle TPU deal for up to one million chips
In October 2025 Anthropic agreed to expand its use of Google Cloud to as many as one million TPUs, adding well over a gigawatt of capacity in 2026.
Source: anthropic.comMicrosoft and NVIDIA commit up to $15 billion
In November 2025 NVIDIA committed up to $10 billion and Microsoft up to $5 billion, while Anthropic agreed to buy about $30 billion of Azure compute and put Claude on Azure Foundry.
Source: anthropic.comBun acquired as Claude Code passes $1 billion run-rate
Anthropic acquired Bun, the JavaScript and TypeScript runtime behind Claude Code and the Claude Agent SDK, in December 2025, and said Claude Code had reached $1 billion of run-rate revenue.
Source: anthropic.comSeries G of $30 billion at a $380 billion valuation
Anthropic raised $30 billion in February 2026 led by GIC and Coatue at a $380 billion post-money valuation, and reported run-rate revenue of $14 billion and more than 500 customers spending over $1 million a year.
Source: anthropic.comPentagon dispute over surveillance and weapons limits
Anthropic refused Department of Defense demands in February 2026 to drop restrictions on mass domestic surveillance and fully autonomous weapons. The Pentagon designated it a supply chain risk in March 2026 and a federal judge then granted Anthropic a preliminary injunction.
Source: en.wikipedia.orgSeries H of $65 billion at a $965 billion valuation
Anthropic raised $65 billion in May 2026 led by Altimeter, Dragoneer, Greenoaks and Sequoia at a $965 billion post-money valuation, and said run-rate revenue had crossed $47 billion that month.
Source: anthropic.comConfidential draft S-1 submitted to the SEC
On June 1, 2026 Anthropic confidentially submitted a draft registration statement on Form S-1 for a proposed initial public offering of common stock.
Source: anthropic.comClaude Opus 5 and Opus 5.5 released
Claude Opus 5 replaced Opus 4.8 as the flagship in July 2026, and Opus 5.5 followed on September 22, 2026 at $4 per million input tokens and $20 per million output tokens, about 40 percent cheaper to run than Opus 5.
Source: anthropic.comIPO prospectus figures reported
Reporting on the draft prospectus in late September 2026 disclosed 2025 revenue of $4.59 billion, an $8.06 billion operating loss, a $41.97 billion net loss and about $518 billion of long-term compute commitments.
Source: techcrunch.comPublic Form S-1 filed with the SEC
On October 26, 2026, Anthropic publicly filed its Form S-1 for a proposed initial public offering. The filing reports $16.286 billion of revenue for the six months ended June 30, 2026, an unaudited interim period.
Source: sec.govMajor Acquisitions
Humanloop
2025Team acquisition covering prompt management, model evaluation and observability for enterprise customers
Bun
2025JavaScript and TypeScript runtime and toolchain that Claude Code and the Claude Agent SDK are built on
Vercept
2026Computer use research, so Claude can operate live desktop and web applications
Stainless
2026Compiler that turns API specifications into SDKs, command line tools and MCP servers
Mendral
2026Team working on automated continuous integration fixes, flaky test triage and dependency review
Keep researching Anthropic
Anthropic Company History FAQ
What happened in Anthropic's first two years, from 2021 to 2022?
Anthropic was incorporated on January 26, 2021, raised $124 million in May 2021 led by Jaan Tallinn and $580 million in April 2022, published its Constitutional AI method, and finished training the first version of Claude in the summer of 2022 without releasing it.
When did Anthropic launch Claude?
Claude and Claude Instant were released in March 2023 to approved users, after months of internal safety testing, and Claude 2 opened to the public in July 2023.
How did the Amazon and Google investments develop?
Amazon announced a collaboration in September 2023 with an initial $1.25 billion, extended it by $2.75 billion in March 2024 and by a further $4 billion in November 2024 to reach $8 billion. Google invested $500 million in October 2023, committed $1.5 billion more over time and added $1 billion in March 2025.
When did Anthropic release the Claude 3 family?
Anthropic released Claude 3 Opus, Claude 3 Sonnet and Claude 3 Haiku on March 4, 2024, and followed with Claude 3.5 Sonnet in June 2024.
What changed for Anthropic in 2025?
Claude 4 arrived in May 2025 alongside general availability for Claude Code, the authors' copyright case settled for $1.5 billion in September, the Series F raised $13 billion at a $183 billion valuation, Google agreed to supply up to one million TPUs in October, NVIDIA and Microsoft committed up to $15 billion in November, and Anthropic acquired Bun in December.
What were the main events of 2026?
A $30 billion Series G at $380 billion in February, the Pentagon dispute and a court injunction in the spring, the Mythos model announcement in April, a $65 billion Series H at $965 billion in May, a confidential draft S-1 on June 1, Claude Opus 5 in July and Opus 5.5 on September 22.
How does the Long-Term Benefit Trust fit into Anthropic's history?
The trust was created with the company in 2021 and staffed with trustees who hold no financial interest in it. It has the right to appoint and replace part of the board, which is how Anthropic ties its public benefit purpose to its governance.
Is Anthropic still private?
Yes. It confidentially submitted a draft S-1 on June 1, 2026, and reporting in late September 2026 pointed to a listing after the November midterm elections, so as of September 30, 2026 it remains a private public benefit corporation.
Why was Anthropic founded?
Anthropic was founded in 2021 by a group of senior OpenAI researchers who left because they believed OpenAI was prioritizing commercialization and rapid growth over AI safety and alignment.
Related Companies
Compare Anthropic PBC History With Competitors
Related Strategic History
Compare founding milestones and corporate evolution
Explore Anthropic
Core profile pages, founders and related research hubs for this company.