ChronlinesOpen Chronlines

Anthropic Timeline

From the Amodeis' path through Baidu, Google Brain and OpenAI to the 2021 founding, Claude, the Pentagon fight and the October 2026 S-1, with a reported IPO raise of $100 billion or more at $1.5 to $2 trillion.

106 events · 2014 – 2026

Open interactive timeline Anthropic: a timeline on Chronlines

Key events

  1. Dario Amodei joins OpenAI

    Amodei joins the young nonprofit lab as an early employee focused on AI safety, later rising to vice president of research.

  2. "Concrete Problems in AI Safety" published

    Amodei and Chris Olah are the lead authors of a paper that turns AI safety into specific engineering research problems. Olah later becomes an Anthropic co-founder.

  3. Daniela Amodei joins OpenAI

    After five years building recruiting at Stripe, Daniela Amodei joins OpenAI. She rises to vice president of safety and policy in 2020.

  4. GPT-3 paper lists future Anthropic founders

    Tom Brown is first author and Dario Amodei last author on the GPT-3 paper. Benjamin Mann, Jared Kaplan, Jack Clark and Sam McCandlish also appear among the authors.

  5. Dario Amodei leaves OpenAI

    OpenAI announces the departure of its vice president of research. Daniela Amodei and other senior staff also leave around this time, citing differences over direction. Amodei has since also cited a loss of trust in OpenAI's leadership.

  6. Anthropic incorporated

    Anthropic is incorporated as a Delaware public benefit corporation in San Francisco. Its seven co-founders are Dario Amodei, Daniela Amodei, Tom Brown, Jack Clark, Jared Kaplan, Sam McCandlish and Christopher Olah.

  7. $124 million Series A

    Skype co-founder Jaan Tallinn leads the first round, which funds research into steerable, interpretable and robust large-scale AI systems.

  8. $580 million Series B led by Sam Bankman-Fried

    The round is led by the FTX founder and values Anthropic at about $4 billion, reportedly leaving FTX with a stake of about 8%.

  9. Constitutional AI paper

    Anthropic describes training a model to critique and revise its own answers against written principles. The method becomes a signature of Claude.

  10. Google Cloud partnership and early Google stake

    Anthropic names Google Cloud its preferred cloud provider. The Financial Times reports Google paid about $300 million in late 2022 for a stake of about 10%.

  11. Claude launches

    Anthropic releases Claude, its first commercial model, through a chat interface and API, with early partners including Notion, Quora and DuckDuckGo.

  12. Claude 2 and public claude.ai

    Claude 2 arrives with a 100,000-token context window, and claude.ai opens to the public in the US and UK.

  13. Long-Term Benefit Trust unveiled

    Anthropic discloses an independent trust of financially disinterested trustees that can elect and remove a growing share of its board. The trust becomes central to the governance described in the IPO filing.

  14. Responsible Scaling Policy

    Anthropic commits to AI Safety Levels that tie model development and deployment to safeguards.

  15. Amazon commits up to $4 billion

    Amazon takes a minority stake and becomes Anthropic's primary cloud provider, with models trained on its Trainium and Inferentia chips.

  16. Google commits up to $2 billion

    The Wall Street Journal reports, and Anthropic confirms, that Google will invest $500 million up front and up to $1.5 billion more over time.

  17. Claude 3 family

    The Haiku, Sonnet and Opus tiers debut, and Anthropic says the family sets new industry benchmarks across cognitive tasks.

  18. Claude 3.5 Sonnet

    A mid-tier model outperforms Claude 3 Opus on key evaluations at twice the speed, and Artifacts debuts in claude.ai.

  19. Computer use and upgraded Claude 3.5 Sonnet

    Claude gains an experimental ability to operate a computer, an early step toward the agents that later drive its revenue.

  20. Amazon invests another $4 billion

    Amazon's total reaches $8 billion, and AWS becomes Anthropic's primary cloud and training partner, with joint work on Trainium chips.

  21. Model Context Protocol

    Anthropic releases an open standard for connecting AI assistants to data and tools, which rivals later adopt.

  22. Claude 3.7 Sonnet and Claude Code

    Anthropic previews Claude Code, a command-line coding agent. It becomes the product that most reshapes the company's revenue.

  23. $3.5 billion Series E at $61.5 billion

    Lightspeed Venture Partners leads the round.

  24. Claude 4 and ASL-3 protections

    Opus 4 and Sonnet 4 launch at the first Code with Claude conference, and Claude Code becomes generally available. Anthropic activates ASL-3 safeguards for the first time.

  25. Alsup splits on fair use in Bartz

    Judge William Alsup rules that training on lawfully bought, scanned books is fair use, but that keeping pirated copies in a central library is not. The ruling sets up the settlement.

  26. $200 million Defense Department agreement

    The Pentagon awards a two-year prototype agreement with a $200 million ceiling, the same relationship that collapses in early 2026.

  27. $13 billion Series F at $183 billion

    ICONIQ leads, with Fidelity and Lightspeed co-leading. Run-rate revenue had passed $5 billion in August.

  28. $1.5 billion authors settlement

    Anthropic agrees to pay about $3,000 per book to settle piracy claims, the largest copyright settlement on record. The charge appears as a separate line in the S-1.

  29. Deal for up to one million Google TPUs

    An expansion worth tens of billions of dollars is expected to bring more than a gigawatt of capacity online in 2026.

  30. $50 billion for US data centers

    Anthropic partners with Fluidstack on its own sites in Texas and New York.

  31. Microsoft and Nvidia invest up to $15 billion

    Anthropic commits $30 billion to Azure compute in a deal that puts Claude on all three major clouds.

  32. Bun acquired as Claude Code hits $1 billion

    Claude Code reaches $1 billion in run-rate revenue six months after general availability. Anthropic buys the JavaScript runtime that supports it.

  33. Reports: Wilson Sonsini hired to prepare an IPO

    The Financial Times reports Anthropic has engaged the law firm for a listing that could come as early as 2026, and has held early talks with banks. A spokesperson plays down the speculation.

  34. Claude Cowork launches

    A research preview brings Claude Code-style agents to non-coding office work.

  35. New constitution for Claude

    Anthropic publishes a long revised document, released under a public domain dedication, explaining the values it wants Claude to hold.

  36. Cowork plugins trigger a software selloff

    Software and data stocks fall sharply worldwide as investors react to Anthropic's legal and other Cowork plugins, the first of several market shocks tied to its releases.

  37. $30 billion Series G at $380 billion

    GIC and Coatue lead the round. Run-rate revenue stands at $14 billion, Claude Code at more than $2.5 billion, and more than 500 customers spend over $1 million a year.

  38. Amodei refuses Pentagon terms

    Anthropic says it cannot accept "any lawful use" terms that would drop two limits: no mass domestic surveillance and no fully autonomous weapons.

  39. Trump orders agencies off Claude

    The president directs federal agencies to stop using Anthropic's technology, with a six-month phase-out, and Defense Secretary Pete Hegseth says he will designate the company a supply chain risk.

  40. Formal supply chain risk designation

    The Department of War sends two letters, under separate statutes, designating Anthropic a supply chain risk, a label normally used for foreign adversaries.

  41. Anthropic sues the government

    The company files in federal court in San Francisco and petitions the D.C. Circuit, calling the actions unprecedented and unlawful.

  42. Judge blocks the Pentagon ban

    US District Judge Rita Lin grants a preliminary injunction in San Francisco, restoring the status quo while the case proceeds.

  43. Reports: IPO as soon as October, raising $60 billion or more

    Bloomberg and The Information report early talks with Goldman Sachs, JPMorgan and Morgan Stanley. The first reported size is well under what is discussed by autumn.

  44. Run-rate revenue passes $30 billion

    Announcing multiple gigawatts of Google and Broadcom TPU capacity from 2027, Anthropic says run-rate revenue has more than tripled since year-end and more than 1,000 customers spend over $1 million a year.

  45. Claude Mythos Preview and Project Glasswing

    Anthropic withholds its most capable model from general release because of its cyber abilities, and gives a coalition of 12 launch partners access to find and fix software flaws, backed by up to $100 million in credits.

  46. Trust-appointed directors become a board majority

    The Long-Term Benefit Trust appoints Novartis CEO Vas Narasimhan, giving its appointees control of the board ahead of a listing.

  47. Amazon adds $5 billion and up to $20 billion more

    Anthropic commits more than $100 billion to AWS over ten years and secures up to 5 gigawatts of capacity.

  48. Google commits up to $40 billion

    Google invests $10 billion now and up to $30 billion more if performance targets are met.

  49. Compute deal with SpaceX

    Anthropic takes all of the Colossus 1 data center in Memphis, more than 300 megawatts and 220,000 Nvidia GPUs, and raises Claude usage limits. The data center was built by xAI, now a SpaceX subsidiary.

  50. $65 billion Series H at $965 billion

    Altimeter, Dragoneer, Greenoaks and Sequoia lead the round. Run-rate revenue has passed $47 billion, and the valuation tops OpenAI's, making it the last private mark before the IPO.

  51. Confidential draft S-1 submitted

    Anthropic says it has privately submitted a draft registration statement, giving it the option to go public after SEC review, subject to market conditions. No size or price is given.

  52. Claude Fable 5 and Mythos 5

    Fable 5 is a Mythos-class model made safe for general use, with requests on sensitive topics handed to Opus 4.8. Mythos 5 is restricted to cyber defenders and infrastructure providers.

  53. Export directive forces Fable 5 offline

    Days after launch and within two weeks of the draft S-1, a US directive bars foreign nationals from Fable 5 and Mythos 5. Unable to verify nationality in real time, Anthropic shuts both off for everyone and says it disagrees with the order.

  54. SpaceX sets the IPO record

    SpaceX raises $75 billion at $135 a share, a $1.77 trillion valuation, and closes up about 19%. With its overallotment the raise reaches about $86 billion, the benchmark Anthropic is measured against.

  55. Export controls lifted

    The restrictions end after Anthropic adds a classifier that blocks the jailbreak in question, and Fable 5 returns globally on July 1.

  56. Bartz settlement wins final approval

    Judge Araceli Martínez-Olguín approves the $1.5 billion settlement over objections, removing a major legal overhang before the IPO. Payments run in installments through September 2027.

  57. Run-rate revenue tops $65 billion

    Bloomberg reports Anthropic told investors its annualized rate passed $65 billion at the end of July, seven times the end of 2025 level.

  58. Report: IPO to match or beat SpaceX

    Bloomberg reports Anthropic expects its debut to equal or exceed the largest ever, with a public filing possibly weeks away.

  59. Court rules the designation unlawful

    Judge Lin largely grants Anthropic summary judgment, finding the government's actions were retaliation for protected speech. The fight continues on appeal.

  60. Report: marketing to start in mid-October

    Reuters reports the launch has shifted to mid-October at the earliest, with Morgan Stanley, Goldman Sachs, JPMorgan and Citi working on the deal and a $15 billion credit line being finalized.

  61. Report: IPO slips to November

    The Wall Street Journal reports a November launch, later than many expected, so third-quarter results can be shown first.

  62. D.C. Circuit sides with the Pentagon

    In a 2-1 decision, the appeals court upholds the Section 4713 designation and rejects Anthropic's First and Fifth Amendment claims. Anthropic seeks rehearing by the full court and flags the risk in its S-1.

  63. Report: mid-November at $1.8 to $2 trillion

    Bloomberg reports formal marketing could begin the week of November 9, allowing trading before Thanksgiving. The Wall Street Journal puts the raise at up to $100 billion at about $2 trillion.

  64. Investor day: Q3 revenue of $16.8 billion

    Anthropic tells investors in San Francisco that September quarter revenue rose about 45% from the prior quarter, with an adjusted 11% operating margin. The slowdown from Q2 unsettles some investors.

  65. NYC comptroller urges an IPO pause

    Mark Levine, who oversees more than $325 billion in pension assets, asks Amodei to pause the IPO until safety audits and governance concerns are addressed. Anthropic opens talks but does not agree to delay.

  66. Report: investors talk down to $1.5 trillion

    Reuters reports investors who attended the investor day are discussing a $1.5 trillion valuation, down from about $2 trillion, after disclosures on slowing growth and more than $518 billion in compute commitments.

  67. Anthropic Public Benefit Fund

    Eight days before the filing, Anthropic commits 5% of its shares, subject to stockholder approval, and may add 2% for each 50% rise in value. With founder and employee pledges, it says about 20% of shares go to the public good.

  68. Anthropic files its public S-1

    Anthropic files to list Class A shares on the Nasdaq Global Select Market as "ANT," with Goldman Sachs and Morgan Stanley leading 22 banks. Share count, price range and raise are left blank. The filing has 94 pages of risk factors.

  69. S-1 reveals about $417 billion in compute obligations

    CNBC reports commitments including Amazon, Google, Microsoft, SpaceX and AMD. Net revenue retention is about 900%, and about 48% of revenue flows through cloud marketplaces.

  70. S-1 reveals founder and trust control

    A company owned by the seven co-founders gets 50.1% of the vote on most matters, while the Long-Term Benefit Trust elects a board majority. The filing says the structure has no direct public company precedent.

  71. S-1 reveals soaring revenue and large losses

    Revenue was $41 million in 2023, $386 million in 2024, $4.6 billion in 2025 and $16.3 billion in the first half of 2026. The first half net loss of $80.7 billion was mostly noncash, and Q2 brought a first operating profit of about $1 billion.

  72. Expected: IPO raises about $100 billion at $1.5 to $2 trillion

    Press estimates, not company guidance: a raise of $100 billion or more at a valuation of $1.5 trillion to $2 trillion, up from $965 billion in May. That would be the largest IPO ever, above SpaceX's $75 billion base deal and about $86 billion total. November 19 or 20 are the reported listing dates. Terms are not set.

  73. Planned: IPO roadshow begins

    IFR reports the roadshow will launch November 9, with pricing the following week. The date comes from an unnamed person familiar with the process, and Anthropic has not confirmed it.

33 more in the interactive timeline.

Open interactive timeline