An AI safety and research company that builds and sells access to Claude, a family of large language model-based AI assistants.
Named as a constituent of Morningstar PitchBook Unicorn 20 Index, Morningstar PitchBook Unicorn 30, Morningstar PitchBook Global Unicorn, Morningstar PitchBook US Unicorn 100, Morningstar PitchBook Global Artificial Intelligence Unicorn, Morningstar PitchBook US Modern Market 100, Hiive50 Index and Forge Accuidity Private Market Index — a fact about those indexes, not an endorsement.
Holding an offer for Anthropic? Put its numbers next to these. Check my offer →
PrePost's proprietary index, combining capital-raising momentum, market-pricing confidence, and investor strength into one comparable score from 0 to 100. It lets you compare companies on those three measures; it is not a rating of the investment. How it works
SEC-filed mutual fund marks (Growth Fund of America, BlackRock) independently confirm the 589.01 Forge Price at the Series H issue price.
Low: LinkedIn-derived estimate via JobsByCulture, mid-2026. High: Tracxn, Aug 2026.
As of 2026-09-13 · live via greenhouse's public job board · ≈14.4% of the 2,300–6,000 estimated team
+143 more roles across 12 other teams.
See all open roles at Anthropic →Anthropic builds the Claude family of large language models, trained with human and AI feedback, including a Constitutional AI method in which the model critiques and revises its own outputs against a written list of principles. Releases since October 2024 include computer use, the open-sourced Model Context Protocol and the Claude Code coding agent.
| Date | Type | What |
|---|---|---|
| Jul 2026 | Milestone | Anthropic's post records Fable 5 access restored on July 1; Opus 5 released July 24 |
| Jun 2026 | Milestone | Fable 5 released June 9, with Anthropic's post recording access suspended on June 12; Sonnet 5 released June 30 |
| May 18, 2026 | Acquisition | Stainless |
| Apr 2026 | Acquisition | Coefficient Bio |
| Feb 25, 2026 | Acquisition | Vercept |
| Dec 2, 2025 | Acquisition | Bun |
| Aug 2025 | Milestone | Hired the co-founders and most of the team of Humanloop; a spokesperson said it did not acquire Humanloop's assets or intellectual property |
| May 22, 2025 | Milestone | Claude Code became generally available alongside Claude Opus 4 and Sonnet 4 |
| Feb 24, 2025 | Milestone | Claude Code, a command-line coding agent, entered research preview |
| Nov 25, 2024 | Milestone | Model Context Protocol open-sourced, a standard for connecting assistants to data sources and tools |
| Oct 22, 2024 | Milestone | Computer use released in public beta: the model operates a computer by looking at the screen, moving a cursor, clicking and typing |
| Mar 20, 2024 | Patent filed | US 12,387,036 Multimodal agent for efficient image-text interface automation |
| Mar 20, 2024 | Patent filed | US 12,430,150 Runtime architecture for interfacing with agents to automate multimodal interface workflows |
| Mar 20, 2024 | Patent filed | US 12,437,238 Generation of agentic trajectories for training artificial intelligence agents to automate multimodal interface task workflows |
| Mar 4, 2024 | Milestone | Claude 3 family released, after Claude 2 on July 11, 2023; Claude 3.5 Sonnet followed in June 2024 |
| 2024 | Paper | Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training |
| 2024 | Paper | Alignment faking in large language models |
| Mar 14, 2023 | Milestone | Claude and its API announced; Anthropic's launch post describes Claude as based on its feedback-training research |
| Dec 2022 | Milestone | Constitutional AI paper: a written list of principles is the only human oversight of harmlessness; the model critiques and revises its own outputs |
| Apr 2022 | Milestone | Company paper applies preference modelling and reinforcement learning from human feedback to make an assistant helpful and harmless |
| 2022 | Paper | Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback |
| 2022 | Paper | Constitutional AI: Harmlessness from AI Feedback |
| Jul 12, 2021 | Patent filed | US 11,763,082 Accelerating inference of transformer-based models |
| 2021 | Milestone | Started by Dario Amodei, previously vice president of research at OpenAI, with his sister Daniela Amodei, as a public benefit corporation |
| 2021 | Paper | A General Language Assistant as a Laboratory for Alignment |
| Nov 14, 2019 | Patent filed | US 11,443,235 Identifying optimal weights to improve prediction accuracy in machine learning techniques |
Patents. All five listed patents reached Anthropic by assignment: three covering a multimodal agent that automates software interfaces, its runtime and its training came from Adept AI Labs on April 9, 2025, and two older patents came from International Business Machines Corp. on May 22, 2025.
Anthropic was started in 2021 by Dario Amodei, previously vice president of research at OpenAI, with his sister Daniela Amodei, as a public benefit corporation (TechCrunch, May 28, 2021). Its work centres on large language models trained with human and AI feedback. One company paper applies preference modelling and reinforcement learning from human feedback to make an assistant helpful and harmless (Bai et al., April 2022, arXiv 2204.05862). A second, Constitutional AI (Bai et al., December 2022, arXiv 2212.08073), uses a written list of principles as the only human oversight of harmlessness: the model critiques and revises its own outputs, and a preference model is trained on AI-generated comparisons. Anthropic's March 2023 launch post describes Claude as based on this research.
Claude and its API were announced on March 14, 2023, followed by Claude 2 (July 11, 2023), the Claude 3 family (March 4, 2024) and Claude 3.5 Sonnet (June 2024). On October 22, 2024 Anthropic released 'computer use' in public beta: the model operates a computer by looking at the screen, moving a cursor, clicking and typing. On November 25, 2024 it open-sourced the Model Context Protocol, a standard for connecting assistants to data sources and tools. Claude Code, a command-line coding agent, entered research preview on February 24, 2025 and became generally available on May 22, 2025 alongside Claude Opus 4 and Sonnet 4. Anthropic says Claude Code reached $1 billion in run-rate revenue in November 2025. Later releases include Opus 4.5 (November 24, 2025), Sonnet 4.6 (February 17, 2026), Fable 5 (June 9, 2026; Anthropic's post records access suspended on June 12 and restored on July 1), Sonnet 5 (June 30, 2026) and Opus 5 (July 24, 2026).
All five patents listed here reached Anthropic by assignment. Three US grants with a March 20, 2024 priority date (US 12,387,036; US 12,430,150; US 12,437,238), granted between August 12 and October 7, 2025, cover a multimodal agent that reads screen images and text to automate software interfaces, a runtime for such agents, and generation of 'agentic trajectories' to train them. Google Patents' assignment entries show the inventors assigned each application to Adept AI Labs Inc. on March 17, 2025, and Adept assigned them to Anthropic on April 9, 2025. Two older patents, US 11,443,235 and US 11,763,082, were filed by International Business Machines Corp. and assigned to Anthropic on May 22, 2025. The same records show Anthropic granted a security interest in both to Morgan Stanley Senior Funding, Inc. on July 8, 2025.
In August 2025 Anthropic hired the co-founders and most of the team of Humanloop; a spokesperson told TechCrunch it did not acquire Humanloop's assets or intellectual property. It then acquired the JavaScript runtime Bun (December 2, 2025; Bun remains open source under the MIT licence), the computer-use startup Vercept (February 25, 2026; its product was scheduled to close on March 25), the biotech AI startup Coefficient Bio (reported in April 2026 by The Information at about $400 million in stock, a figure Anthropic has not confirmed) and the SDK-generation company Stainless (May 18, 2026). Anthropic disclosed no price for any of these.
Profile as of Sep 19, 2026.
| Round | Date | Valuation | Lead |
|---|---|---|---|
| Series H | 2026-05-28 | $965.0B+154% | Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital (co-led by Capital Group, Coatue, D1 Capital Partners, GIC, ICONIQ, and XN) |
| Series G (close) | 2026-02-12 | $380.0B+2% | GIC and Coatue (co-led by D.E. Shaw Ventures, Dragoneer, Founders Fund, ICONIQ, and MGX) |
| Series G | 2026-01-26 | $373.8B+104% | GIC and Coatue |
| Series F | 2025-09-02 | $183.0B+198% | ICONIQ (co-led with Fidelity Management & Research Company and Lightspeed Venture Partners) |
| Series E (ext.) | 2025-03-03 | $61.5B+50% | Lightspeed Venture Partners |
| Series E | 2024-11-22 | $41.1B+123% | Amazon (sole investor, additional $4B) |
| Series D-2/3 | 2024-03-27 | $18.4B+26% | Amazon (sole investor, completing its $4B commitment) |
| Series D-1 | 2023-10-27 | $14.6B+220% | Google (sole investor via convertible note) |
| Series C | 2023-05-23 | $4.5B+13% | Spark Capital |
| Series B | 2023-02-03 | $4.0B+545% | Alameda Research / Sam Bankman-Fried (original Apr 2022 close); Google added a ~$300-400M top-up in Feb 2023, not described as a new lead |
| Series A | 2021-05-28 | $623M | Jaan Tallinn |
+10 more confirmed lead investors.
No upcoming rounds currently reported.
These are on the public record as giving exposure to Anthropic. Buying one means owning a stake in the vehicle, not in Anthropic, and the vehicle takes fees along the way.
| Vehicle | How you get exposure | Share of the vehicle | Who can buy |
|---|---|---|---|
| Destiny Tech100 Inc. Listed closed-end fund | Through a wrapper via Magnitude ANC III, LLC | 14.42% of the fund | Anyone with a brokerage account that can trade US-listed shares. |
| BlackRock Science and Technology Trust Listed closed-end fund | Held directly | 5.13% of the fund | Anyone with a brokerage account that can trade US-listed shares. |
| ARK Venture Fund Interval fund | Held directly | 4.19% of the fund | US investors, through brokers that carry the fund. Not exchange-traded. |
| BlackRock Science and Technology Term Trust Listed closed-end fund | Held directly | 4.07% of the fund | Anyone with a brokerage account that can trade US-listed shares. |
| T. Rowe Price Science and Technology Fund Open-end mutual fund | Held directly | 1.32% of the fund | US investors, through a broker or directly from the manager. |
| Stableton Morningstar PitchBook Unicorn 20 Certificate or fund, by country | Through the index it tracks one of 19 constituents of Morningstar PitchBook Unicorn 20 Index | — | Not published |
On the public record means the vehicle exists in a regulatory filing or a public product document. It does not mean PrePost has vetted it, or checked that it suits you. What each of these really is, and what comes off your return, is in the instruments primer.
Ordered by how much of the vehicle this company is, never by payment. How we stay independent.
| Status | S-1 Filed |
|---|---|
| Banks | Morgan Stanley, Goldman Sachs (JPMorgan also reportedly involved) |
| Key hire | Krishna Rao -- former Airbnb Global Head of Corporate & Business Development who played a key role in raising over $10B in capital including Airbnb's 2020 IPO -- hired as Anthropic's first CFO, announced May 21, 2024 |
| Polymarket | ~50/50 odds by Oct 29, 2026 (as of Sep 13, 2026 · market) |
| Kalshi | ~50/50 odds by Nov 6, 2026 (as of Sep 13, 2026 · market) |
| Target val. | $2.0T |
| Timing | Oct-Dec 2026 (not firm) |
| Listing | Nasdaq |
| Notes | No share count, price range, or firm date has been set. |
| Source | anthropic.com/news/confidential-draft-s1-sec, Jun 1 2026; CNBC; Bloomberg, Jun 3 2026; Anthropic newsroom, May 21 2024 (Krishna Rao hire) |
Polymarket/Kalshi rows are live market-implied odds, not a PrePost prediction; everything else is a cited news signal. PrePost doesn't publish its own IPO forecast.