Anthropic owns Claude AI. Anthropic is a private American company, registered as a public benefit corporation, that builds and sells the Claude models and the Claude app. So the short answer to who owns Claude AI is one company. The longer answer is more interesting, because Anthropic itself is owned by its founders and a long list of investors, two of the biggest tech companies on earth among them. An independent trust, meanwhile, elects much of its board.
That mix matters if you use Claude for work. Ownership decides who profits from the product and who can overrule whom when safety and revenue pull in different directions. As of October 2026 the picture is also about to change, since Anthropic is preparing to go public.
Who created Claude AI: Anthropic and its founders
Claude was created by Anthropic, a company founded in 2021 by a group of former OpenAI researchers. The best-known co-founders are the siblings Dario Amodei, who runs the company as chief executive, and Daniela Amodei, its president. The first Claude model reached the public in March 2023.
The founders left OpenAI with a specific thesis: frontier AI would arrive fast, and the lab building it should treat safety research as part of the product. That thesis shaped the legal structure they chose, which is why the ownership question has three layers.
- Anthropic, PBC owns Claude outright, from the model weights to the apps.
- Shareholders own Anthropic: founders and staff, plus outside investors.
- The Long-Term Benefit Trust holds a special class of shares that lets it elect board members.
A public benefit corporation, or PBC, is a for-profit company whose charter names a public purpose that directors must weigh alongside shareholder returns. Anthropic's stated purpose is the responsible development of advanced AI for the long-term benefit of humanity. It is a lofty sentence to put in a Delaware filing, and Delaware has seen stranger ones.
Who owns Anthropic: the investors behind Claude
Anthropic has raised money in a long series of rounds, and each round sold a slice of the company. Nobody outside the company can give an exact percentage for each owner yet, because Anthropic has never published its cap table, meaning the list of who holds which shares. The numbers below come from company announcements and reporting on court and securities filings.
The most recent priced round was a $65 billion Series H, announced on May 28, 2026, at a $965 billion post-money valuation. Anthropic's Series H announcement names Altimeter Capital and Sequoia Capital among the four lead investors, with dozens of funds joining. The same announcement says run-rate revenue crossed $47 billion in May.
Most of that money goes straight back out the door as compute. Training a frontier model takes data centres measured in gigawatts, and the Series H announcement pairs the funding with agreements for up to five gigawatts of new capacity from Amazon, plus five gigawatts of TPU capacity from Google and Broadcom. Investors fund the chips, and the chips train Claude. Everyone in the industry calls this loop a flywheel, which sounds better than a very large bill.
| Owner | What is publicly known |
|---|---|
| Amazon | About $8 billion invested from 2023, plus $5 billion more in 2026. Held as convertible notes and nonvoting preferred stock. Estimates put the stake in the mid-to-high teens. |
| Roughly 14% per court documents, contractually capped at 15%. Committed up to $40 billion more in April 2026, with $10 billion up front. | |
| Venture and growth funds | Lead investors across the rounds include Sequoia and Altimeter, with sovereign funds such as GIC among others. |
| Founders and employees | Undisclosed. Expected to appear in the public IPO filing. |
The Amazon figures come from Fortune's analysis of the Amazon and Google stakes, which notes that an April Amazon filing valued its position at about $74 billion. Microsoft and Nvidia also announced investment commitments in late 2025, which is how Claude ended up on all three of the largest cloud platforms.
Notice the word nonvoting. Amazon has put more than $13 billion into Anthropic, and its preferred shares still carry no vote. That is a deliberate design choice. It is also the corporate equivalent of buying a very expensive car and agreeing to sit in the back.
Claude AI who owns versus who controls: the Long-Term Benefit Trust
Owning shares and controlling a company are two different things, and Anthropic separates them on purpose. The tool for that is the Long-Term Benefit Trust, a body of five financially disinterested trustees who hold no equity stake in Anthropic.
According to Anthropic's explanation of the Long-Term Benefit Trust, the company amended its charter in 2023 to create Class T stock, held only by the trust. Class T shares give the trust the power to elect and remove a growing number of directors, phasing in over time and funding milestones, with the stated goal of a board majority within four years. Trustees serve one-year terms, and the existing trustees pick their successors.
The trust has turned over since 2023. Neil Buddy Shah chairs it, national security expert Richard Fontaine joined in 2025, and former Federal Reserve chair Ben Bernanke joined in July 2026.
There is a safety valve in the other direction. A large enough supermajority of stockholders can change the trust and its powers without the trustees' consent, and the threshold rises as the trust's power grows. So the honest summary reads like this: investors own most of the economic value while the trust shapes the board. A big enough coalition of shareholders keeps the final say. If you have followed the wider debate in why AI transformation is a governance problem, this is that argument written into a corporate charter.
The Anthropic IPO: what changes when Claude's owner goes public
Anthropic confidentially submitted a draft registration statement to the US Securities and Exchange Commission in June 2026. A registration statement, usually called an S-1, is the document a company must publish before it can sell shares to the public. In late October, reports said the public version could appear within days and the listing could follow in November, with talk of raising up to $100 billion at a valuation near $2 trillion. As of October 26, 2026, the public filing had not yet appeared, so treat every IPO figure as reported and unconfirmed.
When the S-1 goes public, the ownership question gets a precise answer for the first time. The filing has to list every holder of 5% or more and the voting rights attached to each share class, Class T included. Analysts will spend the first afternoon recalculating Amazon's stake, and the second afternoon arguing about it.
Public companies also report every quarter. Revenue, losses and compute spending that today surface through leaks and secondary reporting would arrive in audited filings every three months. For anyone weighing Claude as a long-term platform, that disclosure is useful, since it shows whether the business paying for the model stands on solid ground.
After a listing, anyone with a brokerage account could own a sliver of Claude's parent company. Day-to-day control would stay with management and the board, and the trust structure is designed to survive the transition to public markets. Expect the prospectus to spell out how.
Is Claude AI safe, given who owns it?
For most users, ownership changes less than the headlines suggest. Your data is governed by Anthropic's terms and privacy policy, which apply whichever investor holds which slice.
Ownership does matter in two places:
- Where your requests run: AWS is Anthropic's primary cloud provider and training partner, and Claude also runs on Google Cloud and Microsoft Azure. If you reach Claude through Amazon Bedrock or Vertex AI, that platform's data terms apply too.
- Who can change the rules: pricing, plans and model retirements are Anthropic's call. If you are budgeting for a team, our breakdown of Claude AI pricing plans covers what each tier costs today.
Before you move sensitive work onto Claude, settle three things for your team:
- Which plan you are on.
- Whether your conversations can be used for training under that plan.
- Who in your organisation owns the account.
Those three answers protect you more than any shareholder list. Then learn the product properly. Claude AI for Beginners walks through choosing a model and writing prompts that hold up, and if you are comparing assistants, our explainer on whether ChatGPT is an LLM covers what Claude and ChatGPT share under the hood.
So, who owns Claude AI? Anthropic does. Amazon and Google hold large economic stakes in Anthropic, with funds and staff owning much of the rest. A five-person trust holds the keys to the boardroom. Check back after the S-1 lands, when the percentages stop being estimates.