Anthropic's IPO shows why buying AI shares and controlling an AI company are different. We explain founder votes, board powers and public accountability.
Public Shares, Different Powers
A publicly traded AI company is governed through its board, management and the rights attached to its shares. Public investors can own part of the business without holding decisive voting power or choosing most directors. The answer depends on the company's governing documents and applicable law, rather than simply who has invested the most money.
Anthropic provides a timely example. According to its IPO closing announcement, its Class A shares began trading on Nasdaq as ANT on 19 November 2026. The offering closed the following day. For those of us interested in where AI goes next, the useful question is what changed about decision-making when those shares became publicly traded.
What Buying a Share Actually Means
An initial public offering, or IPO, is a company's first public offering of shares. The SEC's investor guidance describes shares as an ownership interest in a business. That interest can carry financial benefits and voting rights, but different types of shares can come with different rights.
Think of three separate questions: who participates financially, who votes on shareholder matters, and who appoints the board? A share class is a category of shares with defined rights. When a company uses several classes, the answers to those three questions can point to different people. Buying shares does not automatically put an investor in charge of product launches, research priorities or the next model's behaviour.
How Anthropic Divides Control
Anthropic's final SEC prospectus describes distinct founder and trust powers. A single Class F share held by the co-founders' Founder LLC carries 50.1% of the voting power on general shareholder matters, including director elections outside the trust's remit. The Long-Term Benefit Trust's Class T rights generally allow it to elect and remove directors constituting a majority of the authorised board.
Class A shares carry one vote each. Neither Class F nor Class T carries economic rights. This separates financial ownership, general voting control and board appointment authority; describing the arrangement simply as "the founders control the board" would miss an important distinction.
The board oversees corporate leadership, while management runs the business. In its original explanation of the trust, Anthropic said it did not expect trustees to intervene in ordinary day-to-day commercial decisions. That explains the intended division of work, rather than guaranteeing how every future disagreement will be resolved.
Public Benefit Is a Legal Structure
Anthropic is also a public benefit corporation. In Delaware, that remains a for-profit company: the law requires directors to balance shareholders' financial interests, the interests of people materially affected by the business, and the public benefit specified in its charter. "Public benefit" and "publicly traded" describe different things.
That balancing requirement matters, but it does not give everyone affected by AI a vote. Nor does the label certify that a company's models are safe or that its leaders will always make the right call. Our assessment is that a mission becomes more credible when we can examine both the powers supporting it and the decisions made under those powers.
The Money Matters, Too
Anthropic reported approximately US$93.5 billion in gross IPO proceeds, including shares sold by existing stockholders. About US$91.6 billion went to the company, before the relevant deductions. A separate private placement raised approximately US$26 billion. These are financing figures, not annual revenue, profit or the company's valuation.
The governance question is what happens when that capital meets competing priorities. In principle, public funding can support longer research programmes and infrastructure. Investors can reasonably want commercial progress, while leadership may judge that a particular release needs more work. An arrangement that preserves long-term decision-making can be valuable, but we should examine who gets to make that judgement and how they are held accountable.
What Public Scrutiny Can Change
Public-company status brings a continuing information obligation. The SEC explains that US public companies disclose financial results and other important developments through annual, quarterly and event-driven filings. Those disclosures give investors and other readers evidence with which to question management's account of the business.
Our view is that scrutiny and control should be assessed separately. Better disclosure can expose tensions between a stated mission and commercial decisions, even where ordinary shareholders cannot dictate the outcome. Conversely, preserving a mission through concentrated authority raises its own questions about oversight and whose interests are represented.
We would therefore judge an AI governance structure by how it handles disagreement. Who can replace leadership? What information reaches the people making decisions? What happens when those people disagree about a release? A thoughtful structure deserves attention, followed by evidence that it works.
Tanizzle Says: Follow the Decision Rights
We want AI companies to have the resources to build useful technology and the patience to make difficult decisions properly. Public investment can help finance that ambition. It also makes understanding the terms of participation more important.
Our position is straightforward: assess financial ownership, voting rights and oversight together. A large funding announcement tells us that capital has arrived. The governance documents tell us where authority sits, and the company's subsequent decisions show how responsibly it is used.
From Tanizzle: For You
For the technical question alongside this governance debate, our explanation of recursive self-improvement in AI explores AI's role in improving AI. Understanding the capability question helps put the argument about decision-making in context.
Tanizzle FAQs: Ownership and Accountability
Does publicly traded mean publicly owned by the government?
No. Here, it means the company's shares trade in public markets. Investors own the shares they hold; the term does not itself establish government ownership or democratic control.
Does an IPO make AI models open source?
No. Offering company shares to public investors does not itself release model weights or source code. Access to those materials is a separate decision governed by the terms under which they are released.
Does a public benefit corporation have to ignore profit?
No. Delaware's public benefit corporation framework includes shareholders' financial interests within a broader balancing duty. It does not turn the business into a charity or remove the need for commercial judgement.
Do all publicly traded AI companies use Anthropic's structure?
No. Anthropic is a specific example. Assess each company's share classes, board appointment rules and governing documents before making claims about who controls it.