Who Really Controls Anthropic After It Becomes a Public Company?

Table of Contents

• You might want to know
• Main Topic
• Key Insights Table
• Afterwards...

You might want to know

• How much influence could ordinary shareholders have over Anthropic’s major decisions after an IPO?
• Who could shape the company’s board if its founders no longer act together?

Main Topic

Anthropic was founded in 2021 with a stated aim: to ensure that humanity can control artificial intelligence and benefit from it over the long term. As the frontier AI company approaches an initial public offering, that mission raises a question about who will have a meaningful say in its future. The people most likely to be excluded from direct control may be the very investors who buy its shares on public markets.

It is not unusual for Silicon Valley founders to retain substantial influence over a public company without owning most of its economic value. But Anthropic’s approach is especially elaborate. Its governance design combines public-benefit status, a trust with significant authority over board membership, special founder voting rights, and a mechanism that can give employees a role in certain board decisions. Taken together, these arrangements could leave common shareholders with limited influence over major matters, including who serves as a director.

First, Anthropic is a public benefit company. This is a for-profit corporation that is legally allowed to consider a public mission alongside profit. Anthropic’s stated purpose includes promoting AI safety. Public-benefit status can help protect directors when they make decisions in pursuit of that mission, even when those decisions do not maximize shareholder value in the short term. In practical terms, it may make it harder for investors to challenge the board simply because they believe a different choice would have produced higher financial returns.

That distinction matters for an AI company facing decisions whose consequences may extend well beyond its financial performance. A board could, for example, weigh safety concerns and its public mission when considering business choices. The structure does not mean that profits are irrelevant or that directors can ignore their responsibilities. Rather, it formally recognizes that the company may balance financial interests with its stated public benefit.

Anthropic’s current board is chaired by company president Daniela Amodei and includes CEO Dario Amodei, her brother; Netflix cofounder Reed Hastings; former Microsoft and GM CFO Christopher Liddell; Novartis CEO Vas Narasimhan; and Spark Capital partner Yasmin Razavi. There is one vacancy on the board. The composition brings together company leadership and people with experience in technology, finance, health care, and investing, but the company’s other governance arrangements also affect who may choose future directors.

Second, a Long Term Benefit Trust (LTBT) has been given significant influence over the board. It is an independent panel intended to keep Anthropic focused on its mission to benefit humanity. The trust holds a special class of stock that gives it the power to elect and remove a majority of the company’s board, the directors known as Class T directors. This means the trust, rather than ordinary public investors, could have a central role in determining much of the board’s composition.

The LTBT is chaired by Neil Shah, a global health expert who was managing director at GiveWell, a nonprofit that ranks charities. Its other members include former Federal Reserve chair Ben Bernanke, national security expert Richard Fontaine, and Harvard physicist Christopher Stubbs. One position on the LTBT is currently vacant, and the existing members will choose who fills it. That succession arrangement gives the trust a degree of influence over its own future membership as well as over the board.

Third, Anthropic’s seven founders jointly hold a special class of stock. A majority of the founders can exercise 50.1% of shareholder voting power, allowing them to elect three members of the board. The founders’ influence is therefore not limited to their executive roles or economic ownership; it is also built into the voting structure. As long as a majority of them act together, their special voting rights could provide them with substantial leverage over board selection.

There is an important qualification. If the founders cannot form a majority, their shares are voted in proportion with the holders of common stock. In that circumstance, the founders’ special class would no longer operate as a unified voting bloc in the same way, potentially increasing the relative influence of ordinary shareholders. The effect would depend on the circumstances and on how the relevant votes are cast.

The board also has a limited check on the LTBT. If the trust nominates a “Specified” one of its four directors, or removes that director without cause—that is, without alleging misconduct—the board’s nominating committee can decline to approve the action. If it does, the question is referred to the company’s employees for a vote. This creates a route for employees to influence a particular board decision, rather than leaving it solely to the trust or the board.

That employee role is notable because it makes the governance system more than a contest between founders, trustees, and public investors. Employees could become decisive when the board’s nominating committee objects to a specified trust action. The company’s workforce would not necessarily control the board as a whole, but in the circumstances described, its vote could affect whether a director is appointed or removed.

The consequences of a system this complex are difficult to predict. One relevant comparison is OpenAI’s so-called blip, when its board briefly fired CEO Sam Altman over concerns that he had not been candid with them. Employees demonstrated loyalty to Altman, and the board backed down, reinstated him, and a new governance structure was put in place. The episode showed how employee support can matter in a corporate power struggle. Anthropic’s framework appears to formalize an employee role in at least one type of board dispute, though the two companies’ structures and events are not identical.

Anthropic’s system could also evolve in unexpected ways if the founders stop acting as a unified group. A division among them could mean that their special voting rights no longer produce a majority, giving common shareholders more influence than they might otherwise have. Likewise, employees could overrule the LTBT on certain board seats if they support the board’s position over that of the trust. These possibilities make the practical balance of power dependent not only on formal rules, but also on relationships and decisions made during future disputes.

The founders’ control is not designed to last forever. Ten years after the IPO, they can choose to end their control of the company. Their control will also end automatically if fewer than three of them remain. This creates a defined transition point, although it does not immediately transfer all authority to common shareholders. After the founders’ control ends, the LTBT would continue to have a role.

The trust’s authority could eventually end in one of two ways: 85% of common shareholders could vote to terminate it, or the LTBT could decide to step down with the board’s approval. The required shareholder threshold is high, meaning ordinary investors may need broad agreement to remove the trust’s role themselves. Alternatively, the trust and board could jointly support its departure. Until one of those conditions is met, the LTBT would remain part of the governance structure after the founders’ special control ends.

Only when neither special class of shareholder exists would common shareholders finally gain control under the arrangement described. That outcome is a distant possibility rather than an immediate consequence of an IPO. The structure therefore asks public investors to distinguish between owning shares and possessing decisive governance power. Share ownership may provide economic exposure, but it does not necessarily give investors the final say on board appointments or the company’s mission.

The system is intended to protect Anthropic’s long-term purpose, particularly its focus on AI safety, from pressure to prioritize short-term financial returns. At the same time, it distributes authority among several groups whose interests may not always align. Founders may emphasize the company’s direction, trustees may focus on its public mission, directors may weigh their legal and strategic responsibilities, and employees may take positions shaped by their confidence in leadership. Common shareholders, meanwhile, may have less immediate influence than they would at a company with a more conventional voting structure.

The central point is that an Anthropic IPO would not automatically place the company under the control of its public shareholders. The trust’s board powers, the founders’ special voting rights, and the employee vote mechanism create a layered system in which control depends on who is acting, which decision is at issue, and whether special conditions have been met. Investors assessing the company would therefore need to consider not only its business prospects, but also the governance rights attached to different classes of shares.

Key Insights Table

Afterwards...

As Anthropic moves toward a possible public listing, the key issue will be how its governance design operates in practice. The formal rules define who can nominate directors, exercise special voting rights, and participate in certain disputes, but future relationships among founders, trustees, directors, employees, and investors will shape how those rules are applied.

For prospective shareholders, the distinction between financial ownership and control will remain important. Anthropic’s structure is intended to preserve a mission-driven approach to AI safety over time, while also creating processes through which different groups may challenge or influence decisions. Whether that balance proves durable will depend on how the company handles disagreements and how, over time, its special governance arrangements change.