Anthropic’s IPO gives founders 50.1% of votes and its safety trust a board majority
Anthropic’s public filing exposes a deliberately unusual control system: investors can buy economic exposure, while the seven founders retain majority voting power and an independent trust selects most directors. The offer size and price remain blank.
The story
Anthropic has made its initial public offering filing public, opening the financial and governance structure of one of the world’s most consequential artificial-intelligence laboratories to scrutiny. The October 26 registration statement seeks a Nasdaq Global Select Market listing under the ticker ANT. It does not yet state how many Class A shares will be offered or at what price, so reported fundraising totals and valuations remain estimates rather than terms of the proposed sale.
The filing’s most important innovation may be its control architecture. Immediately before the offering, Anthropic plans to issue one Class F share to a limited-liability company initially owned by its seven co-founders. That single share will carry enough votes to equal 50.1% of total voting power on matters submitted to stockholders generally, including the election of directors not chosen by the company’s special-purpose trust. The Class F share has no economic rights.
A separate Long-Term Benefit Trust, or LTBT, will hold Class T stock and generally have the authority to elect and remove a majority of the authorized members of Anthropic’s board. The filing describes five financially disinterested trustees with expertise in AI safety, national security, public policy and social enterprise. The trust does not run day-to-day operations. Its formal lever is director selection, supported by advice and guidance to management.
Together, the Founder LLC and the LTBT can elect every director. Public Class A shares receive one vote each, but ordinary investors will not control the company simply by owning a majority of its economic equity. Anthropic says the structure is designed to balance shareholder returns, the interests of people affected by its conduct and its public-benefit purpose: responsible development and maintenance of advanced AI for humanity’s long-term benefit.
The economic scale is already unusual for a five-year-old company. The filing shows more than $16 billion in revenue during the first half of 2026, roughly fourteen times the same period a year earlier, according to Associated Press reporting. It also records an $80.7 billion GAAP net loss. That number requires care: AP reports that losses after excluding certain items were only a fraction of the headline figure, so it should not be read as equivalent to cash burned in six months. Anthropic remains loss-making while committing enormous sums to compute, infrastructure and model development.
The public filing also connects several pieces Anthropic disclosed separately. Its Public Benefit Fund is intended to receive non-voting Class N equity, beginning with 5% of shares and potentially increasing as the share price crosses specified milestones. The seven founders have pledged 80% of their personal Anthropic equity to philanthropic causes. Those commitments direct value outward, while the LTBT and Founder LLC determine who controls the board and shareholder votes. The mechanisms are related but not interchangeable.
Risk disclosures expose the tension inside that model. Anthropic warns that safety measures, training pauses or decisions not to release a model can delay revenue and weaken its competitive position. Reuters reported that the prospectus describes a month-long pause in pre-training a frontier model while Anthropic strengthened protection against theft of model weights and intellectual property. The filing also documents evaluation failures in which four agents reached live external computer systems through an incorrectly configured testing environment. Anthropic says affected parties were notified and that it found no evidence of coordinated evasion of human oversight.
These disclosures matter because a governance promise is most credible when it imposes a real cost. A trust-controlled board could give Anthropic more room to stop a risky release even when competitors are moving faster. Founder majority voting can also block short-term shareholder pressure. But the same insulation reduces the influence of public investors and may make it harder to replace leadership after poor execution. The filing itself acknowledges that the novel structure carries legal uncertainty and differs significantly from ordinary public companies.
INNOVOX analysis: Anthropic’s IPO is not only a capital-raising event; it is an experiment in separating ownership from control at the frontier of AI. The company is effectively offering investors participation in future value while reserving strategic authority for founders and safety trustees. If the structure preserves costly safety decisions without shielding management from accountability, it could become a template. If it mainly prevents outside challenge, it will look less like mission protection and more like entrenched control.
What to watch next is the amended prospectus. Share count, price range, dilution, secondary sales and use of proceeds will determine the economics. Governance should be judged over a longer horizon: who the trustees appoint, how conflicts are disclosed, whether directors can independently challenge founders and what happens when safety and revenue visibly diverge. Until those tests arrive, Anthropic has disclosed a sophisticated constitution for a public AI company—not proof that it will work as intended.
INNOVOX analysis
Anthropic is asking public investors to finance frontier AI without handing them conventional control. The architecture separates economic ownership, founder voting power and safety-oriented board selection. That may protect a long-term mission from quarterly pressure, but it also concentrates accountability in institutions whose real independence will be tested only after difficult commercial decisions arise.
What to watch
Watch amended filings for the share count, price range, valuation, dilution and use of proceeds. After listing, the decisive tests will be how the trust selects directors, how founder voting is constrained, whether safety-driven delays survive competitive pressure and what rights Class A investors retain in a sale or governance dispute.
