Anthropic’s IPO Governance Test: Safety Control Versus Shareholder Power

A control structure built around the safety mission

The central innovation is the proposed Founder LLC. According to Reuters, the entity would initially be controlled by Anthropic’s seven co-founders and would hold a Class F share giving it 50.1% voting power on key matters.¹ A syndicated Reuters version described the filing as showing how Anthropic is creating a new mechanism to preserve its public-good orientation as it moves toward public markets.²

The design differs from a standard dual-class share structure in its stated purpose. Founder control at technology companies is often defended as a way to protect long-term product vision from quarterly earnings pressure. Anthropic’s argument is broader: it says its decisions may involve public-safety consequences that ordinary shareholder voting may not be suited to resolve.¹⁵

That distinction matters. If investors accept the structure, Anthropic could establish a template for frontier AI listings in which governance is not a secondary risk factor but a defining asset. The company is effectively saying that safety authority, founder alignment and public-benefit obligations are part of what investors are buying.

The investor bargain: less say, more mission lock-in

For outside shareholders, the proposed bargain is clear. They could gain exposure to one of the most closely watched AI companies in public markets, but with less power over major corporate matters. AI Affairs, summarizing earlier reporting on Anthropic’s founder-vote plan, said the proposed special share class would give the seven co-founders collective control of most corporate votes, with board elections as an exception. It also noted that the arrangement would depend on at least three co-founders retaining a minimum holding.⁶

Those limits may be tolerable for investors who see mission lock-in as economically valuable. If stronger safety governance reduces regulatory, reputational or catastrophic product risk, investors may view founder control as protection against value-destructive shortcuts. In that interpretation, Anthropic’s reduced shareholder democracy is not a concession but a risk-management feature.

The opposite interpretation is that the structure weakens accountability precisely when Anthropic’s public shareholders would be funding an unusually capital-intensive business. Reuters reported that Anthropic’s broader IPO prospectus showed a net loss of $42 billion in 2025, nearly $4.6 billion in revenue, and $518 billion in planned cloud, computing and infrastructure obligations in coming years.³ Those numbers mean investors would be underwriting not only an AI-safety mission but also a vast financing plan whose returns depend on execution, customer durability and continued demand for frontier models.

Public benefit, private control

Anthropic’s governance stack includes more than the Founder LLC. Reuters’ breakdown of the prospectus identified the company’s leadership and board structure, including co-founder and CEO Dario Amodei, co-founder and president Daniela Amodei, co-founder and chief science officer Jared Kaplan, and trustees connected to the Long-Term Benefit Trust, which is designed to balance Anthropic’s public-benefit goals.⁴

That structure suggests Anthropic is trying to separate several forms of authority: operational control by executives, oversight by directors, mission balancing through trust-linked governance, and founder voting power through the proposed Class F mechanism. The result is a layered model that may appeal to governance specialists who believe frontier AI companies need unusual safeguards.

But complexity has costs. Public investors will need to understand which body has authority in a crisis, which decisions fall under the Founder LLC’s voting power, how the Long-Term Benefit Trust interacts with directors, and what remedies shareholders would have if they believe safety language is being used to justify poor commercial decisions. The more Anthropic embeds mission protection into its corporate machinery, the more investors will scrutinize whether those mechanisms are enforceable, transparent and bounded.

Safety claim or valuation strategy?

The timing of the governance pitch is also important. Anthropic is reportedly seeking to become one of the first pure-play frontier AI labs to enter public markets, potentially setting valuation benchmarks for the sector.³ Reuters reported that the public sale could value the company at more than $2 trillion, even as the prospectus showed steep losses and major infrastructure obligations.³

That creates an unavoidable question: is the safety-governance structure primarily a mission safeguard, or is it also part of the valuation narrative?

Critics are already challenging the industry’s use of safety rhetoric. Axios reported that Uljan Sharka, CEO of European AI company Domyn, accused OpenAI and Anthropic of overstating safety concerns and argued that doom-focused narratives could support valuations ahead of IPOs.⁷ Sharka is a competitor with his own commercial incentives, and Anthropic and OpenAI say their concerns are grounded in private testing and model behavior that outsiders cannot fully evaluate.⁷ Still, the critique highlights a central tension: the stronger the safety claim, the stronger the case for founder control — but also the greater the burden on Anthropic to show that the claim is not merely governance branding.

What public investors will test

Anthropic’s IPO could become a referendum on whether public markets will accept AI-specific governance exceptionalism. Investors have tolerated founder control before, especially in high-growth technology companies. But Anthropic’s structure asks for something more sensitive: deference to a mission-led control system in a company whose products may affect public safety, labor markets, national security and information systems.

The likely investor questions are straightforward. Which corporate matters does the Class F share control? Can the Founder LLC block a sale, strategy shift, model release or change in safety policy? How are conflicts among the seven co-founders resolved? What happens if founders leave, sell down or disagree? How much power remains with ordinary shareholders through board elections? And how will Anthropic disclose safety decisions that materially affect revenue or margins?

If Anthropic answers those questions clearly, its governance could become a differentiator: a credible attempt to reconcile public-market capital with public-benefit AI development. If it does not, investors may view the structure as a high-minded version of entrenchment, leaving them exposed to the risks of a costly AI race without proportional control.

For now, the company’s planned IPO is shaping up as a test of market appetite for mission-controlled AI. Anthropic is betting that investors will accept weaker voting power in exchange for a governance system designed to keep frontier AI decisions from being dictated solely by market forces. The price of that bargain will be measured not only in valuation, but also in how much public shareholders are willing to trust founders to define the public good.