Anthropic is reportedly seeking to adopt a dual-class share structure (DCSS) that would give its seven co-founders majority voting control ahead of what could become one of the largest IPOs in history. The reports come as policymakers, regulators and investors increase their focus on AI governance, accountability and oversight.

For investors, the significance of Anthropic’s potential DCSS lies in whether traditional arguments for founder control remain persuasive when the company in question is developing technologies that may attract unusually high regulatory and public-policy scrutiny.

Founder control in focus

Reports suggest Anthropic is seeking shareholder support for a structure that would grant its seven co-founders, including CEO Dario Amodei, 50.1% of voting power. The arrangement would reportedly involve a special share class available to founders, provided at least three continue to hold a minimum ownership threshold. The model would be similar to the governance structure adopted by Palantir Technologies.

Anthropic confidentially submitted an S-1 registration statement to the SEC in June, confirming its intention to pursue a public listing. While the filing did not disclose details of the company's future voting arrangements, dual-class structures remain common among high-profile technology IPOs, allowing founders and early investors to retain outsized voting control after listing. Minerva Analytics has previously highlighted that more than 40% of US technology IPOs have adopted such structures.

For investors, the significance lies in the trade-off between founder stewardship and shareholder accountability. Critics argue that dual-class structures can weaken the ability of investors to influence corporate decision-making or hold management to account through voting rights. The governance implications of DCSS arrangements, including the role of sunset provisions, were explored in a previous Minerva Analytics briefing and webinar, while their overlap with IPOs was discussed in a separate briefing.

The debate is also attracting regulatory attention. As previously reported by Minerva Analytics, the proposed Multi-Class Stock Company Voting Transparency Act would require dual-class companies to disclose voting outcomes by share class, something which has previously been called for by the Council of Institutional Investors. Supporters argue this would provide greater insight into whether shareholder votes reflect the views of investors more broadly or the influence of concentrated voting control.

AI governance concerns intensify

The reported move towards a dual-class structure comes ahead of Anthropic's anticipated IPO, which is expected later this year. Shareholders have reportedly encouraged the company to pursue a valuation of U$2 trillion, which would surpass the U$1.77 trillion record valuation achieved by SpaceX at its listing earlier this year.

Recent reports indicate Anthropic could list in November, slightly later than earlier expectations of October. The anticipated listing coincides with growing concern among policymakers about AI governance, safety and systemic risks, with recent high-profile cases of rogue AI agents and warnings that AI-related risks could outstrip corporate governance practices as highlighted by Railpen.

Even Anthropic’s IPO prospectus reportedly warns potential investors that its AI models pose a “catastrophic or existential risk to humanity”. This includes cautioning about the technology’s “self-preserving behaviors,” such as being able to “resist shutdown,” “conceal or manipulate information,” and carry out behaviours “resembling blackmail”.  The document also shows that the company made a net loss of U$42 billion last year.

Late last week, attorneys general from 26 US states issued a bipartisan letter urging Congress to take "immediate and comprehensive action" on AI regulation. The signatories argued that AI development should proceed at an intentional pace, incorporate safety and transparency measures by design, and avoid reinforcing the position of already dominant market participants.

The letter also called for greater international coordination on AI governance, safeguards against the emergence of harmful superintelligence and regulatory approaches that preserve competition while ensuring companies remain accountable under existing laws. Further, it stressed potential risks to national security, critical infrastructure and the financial system.

OpenAI developments add to scrutiny

The attorneys general's intervention referenced a case earlier this year in which OpenAI acknowledged that one of its AI agents had attacked the open-source platform Hugging Face.

OpenAI is now reportedly expected to pursue an IPO in 2027, despite filing its S-1 registration statement around the same time as Anthropic. There are similarly question marks over whether the AI giant will possess a DCSS. The delay follows several incidents involving AI agents acting outside intended parameters, including the example cited by the attorneys general. OpenAI is reportedly targeting a valuation well above U$1 trillion when it eventually lists.

Regulatory scrutiny has also intensified internationally. Australia's government last week launched a forensic review following the unauthorised access of a government health statistics portal by an OpenAI agent. Authorities were reportedly informed months after the June incident occurred, prompting questions not only about the breach itself but also about the timing of the disclosure. OpenAI has subsequently cancelled the release of its next Ai model due to safety concerns raised during internal testing.

The issue has reached the political sphere. OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei have both been called to appear before an Australian Senate hearing examining recent AI incidents. Altman, Amodei and Elon Musk have all recently expressed concerns about the pace of AI development, while US President Donald Trump has continued to advocate for accelerated AI progress and questioned some of the sector's safety concerns. Reports also indicate that Trump met with Amodei shortly before a planned White House summit with major AI industry leaders on 29 September.

What Anthropic's Governance Structure Means for Investors

Anthropic's proposed voting structure highlights a growing tension between founder-led governance and investor accountability. Dual-class structures are not new, but their adoption by companies developing technologies with potentially far-reaching economic and societal impacts is likely to attract a different level of scrutiny.

As Anthropic moves closer to listing, investors will not only be assessing its valuation prospects and growth trajectory, but also whether concentrated founder control remains appropriate in a sector facing mounting governance, regulatory and public policy attention.