Who Really Controls Anthropic After the IPO? Governance Explained
TL;DR
- Anthropic retains a founder-led, mission-locked governance model post-IPO with Class B supervoting shares and the Long-Term Benefit Trust holding decisive power over board control and mission compliance.
- Public shareholders gain economic exposure through Class A shares but have very limited voting influence, with no ability to elect most directors or override trust safeguards.
- Major investors like Amazon and Google hold substantial equity and commercial partnerships but remain structurally non-controlling, with key decisions ultimately gated by founders and the independent Trust.
The IPO That Didn't Change Who's In Charge
Anthropic's long-awaited public market debut was pitched as a landmark moment for AI transparency — an S-1 packed with safety commitments, capped returns language, and unusual disclosures about who can fire whom. But for governance watchers, the headline was not the valuation or the ticker. It was how little control public shareholders actually bought.
Unlike a conventional tech IPO where one-share-one-vote gradually gives way to market discipline, Anthropic went public with a deliberately layered system designed to keep its public benefit mission insulated from short-term market pressure. Founders, early backers, and an independent trust each hold a different piece of the puzzle.
Founders Still At The Center
Co-founders Dario Amodei and Daniela Amodei remain the operational and strategic core of the company after the IPO. Through a dual-class share structure, they and other early insiders hold Class B shares carrying supervoting rights, giving them outsized influence over director elections, executive compensation, acquisitions, and charter amendments.
Dario Amodei continues as CEO and board member, with Daniela Amodei in a senior leadership and board-adjacent role focused on safety, culture, and alignment. Company filings emphasize that founder consent is required for major changes to Anthropic's mission, safety frameworks, and corporate form as a Public Benefit Corporation. In practice, that means no activist investor or hostile acquirer can easily pivot the company away from its safety-first charter without founder sign-off.
Inside the Dual-Class Share System
The mechanics will feel familiar to anyone who followed Google, Meta, or Snap, but Anthropic's version is stricter.
Public investors in the IPO received Class A common stock with one vote per share. Insiders hold Class B shares with multiple votes per share, which elect the majority of the board and dominate any shareholder vote. The Class B pool converts to Class A if sold or transferred outside permitted holders, preventing voting power from leaking into the public market over time.
The result is a permanent control gap: public shareholders can own a growing share of the economics while holding a minority share of the votes. Proxy advisors have flagged the structure as among the most founder-entrenching among recent large tech listings, while supporters argue it is necessary to prevent race-to-the-bottom pressure on AI safety.
The Long-Term Benefit Trust: The Real Veto Power
The most unusual piece — and the one that truly sets Anthropic apart — is the Long-Term Benefit Trust, or LTBT.
Created years before the IPO, the Trust is an independent body tasked with upholding Anthropic's long-term mission to develop safe and beneficial artificial general intelligence. Post-IPO disclosures confirm the Trust's powers were not diluted — they were reinforced.
The Trust has the power to appoint and remove a dedicated class of directors, often described as the Trust directors or independent mission directors, who sit alongside founder and investor directors. It can block changes to the company's mission, enforce compliance with Responsible Scaling and safety policies, and intervene in a deadlock or mission breach. In extreme scenarios outlined in corporate documents, the Trust can expand its board appointment power and act as a backstop against a buyer, board majority, or CEO action that would undermine public benefit obligations.
Legal experts describe it as a hybrid between a nonprofit steward and a golden share: it owns no large economic stake, but it controls governance outcomes that matter most for AI safety.
Where Investors Like Amazon and Google Fit In
Anthropic entered its IPO with two of the most powerful strategic investors in tech history — Amazon with up to multi-billion dollar investments and a primary cloud and chip partnership via AWS and Trainium, and Google with a major equity stake and cloud partnership.
Yet neither holds control. Filings and partnership agreements stress that both are non-controlling financial and commercial partners. They do not control a board majority, cannot unilaterally dictate model deployment decisions, and are subject to the same mission and Trust constraints as other shareholders.
Their leverage is commercial, not constitutional: discounted compute, distribution through Bedrock and Google Cloud, and deep integration into enterprise AI stacks. That gives them influence and board observation or minority director rights at different periods, but not veto power over safety governance. Both companies have publicly framed their stakes as consistent with responsible AI development rather than a path to acquisition.
What Public Shareholders Can — and Can't — Do
So what did IPO buyers actually get?
They got liquidity, quarterly disclosures, and economic upside tied to one of the fastest-growing AI labs behind Claude. They can vote on routine Class A matters, submit shareholder proposals under SEC rules, and sue for fiduciary breaches like shareholders of any public company.
What they cannot do is more telling. Public shareholders cannot elect the Trust-appointed directors, cannot remove founders through a simple majority vote due to supervoting math, cannot amend the mission or benefit corporation provisions without clearing Trust and founder approvals, and cannot force a sale of the company.
In short: voice without control. Anthropic is explicit about this trade-off in its risk factors, warning that its governance structure may deter takeovers, limit shareholder influence, and cause its stock to trade at a governance discount relative to one-share-one-vote peers.
Why Anthropic Built It This Way
Anthropic's leadership argues complexity is the point. The company was founded after an OpenAI split over safety and commercialization, and its charter explicitly prioritizes humanity's long-term benefit over profit maximization.
The layered model — Public Benefit Corporation status under Delaware law, founder supervoting control, and an independent Trust with board-level teeth — is designed to solve three problems at once: prevent a race to deploy unsafe models to satisfy quarterly earnings, prevent any single Big Tech backer from capturing frontier AI, and give safety researchers and staff confidence that mission commitments will survive leadership changes and market downturns.
Critics counter that it creates an accountability vacuum: a self-perpetuating system where founders pick the Trust, the Trust protects the founders, and public shareholders foot the bill without recourse if strategy fails. Corporate governance advocates have called for sunset provisions that would phase out supervoting shares after seven to ten years, but Anthropic has so far resisted a hard sunset, offering instead periodic mission reviews and enhanced Trust independence disclosures.
The Bottom Line On Who Holds Power
Power at Anthropic after the IPO flows in a clear hierarchy, not a shareholder democracy.
Day-to-day control sits with Dario Amodei and the executive team. Structural control sits with Class B holders, led by the founders. Ultimate mission control sits with the Long-Term Benefit Trust, which can overrule even a board majority on existential safety and purpose questions. Strategic investors provide capital and compute leverage. Public shareholders provide capital and market legitimacy — but not command.
For investors, that means betting on Anthropic stock is less a bet on shareholder rights and more a bet on trusted stewardship: that founders and Trust stewards will make better long-term decisions about powerful AI than the public market would.
Get All The Latest Updates Delivered Straight To Your Inbox For Free!