AdvertiseFee Deal Desk

TradFi

Anthropic Founders Sold Nearly $2B of Stock in IPO, Kept 50.1% Voting Power

Anthropic's founders sold about $2 billion of stock in its Nasdaq IPO but kept 50.1% voting power through a Class F share, its SEC prospectus shows.

Be a creator
November 21, 2026, 03:14 AM UTC4 min read
AI SummaryAI
  • Anthropic's founders each sold about 3.4 million shares for roughly $280 million in the November 19 IPO.
  • Founders retained 50.1% voting power through a Class F share held by their Founder LLC.
  • Anthropic raised $93.5 billion in the IPO and $26 billion in a private sale, $119.5 billion combined.
  • Anthropic reported 5,087 vulnerabilities across 680 open-source projects as of November 2, with 629 fixed.

Founders Sold $2B, Kept Voting Control

The founders of Anthropic sold roughly $2 billion of their own stock in the company's initial public offering on Nasdaq on November 19, yet kept 50.1% of the voting power, according to the SEC prospectus for the listing. Shares trade under the ticker ANT, and the sell-down ranks among the largest founder cash-outs on record.

Each of the seven founders sold about 3.4 million shares, taking in roughly $280 million apiece before taxes and fees. Selling shareholders collected about $1.95 billion in total after underwriters' deductions, the filing states, before any other costs. The sales left founder wealth largely intact. CEO Dario Amodei's remaining stake is worth about $20.07 billion and Daniela Amodei's about $19.76 billion, while the five other founders hold close to $20 billion each. CFO Krishna Rao still owns roughly $788 million of shares. Lockups and vesting restrictions limit when any of them can sell more, so their remaining wealth moves with how the shares perform after the listing.

Control sits with the founders through a special Class F share held by their Founder LLC, the instrument that carries the 50.1% voting stake. They have also pledged to donate 80% of their personal equity to charity. The pledge is voluntary and can be withdrawn, and its potential value has been put at $156 billion, a figure describing future donations rather than money already given. On Friday, November 20, the company said the IPO produced $93.5 billion in total proceeds, including shares sold by existing holders, of which Anthropic itself received about $91.6 billion. A separate private placement added $26 billion from buyers including Nvidia, Google and Microsoft, lifting combined gross proceeds to $119.5 billion, with other semiconductor manufacturers also among the participants.

30-Day Deadline for High-Risk Bugs

A second change announced on November 16 shortens how long security flaws stay private. Anthropic will cut the public disclosure window for high-risk and critical vulnerabilities reported to software vendors and maintainers from 90 days to 30 days, and to 45 days for less severe issues, applying to vulnerabilities reported from January 1, 2027. Coordinated vulnerability disclosure, the practice being revised, asks researchers to report a flaw privately and give the vendor time to patch before details go public. The 90-day window has been the standard used for years by Google Project Zero and similar teams.

Anthropic argued those conventions were written before AI could find vulnerabilities and rested on two assumptions: that patching moves at human speed and that a flaw one party discovers is unlikely to be found again independently. The company said both assumptions now fail, because models can help write patches and are increasingly rediscovering each other's findings. The revised policy escalates a report to an external coordination body 15 days after a silent maintainer, down from 30 days, and publishes full technical details 30 days after a patch ships, down from 45. Vulnerabilities actively exploited in the wild keep a 7-day deadline.

The scale behind the change is visible in the company's own disclosure dashboard. As of November 2, Anthropic's models, including an early version of Claude Mythos Preview, had found and reported 5,087 vulnerabilities across 680 open-source projects; only 629 are known to be fixed, about 12%. Each report passes human review at external security firms, which the company calls the bottleneck. CEO Dario Amodei has said separately this month that AI is surfacing hundreds of thousands of software flaws and that fixing them is harder than finding them. Hard-to-patch critical infrastructure software and fixes that require specification changes can negotiate separate schedules, while AI findings without human review, such as output from the OSS Scanner service, fall outside the new deadlines. Legacy high-risk reports older than 90 days get an extra 30 days' notice before publication.

A $1.05 Trillion Commitments Schedule

The same SEC filings show what the raise has to cover: $900.9 billion in fixed commitments plus $150 billion to $175 billion in estimated future payments, about $1.05 trillion to $1.08 trillion in total, or roughly 24 times annualized revenue for the first three quarters of 2026, with payments stretching to 2045. Anthropic's share of enterprise AI API spending rose from 40% to 55% over the past year, while a Gartner survey found only 13% of CIOs reporting significant value from AI spending. Read together, the two announcements frame the post-IPO company: founders have locked in 50.1% voting control while public shareholders carry the execution risk, and the shortened disclosure window shows the security workload is scaling faster than maintainers can absorb. Whether revenue growth outruns the commitments schedule through 2045 is the test the $119.5 billion in proceeds was raised to meet.

Primary sources

COINOTAG's editorial and research desk.

AI-Assisted

AI-generated, AI-reviewed, under COINOTAG editorial oversight.