Anthropic co-founders prepare for a multi-billion dollar IPO
The seven researchers who left OpenAI to prioritize AI safety are set to go public, with stakes potentially reaching $18 billion each.
The short version
- The seven co-founders collectively own about 10% of Anthropic but maintain a 50.1% majority voting control.
- Each founder holds about 240,000,000 class A shares, valued at roughly $14 billion in the most recent funding round.
- The founders plan to dedicate 80% of their personal wealth from the IPO to philanthropy, though this is not legally binding.
- Anthropic is setting aside 5% of its shares to create a public benefit fund for nonprofits and community grants.
Why it matters
This move marks a major transition from private research to public accountability for one of the leading AI labs. It also highlights the massive financial stakes now tied to the development of safe artificial intelligence.
A defection over safety
Anthropic was born from a group of researchers who grew frustrated with OpenAI's direction at the end of 2020. While OpenAI sparked the current generative AI trend, several employees felt the company was prioritizing commercial goals over safety protocols. They feared that AI capabilities would advance much faster than the world's ability to understand or control them.
The core of the group included siblings Dario and Daniela Amodei, who were vice presidents at OpenAI. They were joined by a mix of researchers, a journalist, a public policymaker, and an engineer. This diverse team launched Anthropic in early 2021 with the specific goal of shaping AI development more responsibly than their former employer.
The founders have maintained that their primary motivation was not personal wealth. Jeffrey Ladish, a contract safety engineer who joined the company in 2021, noted that they wanted to shape a significant global technology because they believed others might do it worse. Now, however, they are positioned to be among the first to see significant financial rewards from the AI boom.
The diverse team of founders
The leadership team combines deep academic research with practical engineering and policy experience. CEO Dario Amodei holds a doctorate in computational neuroscience from Princeton and worked at Baidu and Google Brain before joining OpenAI. He is credited with co-inventing reinforcement learning from human feedback, a method used to align AI outputs with human needs, such as politeness or safety.
Daniela Amodei serves as the president and board chair, bringing a background in politics and risk management. Other key figures include Tom Brown, an engineer who led training for GPT-3, and Jack Clark, a former journalist who now leads the company's internal think tank. These individuals represent a blend of technical expertise and a focus on the social and economic impacts of new technology.
The group also includes Jared Kaplan, a theoretical physicist who pioneered scaling laws and constitutional AI. Sam McCandlish and Chris Olah joined from academia and Google, respectively, focusing on pre-training and interpretability. Interpretability is a branch of study that attempts to understand the specific data inputs and reasoning methods a model uses to arrive at its outputs.
IPO stakes and voting power
Anthropic is preparing for a major public offering expected to begin on Nov. 19. While the seven co-founders own only about 10% of the total equity, they hold a 50.1% majority voting control through a legal entity called Founder LLC. This structure ensures that the original team maintains significant influence over the company's direction even as it becomes a public entity.
The financial scale of this IPO is massive, with a potential market valuation of nearly $1.4 trillion. At a target price of $70 to $78 per share, each co-founder's stake could be worth between $17 billion and $18 billion. This would make them some of the wealthiest figures in the technology sector, following a period of private funding where their shares were valued at about $14 billion each.
The prospectus indicates that the founders expect to sell over 20 million shares during the IPO, resulting in roughly $250 million in cash per founder. To balance this wealth, they plan to dedicate 80% of their personal equity proceeds to philanthropy. Additionally, the company is earmarking 5% of its shares for a public benefit fund to support nonprofits and community grants.
The path toward 2030
The founders' work focuses heavily on long-term risks, including the possibility of AI recursively self-improving. This refers to a scenario where an AI writes its own code to enhance its capabilities without human intervention. Jared Kaplan has suggested that this specific risk could become a primary concern between 2027 and 2030.
The company's strategy involves balancing these existential risks with the need for commercial viability. By moving to a public market, Anthropic will face increased scrutiny regarding how it manages these safety protocols. The goal remains to ensure that as models become more powerful, they remain aligned with human values and safety principles.
As the company prepares for its debut, the focus remains on the tension between profit and purpose. While the founders have expressed a desire to avoid being influenced by financial incentives, the move to a public market introduces a new layer of investor expectations. The success of Anthropic will likely serve as a bellwether for how other AI labs navigate the transition from private research to public commerce.
What to watch
- Whether the 80% philanthropy pledge becomes a legally binding commitment.
- How the market reacts to Anthropic's valuation compared to other tech giants.
- The specific safety milestones Anthropic aims to reach before 2027.
Try it
Look up the 'Constitutional AI' principles pioneered by Jared Kaplan to see how researchers use lists of rules to guide AI behavior.
Sources and links
- VideoY Combinator's Lightcone podcast youtube.com
- VideoDecember 2024 video youtube.com
- Postemail to a colleague at Bloomberg x.com
- ReportingCNBC: the article we sourced
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.
Listen to the AI Wire headlines
Listen to the AI WireListen to our daily AI shows every morning: on your commute, in the shower, wherever you are
Listen to the daily showsFirst reported by CNBC · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.
