Seven former OpenAI employees with complementary profiles are preparing to make their stock market debut with Anthropic. Their shares could be worth as much as $17 billion to $18 billion each if the price range is reached, and a package of $250 million in cash is planned for each. All claim a course focused on the security of AI, while expressing reservations about the race to commercialization.
Risks publicly discussed by Kaplan, McCandlish, Olah and Dario Amodei
Several co-founders have taken a stand on the risks associated with AI and their motivations. Jared Kaplan believes that allowing an AI to improve recursively would represent “the ultimate risk” and places the possibility of such a decision between 2027 and 2030. Sam McCandlish says that none of the founders initially wanted to start a company, but that they acted out of duty. Chris Olah explains that he has long advocated for a non-profit structure dedicated to security research rather than a commercial company. Dario Amodei says he turned to AI to tackle “almost unmanageable” and “slow” biology problems, and criticizes the use of the term “impact” as a recruitment argument. The co-founders’ work areas cover interpretability, pre-training, and other core topics to monitor model behaviors and document their limitations.
What the IPO would change for them: cash, price per share and valuation
Anthropic’s IPO, scheduled for trading on November 19, promises to be one of the largest ever. The seven co-founders are each due to receive $250 million in cash and could see the value of their stock portfolio cross $18 billion if the price range is reached. Together, they own almost 10% of the capital, with almost 240 million Class A shares per founder, which equates to about 1.4% individual ownership. In the most recent round, the stock price was $58.90, which already equated to about $14 billion for each co-founder. The targeted range of $70 to $78 per share corresponds to a market valuation close to $1.4 trillion, which would bring the individual value of the stakes to around $17 to $18 billion. Anthropic did not respond to requests for comment on these items.
Key technical skills: RLHF, scaling laws, pre-training and interpretability
Anthropic’s co-founders have contributed to several major technical advances. Dario Amodei and Tom Brown are credited with co-inventing RLHF, a tuning method aimed at aligning model results with human expectations, such as politeness or avoiding dangerous responses. Tom Brown also led the formation of GPT-3, which marked a jump in scale from 1.5 billion to 175 billion parameters compared to GPT-2. Jared Kaplan co-pioneered model scaling laws with Dario Amodei and helped formalize “constitutional AI,” which relies on security principles and policies for alignment. Current projects include pre-training on very large corpora and interpretability, which aims to understand what inputs and reasoning methods a model uses to produce its results. Sam McCandlish also led early research into Codex, which now represents a significant portion of OpenAI’s revenue.
AI strategies, movements and tricks deciphered, every evening in 5 minutes. Free.
Included upon registration: our selection of the best AI guides & comparisons.
Choose your pace
Free · No spam · Unsubscribe in 1 click
Roles and governance: CEO, presidency, internal think tank and functions
Dario Amodei, 43, runs the company and owns 245,259,100 shares, including securities placed under voting proxies. Daniela Amodei, 39, serves as president and chairman of the board of directors. Jack Clark is responsible for public benefits and leads the Anthropic Institute, an internal think tank dedicated to the social and economic dimensions of AI. These assignments structure the company’s line in terms of security, public utility and scientific strategy.
Founders’ Journey: From Princeton to Baidu, MIT, Johns Hopkins and Journalism
The profiles of the seven founders span research, engineering and public policy. Dario Amodei defended a thesis in computational neuroscience at Princeton in 2011, did a postdoctoral fellowship at Stanford until 2014, then led speech recognition research at Baidu in China before a stint at Google Brain, and joined OpenAI in 2016. Tom Brown, graduated from MIT in 2010, founded MoPub and Grouper, experienced the acquisition of MoPub by Twitter, then self-trained at machine learning research. Jared Kaplan, scientific director and associate professor at Johns Hopkins, is a Ph.D. in physics (Harvard, 2009) and a former postdoctoral fellow at SLAC. Sam McCandlish, PhD in theoretical and mathematical physics (Stanford, 2017), worked at Boston University; his academic work focused on quantum gravity and tensor networks. Chris Olah, 34, left the University of Toronto after a year and worked at Environment Canada and then Xelerance. Several have passed through OpenAI in key functions: Tom Brown managed the training of GPT-3; Jack Clark left journalism there for AI policy; Jared Kaplan worked there as a consultant; Sam McCandlish and Chris Olah were members of the technical staff there. Jack Clark, a former journalist in the United Kingdom, studied literature and creative writing, writes for various media, and explains that he joined OpenAI out of personal conviction. In terms of governance and origins, Dario and Daniela Amodei, raised in the Mission district of San Francisco and attended Lowell High School, have local roots close to the current headquarters. Daniela graduated from UC Santa Cruz in 2009, worked for Democratic Rep. Matt Cartwright, then spent five years in risk management at Stripe; she indicates that the experience of campaigns naturally led her towards startups.
Why they left OpenAI and started Anthropic
The collective departure at the end of 2020 can be explained by disagreements over security and commercial trajectory. Dario and Daniela Amodei, then vice presidents at OpenAI, were dissatisfied with their employer’s approach to security and feared that the models’ capabilities were advancing faster than their understanding. Early the following year, they founded Anthropic and brought with them Tom Brown, Jack Clark, Jared Kaplan, as well as Sam McCandlish and Chris Olah. The founders were concerned about OpenAI’s commercialization ambitions. Jeffrey Ladish affirms that they did not create Anthropic to enrich themselves but to shape a field considered crucial, while recalling the weight of financial incentives. Co-founders are also touted as early beneficiaries of the rise of AI.

