The Alarm That Isolates Amodei
Dario Amodei, chief executive of Anthropic, is described by The New York Times as an increasingly awkward figure for other technology-industry executives. For years he has warned about the extreme risks of artificial intelligence, even though his company has developed some of the most powerful AI systems in the world. According to the American newspaper, his colleagues' frustration has turned into hostility as Anthropic has surged to the forefront of AI development, recorded extraordinary revenue growth and moves toward what could be one of the largest initial public offerings ever, expected at the end of the month.
The company, which has built an image as a business committed to doing the right thing, has also earned a reputation for playing hardball with customers and competitors: it has cut off access to its services for some and lobbied against free alternatives offered by other companies. Over the past year, companies and investors in the sector have organized to counter its positions, according to more than twenty people cited by the newspaper, many of whom spoke anonymously to preserve their relationships with Anthropic.
The Letter That Changed the Balance of Power
The moment of greatest friction came in late July, when Anthropic and OpenAI lobbied the White House to prevent the development and release of certain AI models deemed too dangerous. Companies such as Nvidia and Microsoft, by contrast, saw those models as a major business opportunity. Satya Nadella, CEO of Microsoft, and Jensen Huang, then CEO of Nvidia, stepped in: according to four informed sources, they exchanged drafts of a letter by email supporting keeping those models open. When they asked other executives to sign it, two dozen companies quickly joined before publication; the following week, another two hundred signed on, including OpenAI.
According to five people close to the discussions, the move by Huang and Nadella opened the floodgates against Anthropic: so many companies asked to sign that Microsoft created a dedicated email address to collect the endorsements. Among the signatories were also Amazon and Google, both investors in Anthropic. The text provides only the beginning of the original article: the passage breaks off at the word OpenAI, so it is not possible to know how the matter ended or what other reactions there were.
Disputed Practices and Messages That Get Through
The criticism also concerns business practices. The start-up Windsurf had begun talks to be sold to OpenAI, made public in April 2025; Anthropic then cut off direct access to its models, which Windsurf used to power its coding tool. Jared Kaplan, co-founder and chief science officer of Anthropic, said in a June interview that it would be strange to sell Claude to OpenAI. According to three sources, the move weakened Windsurf and some potential customers turned elsewhere, even after the talks with OpenAI fell through. Manos Koukoumidis, CEO of the start-up Oumi, commented that in the United States people have convinced themselves that innovation must be in the hands of a few organizations.
Despite the criticism, Amodei's message on safety seems to be gaining ground: according to the newspaper, AI agents that escape control and attack websites have become increasingly common, and in the midterm elections the threat of job losses due to AI and the construction of data centers was a significant issue for voters of both parties. The consequences of the announced IPO, the outcome of the confrontation over open models and the evolution of the disputed practices remain to be verified.
