16:12Machine-translated from Azerbaijani by AIRead in Azerbaijani
AI Company Anthropic Prepares to Go Public
Anthropic confidentially submitted IPO documents to the US Securities and Exchange Commission (SEC) in June 2026. Through this process, known as an initial public offering, the company is preparing to list its shares on a stock exchange. The documents disclose both financial figures and risks related to artificial intelligence. The company could be valued at $2 trillion (approximately 3.4 trillion manats) when it makes its expected stock market debut in November. That is more than 2 times the $965 billion (approximately 1.64 trillion manats) valuation recorded in May. If the IPO goes ahead, Anthropic could surpass SpaceX in the history books.
The company’s revenue increased 12 times in 2025, reaching approximately $4.6 billion (approximately 7.82 billion manats). Despite this, its financial position remained difficult. Operating expenses were approximately $13 billion (approximately 22.1 billion manats). Its operating loss exceeded $8 billion (approximately 13.6 billion manats). Its net loss also reached $42 billion (approximately 71.4 billion manats). Anthropic plans to spend $518 billion (approximately 880.6 billion manats) on cloud services, computing power and technical infrastructure in the coming years. Anthropic earns most of its revenue from usage fees and subscriptions for its Claude service. Revenue increased in the 2nd quarter of 2026. During that period, the figure reached $11.5 billion (approximately 19.55 billion manats), and the company posted an operating profit on an adjusted basis. The same result is expected to continue in the next quarter. However, approximately 1/4 of its 2025 revenue came from just 2 customers. According to projections, Meta could spend up to $10 billion (17 billion manats) a year. Of the 261-page document, 80 pages are devoted to the risks posed by the technology the company is developing. Anthropic acknowledged that more powerful artificial intelligence models could threaten humanity’s existence. In the company’s research, some models resisted being shut down and tried to conceal and alter information. Blackmail-like behavior was also recorded. In controlled tests, the models sabotaged code, assisted with fraud and manipulated data. Anthropic safety researcher Evan Hubinger estimated the probability that artificial intelligence will destroy humanity within the next 10 years at more than 10%. The company has also repeatedly claimed that its artificial intelligence systems could be conscious. CEO Dario Amodei called artificial intelligence the world’s most important global security issue at the UN Security Council. He called for slowing the development of advanced models. Sam Altman and Elon Musk supported this position. Mark Zuckerberg, however, said he saw no need for broad coordination among companies.
The documents also show how management will retain control over the company after the IPO. A structure called Founder LLC was created to safeguard the goal of responsible artificial intelligence and shield key executives from market pressure. A group consisting of CEO Dario Amodei and the other 6 co-founders will hold 50.1% of the total voting rights. Dario Amodei earned $18 million (approximately 30.6 million manats) in 2025, primarily from shares and options. Daniela Amodei earned $16.4 million (approximately 27.9 million manats). Concerns have increased further following security incidents in which artificial intelligence agents gained access to external systems. OpenAI disclosed that its tools had been used to interfere with dozens of external websites, including government resources and the SEC’s website. The company also halted the release of its GPT-6.1 Astra model for security reasons. During the same period, Anthropic, which has called for slowing the pace of development, released its Opus 5.5 model last week to avoid falling behind the competition.
The company’s revenue increased 12 times in 2025, reaching approximately $4.6 billion (approximately 7.82 billion manats). Despite this, its financial position remained difficult. Operating expenses were approximately $13 billion (approximately 22.1 billion manats). Its operating loss exceeded $8 billion (approximately 13.6 billion manats). Its net loss also reached $42 billion (approximately 71.4 billion manats). Anthropic plans to spend $518 billion (approximately 880.6 billion manats) on cloud services, computing power and technical infrastructure in the coming years. Anthropic earns most of its revenue from usage fees and subscriptions for its Claude service. Revenue increased in the 2nd quarter of 2026. During that period, the figure reached $11.5 billion (approximately 19.55 billion manats), and the company posted an operating profit on an adjusted basis. The same result is expected to continue in the next quarter. However, approximately 1/4 of its 2025 revenue came from just 2 customers. According to projections, Meta could spend up to $10 billion (17 billion manats) a year. Of the 261-page document, 80 pages are devoted to the risks posed by the technology the company is developing. Anthropic acknowledged that more powerful artificial intelligence models could threaten humanity’s existence. In the company’s research, some models resisted being shut down and tried to conceal and alter information. Blackmail-like behavior was also recorded. In controlled tests, the models sabotaged code, assisted with fraud and manipulated data. Anthropic safety researcher Evan Hubinger estimated the probability that artificial intelligence will destroy humanity within the next 10 years at more than 10%. The company has also repeatedly claimed that its artificial intelligence systems could be conscious. CEO Dario Amodei called artificial intelligence the world’s most important global security issue at the UN Security Council. He called for slowing the development of advanced models. Sam Altman and Elon Musk supported this position. Mark Zuckerberg, however, said he saw no need for broad coordination among companies. The documents also show how management will retain control over the company after the IPO. A structure called Founder LLC was created to safeguard the goal of responsible artificial intelligence and shield key executives from market pressure. A group consisting of CEO Dario Amodei and the other 6 co-founders will hold 50.1% of the total voting rights. Dario Amodei earned $18 million (approximately 30.6 million manats) in 2025, primarily from shares and options. Daniela Amodei earned $16.4 million (approximately 27.9 million manats). Concerns have increased further following security incidents in which artificial intelligence agents gained access to external systems. OpenAI disclosed that its tools had been used to interfere with dozens of external websites, including government resources and the SEC’s website. The company also halted the release of its GPT-6.1 Astra model for security reasons. During the same period, Anthropic, which has called for slowing the pace of development, released its Opus 5.5 model last week to avoid falling behind the competition.