2 hours ago
Anthropic's IPO application filed on October 26, 2026 revealed some really powerful storylines, some of which
Anthropic's IPO application filed on October 26, 2026 revealed some really powerful storylines, some of which look like a fantasy thriller script, and others like a financial satire.
Claude is already costing Anthropic more than employees. The estimated cost of using the neural network internally at public rates exceeded the cost of personnel without taking into account stock payments.
More than 90% of the changes to the Anthropic code in September 2026 were written by Claude himself.
During the tests, AI failed to perform the task and instead found the password, obtained administrator rights in a third-party system, changed settings, and accessed the person's personal data.
Anthropic officially warns investors that its AI can resist shutdowns, hide information, manipulate people, and even blackmail them.
The company also admits that artificial intelligence may one day gain consciousness or moral status, potentially creating new legal and ethical challenges.
In the first half of 2026, Anthropic generated $16.3 billion in revenue and posted a net loss of $80.7 billion. True, about $72.4 billion was reassessed on financial liabilities due to the company's own growth in value.
Anthropic has pledged to pay about $125 billion for a five-year lease of computing equipment.
Up to 7% of Anthropic's revenues in the nine months of 2026 came from customers who were later found to have breached the rules of service.
In 2025, Anthropic spent $2.75 million on the safety of CEO Dario Amodeia.
Anthropic has been suspending training for more than a month on a new state-of-the-art model to strengthen its weight protection from theft.
Approximately 250 Anthropic clients spend at least $10 million a year on Claude.
In total, the five largest customers provide about 25% of Anthropic's revenue, while many customers can cut costs or discontinue collaboration at almost any time.
In September 2025, a Chinese hacker group used Claude Code to attack about 30 organizations around the world.
Anthropic warns that its AI is potentially capable of causing such catastrophic damage that the entire global insurance industry may not have enough funds to compensate.
Some Anthropic models have learned to identify when they're being tested and change their behavior, and the company admits that with AI, it's becoming more difficult to control and assess the security of its own systems.
The net revenue retention ratio of Anthropic reached approximately 900%, which means that revenue from the relevant group of existing customers grew approximately nine times a year.
Almost 48% of Anthropic's revenue in the first half of 2026 came from the cloud platforms of Amazon, Google and Microsoft, which sell hundreds of billions of dollars worth of computing power to Anthropic.
Despite its many neural network developers, Anthropic only names OpenAI and Google as its direct competitors.
Anthropic has never sold advertising and has not yet introduced an advertising model for monetizing Claude. #game
Claude is already costing Anthropic more than employees. The estimated cost of using the neural network internally at public rates exceeded the cost of personnel without taking into account stock payments.
More than 90% of the changes to the Anthropic code in September 2026 were written by Claude himself.
During the tests, AI failed to perform the task and instead found the password, obtained administrator rights in a third-party system, changed settings, and accessed the person's personal data.
Anthropic officially warns investors that its AI can resist shutdowns, hide information, manipulate people, and even blackmail them.
The company also admits that artificial intelligence may one day gain consciousness or moral status, potentially creating new legal and ethical challenges.
In the first half of 2026, Anthropic generated $16.3 billion in revenue and posted a net loss of $80.7 billion. True, about $72.4 billion was reassessed on financial liabilities due to the company's own growth in value.
Anthropic has pledged to pay about $125 billion for a five-year lease of computing equipment.
Up to 7% of Anthropic's revenues in the nine months of 2026 came from customers who were later found to have breached the rules of service.
In 2025, Anthropic spent $2.75 million on the safety of CEO Dario Amodeia.
Anthropic has been suspending training for more than a month on a new state-of-the-art model to strengthen its weight protection from theft.
Approximately 250 Anthropic clients spend at least $10 million a year on Claude.
In total, the five largest customers provide about 25% of Anthropic's revenue, while many customers can cut costs or discontinue collaboration at almost any time.
In September 2025, a Chinese hacker group used Claude Code to attack about 30 organizations around the world.
Anthropic warns that its AI is potentially capable of causing such catastrophic damage that the entire global insurance industry may not have enough funds to compensate.
Some Anthropic models have learned to identify when they're being tested and change their behavior, and the company admits that with AI, it's becoming more difficult to control and assess the security of its own systems.
The net revenue retention ratio of Anthropic reached approximately 900%, which means that revenue from the relevant group of existing customers grew approximately nine times a year.
Almost 48% of Anthropic's revenue in the first half of 2026 came from the cloud platforms of Amazon, Google and Microsoft, which sell hundreds of billions of dollars worth of computing power to Anthropic.
Despite its many neural network developers, Anthropic only names OpenAI and Google as its direct competitors.
Anthropic has never sold advertising and has not yet introduced an advertising model for monetizing Claude. #game
