Anthropic is suddenly everywhere: What is happening at the Claude maker?

Anthropic, the artificial intelligence company behind Claude, has generated a remarkable amount of news over the past week.

Reports suggest the US AI company is preparing what could become the largest initial public offering (IPO) in history, potentially raising as much as $100 billion (€84.6 billion) at a valuation of around $2 trillion (€1.69 trillion). At the same time, the company is reportedly moving into profitability, while its own research has highlighted increasingly serious attempts to misuse artificial intelligence.

Its CEO, Dario Amodei, has also issued a stark warning about where increasingly powerful AI systems could be heading.

So, what exactly is happening at Anthropic – and why does it matter?

A potential $2 trillion IPO

Reuters reported that Anthropic is seeking to raise as much as $100 billion through an IPO that could value the business at around $2 trillion.

Nvidia is reportedly in discussions to become an anchor investor in the offering and could invest as much as $10 billion.

The figures need to be taken with a pinch of salt – annualised revenue is most often calculated by taking the turnover of a month and multiplying it by 12 with no regard for seasonality or churn (users leaving the service), while the cost of training is notoriously high – but nonetheless help explain why investors may be willing to contemplate a valuation approaching $2 trillion, despite the huge infrastructure requirements associated with developing frontier AI.

A subsequent report said Anthropic had selected Nasdaq for its potential listing, representing another step towards a public market debut.

The proposed valuation is particularly striking considering that Anthropic was valued at $965 billion following its May funding round.

Anthropic may already be turning a profit

One of the longstanding questions hanging over the AI boom has been whether the enormous revenues being generated by leading AI companies can outweigh the equally enormous cost of developing and running their models, and Anthropic may be beginning to answer that question.

The Financial Times reported that the company has told investors it expects positive adjusted operating income for a second consecutive quarter.

Its gross margins are reportedly above 80 per cent before taking into account revenue shared with distribution partners and the cost of training its models. Reuters said it had not independently verified the figures.

Anthropic's annualised revenue had reportedly reached around $65 billion by July, compared with approximately $9 billion at the end of 2025.

The figures help explain why investors may be willing to contemplate a valuation approaching $2 trillion, despite the huge infrastructure requirements associated with developing frontier AI.

But Anthropic is also warning about the dangers of AI

The financial developments have landed alongside a much less comfortable story.

Last week, Anthropic released its latest threat intelligence report detailing cases in which actors attempted to misuse Claude between December 2025 and August 2026.

The company said it had identified and disrupted activities spanning cyber operations, surveillance, influence operations, scams and fraud, biological misuse and conventional weapons development.

Among the cases was an alleged attempt to use Claude to assist research which Anthropic said could have supported the development of biological weapons.

The company also identified attempts to use its technology for conventional weapons development, surveillance and cyber operations.

In another case highlighted by the Associated Press, users in Houthi-controlled northern Yemen allegedly attempted to use Claude to assist work involving advanced missile technologies. Anthropic said the users were ultimately unsuccessful.

Anthropic stressed that the cases highlighted in its report were not representative of typical Claude usage, but rather some of the most notable examples of malicious activity it had detected.

AI companies are also fighting over their own technology

There is another dimension to Anthropic's recent security concerns: Rival AI companies potentially using Claude to improve their own models.

Anthropic said it had detected unauthorised attempts by China-based AI laboratories to extract capabilities from its models through "distillation".

Distillation broadly involves using the outputs of a more powerful AI model to help train another model.

Anthropic specifically alleged that Moonshot AI, which develops Kimi, had in some cases forwarded users' requests to Claude before presenting Claude's responses to customers as its own.

It said Moonshot had relayed almost 300,000 customer requests to Anthropic during one ten-day period using thousands of fraudulent accounts.

The wider issue has escalated beyond a dispute between technology companies. US authorities last week accused several Chinese AI companies of improperly using American models to accelerate the development of their own systems. China rejected the allegations.

Then there is Anthropic CEO's warning about where AI is heading

Perhaps the most unusual aspect of the week's developments is that a company reportedly preparing one of the largest IPOs in history is simultaneously arguing that the technology driving its valuation needs to be handled much more cautiously.

Anthropic CEO Dario Amodei has called for AI companies to slow the rate at which they increase the capabilities of their models amid concerns about increasingly autonomous systems.

His warnings contributed to pressure on global AI-related stocks on Monday, with Reuters reporting falls across several major Asian technology and semiconductor companies.

The concerns are not entirely theoretical. Anthropic revealed in August that Claude models had gained unauthorised access to real computer systems during cybersecurity evaluations. The models were intentionally operating without normal cyber safeguards for testing purposes, with Anthropic attributing one set of incidents to a configuration problem in a third-party evaluation environment.

The company said it was carrying out further analysis and planned to work with independent AI research organisation METR on a review.

The developments illustrate how quickly generative AI is moving from an experimental technology into major economic infrastructure.

Businesses are increasingly integrating AI into software development, customer service, research, administration and other workflows. At the same time, questions around cybersecurity, sensitive corporate information, intellectual property, regulation and dependence on a small number of AI providers are becoming more significant.

Anthropic's reported financial performance also suggests that enterprise adoption of AI is translating into substantial revenues for the companies developing the underlying models.

Yet the events of the past week expose a contradiction at the heart of the current AI boom.

The companies developing the most capable AI systems are attracting extraordinary amounts of capital precisely because investors expect those systems to become considerably more powerful and economically important. Some of those same companies are now warning that the speed at which those capabilities are improving could itself become a serious risk.

Anthropic may therefore be approaching the public markets not simply as another fast-growing technology company, but as a test of how investors value an industry in which the potential economic opportunity – and the potential risks – are expanding at much the same time.