Anthropic reported a revenue of US$ 17 billion, equivalent to approximately R$ 91.8 billion, during the third quarter of 2026. This figure represents a significant increase of 12 times compared to the same period last year. However, despite this accelerated expansion, the developer of Claude also recorded a net loss of US$ 23 billion, which corresponds to about R$ 124.2 billion.
These results were disclosed on Monday (2) and are considered a crucial update to the financial information the company presented in its prospectus for an Initial Public Offering (IPO). Previously, most available data covered only the first six months of 2026.
The performance in the third quarter is particularly important for investors, given the rapid evolution of the artificial intelligence (AI) market. The valuation of a company in this sector is not based solely on revenue growth, but also on its ability to maintain the necessary investments for the development and operation of its models.
If Anthropic's capital opening materializes, the company could be valued at up to US$ 2 trillion (about R$ 10.8 trillion). Furthermore, the offering could mobilize US$ 100 billion (approximately R$ 540 billion), making it one of the largest transactions of this type ever conducted.
Although the quarterly figures slightly exceeded previous forecasts made to investors, Anthropic's growth rate shows signs of moderation. Revenue grew by 47% compared to the second quarter, a percentage slightly higher than the 45% projected in a private presentation by the company.
However, this advance fell short of the 144% growth observed between the first and second quarters. This slowdown, however, does not necessarily imply a loss of commercial strength, but rather raises questions about the company's ability to convert its expansion into lasting financial results.
Brandon Ross, an analyst at LightShed Partners, commented to The New York Times that although the company is efficient, 'the money will start to drain as it builds its own infrastructure.' This concern is not exclusive to Anthropic; large technology corporations are increasing their spending on AI infrastructure and increasingly relying on the debt market to finance their projects. If demand for AI falls below expectations, these commitments could intensify pressure on results.
Anthropic confirmed that it will reserve shares for individual investors in its public offering, without specifying the proportion allocated to this group. This decision follows a similar tactic adopted by SpaceX in its capital opening in June, where retail investors had a relevant stake in the operation.
If carried out, Anthropic's offering will be the second major capital opening of the year, following the operation of Elon Musk's company. OpenAI, the creator of ChatGPT, is also considering a public offering in 2027, with a potential valuation exceeding US$ 1 trillion. This move occurs in a context of greater market caution regarding massive investments in AI, as tech company stocks suffered a strong sell-off the previous week due to questions about the sustainability of the sector's spending.
Parallel to the preparation for the IPO, Anthropic also reinforced the usage limitations of its AI tools. The company modified the guidelines for Claude Code to prohibit the tool from being used in recreating specific existing video games. This change was implemented in the system prompt of the Claude Code Opus 5.5 version.
The new guidelines instruct the model to create original games, productions inspired by genres, or projects based on designs and features provided by the users themselves. The objective of this restriction is to prevent AI from being used to directly replicate intellectual property protected games, even when the user attempts to present this copy as an original work.
Addy Osmani, a representative of Anthropic, confirmed the update to The Verge, explaining that using Claude to violate intellectual property rights contradicts the company's policies. However, the company does not intend to block the creation of AI games; original projects developed with users' own ideas and resources remain permitted. According to Osmani, Anthropic will continuously seek to adjust its instructions to balance intellectual property protection with users' creative freedom.


