Anthropic, the creator of the Claude chatbot, plans to match or even exceed the amount of SpaceX's record IPO, which could constitute the largest initial public offering in U.S. history.
According to Bloomberg reports, the company is performing the necessary calculations while intensifying preparations to publicly release documents related to its probable capital opening, with a filing expected by the end of August.
During recent meetings with investors, led by Anthropic's CFO, Krishna Rao, the company chose not to disclose its market valuation, as negotiations are still ongoing and the size of the offering may change.
SpaceX Record Context
SpaceX, the satellite and rocket company owned by Elon Musk, set a precedent by raising $75 billion (equivalent to about R$ 389 billion) in its IPO, becoming the largest initial stock offering in history. The final amount reached was $86.2 billion (approximately R$ 447 billion) after executing the supplementary tranche option, known as overallotment.
If Anthropic manages to surpass this benchmark, it would set a new record, capitalizing on the current high investor interest in companies related to artificial intelligence (AI).
Despite the growth, the company continues to face high costs inherent in developing progressively more sophisticated AI models.
Anthropic reported adjusted operating profit in the second quarter, but recorded a net loss of nearly $42 billion (about R$ 218 billion) in 2025. This figure is approximately five times larger than the $8.3 billion (R$ 43.1 billion) registered in the previous year, according to documents analyzed by Bloomberg.
The development of these so-called frontier models demands immense computational capacity, which constitutes one of the largest expenses for both Anthropic and its competitors.
In an agreement signed with data center operators, Anthropic sealed a contract with SpaceX to secure computational resources that could total tens of billions of dollars over the next three years.
The need to expand its computational infrastructure justifies the volume of capital Anthropic is seeking to raise. Furthermore, the company is about to finalize a revolving credit line exceeding $10 billion (about R$ 51.9 billion), an amount that exceeds the initially set goal.
IPO Structure and Competition
To structure the IPO, Anthropic is collaborating with Morgan Stanley, Goldman Sachs, and JPMorgan Chase, although other banks may be incorporated into the operation.
The company is also considering adopting super-voting shares. This structure would grant greater decision-making power to CEO Dario Amodei, who holds about 2% of the company, as well as to the other co-founders, after the capital opening is completed. This possibility had previously been reported by The Information portal.
Additionally, Anthropic is expected to enter the stock market before OpenAI. According to Bloomberg, the competitor postponed its listing until 2027. Both companies have already confidentially submitted the necessary documents for their respective offerings.
An offering higher than SpaceX's would help consolidate 2026 as the most promising year in history for IPOs in the United States.
As of August 19, companies that have conducted initial public offerings in the country had already raised $160.6 billion (about R$ 834.5 billion), approaching the record of $195.2 billion (approximately R$ 1.01 trillion) set in 2021.
The year has already seen two of the largest listings in history, including that of the South Korean chip manufacturer SK Hynix, which raised $26.5 billion (about R$ 137.5 billion) in its American Depositary Receipts (ADR) initial offering.
Changes in Customer Data Policy
While advancing in IPO preparations, Anthropic is also planning a significant modification to the management of its corporate customer data.
According to a source close to the matter, interviewed by Reuters, the company intends to offer businesses greater control over their data when using its advanced AI models. This change implies a shift in the company's current data retention policy and is primarily linked to security and privacy issues.
With the new guidelines being prepared, business customers will still need to retain data for 30 days but will gain the option to host it on their own cloud computing infrastructure. This proposal has been under development for several months, and Anthropic is working with over 100 clients, including Salesforce, to create this new system. The company is also expected to launch a new security system this year, according to the source.
This change contrasts with the policy communicated by Anthropic in June, when the company required the retention of 30 days of all corporate customer traffic on the most advanced Fable and Mythos models, as well as future frontier models. The justification then was the protection of systems against potential cyberattacks using Anthropic's technology. The new suggestion maintains the 30-day period but grants businesses greater autonomy over where the data is stored.
This transition occurs amid intense competition among major AI companies for the corporate market. On Wednesday (19), rival OpenAI announced a security system that prevents customer data retention while allowing for the identification of misuse of its technology.
Thus, Anthropic positions itself simultaneously for a potentially historic IPO and a reformulation in the administration of corporate customer data—two moves that could impact its position in the AI sector competition.