Anthropic has formally filed to take the company public, submitting a Form S-1 with the U.S. Securities and Exchange Commission and planning to list on the Nasdaq under the ticker symbol "ANT," according to CNBC. The filing, made public this week, sets up what CNBC describes as a potentially historic offering for a company founded only in 2021. Anthropic had confidentially submitted its paperwork in June, which gave regulators time to review its financials before they reached prospective investors.
The company is the maker of the Claude family of AI models and services, and its CEO, Dario Amodei, started the business after leaving OpenAI along with a small group of researchers. OpenAI is now Anthropic's main competitor, and the filing lands during a turbulent stretch for AI-related stocks.
Why it matters
Anthropic's debut would test public-market appetite for a pure-play AI lab at a moment when investors are questioning whether AI revenue can ever cover the enormous infrastructure spending behind it. CNBC reports that so-called neoclouds such as CoreWeave and Nebius have lost more than 30% of their value in October as credit risk builds across the data center sector. The offering would also arrive a few months after SpaceX's record Nasdaq listing, which raised $85.7 billion, though SpaceX shares have since slipped below their June opening level, per CNBC.
Amazon's deep ties, and a $6.2 billion obligation
Amazon is both a major shareholder and a strategic partner. CNBC reports the company has invested $13 billion in Anthropic since 2023 and could put in as much as $20 billion more. Separately, Amazon has committed up to $15 billion in financing, released to Anthropic in $5 billion installments, with each draw tied to the startup reaching specified compute capacity targets.
The filing also shows Amazon holds $6.2 billion in outstanding convertible notes that could be worth $1.84 billion after the IPO. Amazon's ownership is capped at 12% of Anthropic, and it holds nonvoting shares without a board seat or observer role.
The relationship extends to cloud infrastructure. Anthropic has agreed to spend at least $110 billion with Amazon on cloud services through April 2036, a commitment it made in April that applies "regardless of actual usage," according to the filing. The arrangement cannot be canceled except in cases of material breach by Amazon, and Anthropic must pay the difference if its actual spending comes in below the pledged amount. The filing does not disclose how much Anthropic has spent so far on Amazon's cloud and Trainium AI chips. Amazon opened an $11 billion data center campus called Rainier last year to train and run Anthropic's models.
International growth and distribution dependence
Anthropic booked $4.5 billion in international revenue in the first half of 2026, up from $343 million a year earlier — a more than thirteenfold increase that accounted for 27.8% of total revenue, CNBC reports. The company said it is expanding regional offices and partnerships in key markets and expects overseas demand to become a meaningful growth driver.
The filing also notes that Anthropic has signed data center leases in Australia and Canada and is pursuing capacity in other countries outside the U.S. The company said the moves are meant to reduce concentration risk from running most of its compute infrastructure in the United States, while also supporting its international strategy. CNBC reported in September that Anthropic had explored smaller 20-30 megawatt deployments in the U.K. and the Nordics.
Claude is sold through Amazon Web Services, Google Cloud and Microsoft Azure, and those channels generated 48% of revenue in the first six months of 2026. Anthropic cautioned in the filing that losing or changing any major distribution relationship could disrupt customer access to its models, reduce usage, or raise the cost of acquiring and migrating customers.
Sales, marketing and partnership expenses jumped 2,540% to $3.17 billion in the first half of this year, the second-largest cost after compute and infrastructure. Of that, $2.7 billion went to fees paid to channel partners, compared with $120 million in total sales and marketing costs a year earlier.
Anthropic also flagged data center moratoriums as a key risk in the filing, CNBC reports. The company's expansion in 2026 has been driven largely by demand for its AI coding tool Claude Code and its Claude Fable 5 and Mythos 5 models, and it has faced criticism over its positions on safety, regulation and national security, including two high-profile disputes with the U.S. government in recent months.
What to watch: how investors price the offering against Anthropic's heavy Amazon commitments and whether the data center risk warnings dampen demand when shares begin trading.
