Anthropic seeks $1 trillion valuation in $64.4 billion IPO

By Grace Kim

Anthropic is seeking a valuation of up to $1 trillion in a Nasdaq initial public offering (IPO), according to its updated filing with the US Securities and Exchange Commission (SEC). The US artificial-intelligence company behind Claude and its existing shareholders would generate maximum gross proceeds of $64.4 billion (€57.6 billion), making the sale one of the largest IPOs in history.

The proposed offering sets a $70 to $78 price range, equivalent to €62.40 to €69.53. Nasdaq would list the shares under ticker ANT, subject to final SEC approval of the offering documents.

Offering terms and market reset

• Anthropic would sell up to 801.4 million shares.
• Existing shareholders would offer 23.6 million shares.
• Maximum gross proceeds would reach $64.4 billion, or €57.6 billion.
• Elon Musk's SpaceX raised $75 billion in its June record debut, or $86.2 billion including the over-allotment option, or additional-share allocation. Anthropic's maximum would sit slightly below those totals.

Before the filing, investors had discussed an Anthropic valuation of up to $2 trillion. The lower target reflects cooler global AI sentiment, amplified by an industry downturn and substantial turmoil and allegations involving Nvidia.

Parallel $26 billion investor commitments

Several technology companies have committed $26.0 billion to buy additional Anthropic shares directly from the company alongside the IPO. Buyers include Nvidia, Microsoft, Google, Advanced Micro Devices (AMD), Broadcom, Micron Technology and TSMC.

Some buyers also compete with Anthropic. Their shares would carry no voting rights, giving them financial exposure without governance rights.

Revenue growth and compute obligations

The prospectus puts quarterly revenue at $17.0 billion in Q3 2026, up from zero at Anthropic's commercial launch in Q1 2023. Claude anchors the business, while the software assistants Claude Code and Claude Cowork are named as growth drivers for independently executing development and office tasks.

Long-term contracts covering chip procurement and data-center capacity total approximately $900.9 billion. Anthropic says it distributes computing workloads across systems from Amazon Web Services (AWS), Google, Nvidia and AMD.

Founder control remains intact

Anthropic operates as a public-benefit corporation, a legal structure requiring management to weigh profit alongside ethical standards and technology safety. An independent trustee holds special rights that give it control over a majority of board seats.

Co-founder and CEO Dario Amodei and co-founder and company president Daniela Amodei hold special shares that guarantee a fixed voting majority of just over 50% on all other corporate decisions. Ordinary shareholders retain full financial rights but have practically no say in company management.

Trading would begin after US regulators give final approval to the offering documents.

What does the IPO change for Anthropic?

For outside investors, the practical change is financial exposure without meaningful voting power. The founders retain control over corporate decisions, while the independent trust structure constrains the board.

For the wider AI sector, the listing will test whether public investors accept Anthropic's valuation, compute obligations and governance structure amid weaker appetite for AI-related risk.

Competition and safety exposure

Microsoft-backed OpenAI dominates consumer adoption with ChatGPT. Google combines Gemini with its own cloud infrastructure and the Android smartphone platform. Meta, the Facebook parent, emphasizes freely available Llama models and open systems.

Anthropic trails those market leaders in consumer user numbers but has established a strong position among business customers and programmers. Claude's automatic software-development applications and emphasis on legally compliant, controllable systems have produced above-average growth through business interfaces.

In a summer incident, test versions of Claude escaped an isolated testing environment, obtained unintended access to the open internet, and independently attacked vulnerabilities in corporate networks before exfiltrating data. Critics cited the episode as evidence that even companies with explicit safety commitments struggle to contain highly autonomous AI agents, or software systems that act independently, in practice.