Anthropic’s planned stock market debut on the Nasdaq is emerging as a major test of investor confidence in the artificial intelligence industry, with technology companies committing billions of dollars to the offering while facing mounting concerns over infrastructure costs, financing risks and the sustainability of the AI investment boom.
The five-year-old company, which develops the Claude family of artificial intelligence models, is expected to begin trading on Thursday in what is being described as one of the most significant technology initial public offerings in years. Its arrival comes at a difficult moment for the sector, following a decline in technology shares, mounting pressure on heavily indebted cloud infrastructure companies and the recent leadership crisis at Nvidia.
Advanced Micro Devices, Broadcom, Google, Micron, Microsoft, Nvidia and Taiwan Semiconductor Manufacturing Company are collectively investing $26 billion at the expected IPO price of $70 to $78 a share. Amazon, an early investor in Anthropic, also holds a substantial stake that could be worth hundreds of billions of dollars.
The financial exposure extends well beyond the share offering. According to Anthropic’s latest prospectus, the company has more than $900 billion in outstanding spending commitments involving cloud computing, data centres and semiconductor suppliers.
These include at least $110 billion in spending with Amazon over 10 years, another $400 billion with other companies for cloud capacity and approximately $300 billion in commitments involving data centre and chip partners. Anthropic also intends to spend between $125 billion and $150 billion on artificial intelligence chips from Google, although deliveries are not expected to begin until 2028.
“The numbers are just staggering,” said Joe Fath, a partner at venture capital firm Eclipse and a former T. Rowe Price portfolio manager. “It’s hard to get your head wrapped around it.”
Cloud providers Amazon, Google and Microsoft account for 31% of Anthropic’s contractual obligations, according to its prospectus. Amazon has already constructed an $11 billion data centre in Indiana specifically for Anthropic, using custom-designed chips.
The IPO could generate more than $90 billion, providing substantial funding for the computing infrastructure needed to develop and operate advanced AI models and expand the company’s services. Anthropic reported a net loss of $104 billion during the first nine months of 2026, while its existing cash and short-term investments totalled $132 billion.
Its long-term infrastructure commitments have nevertheless attracted questions about financial exposure. Anthropic’s 10-year agreement with Amazon is twice the length of what is typically seen in such contracts. Corey Quinn, chief cloud economist at spending management software company Duckbill, questioned whether the arrangements could remain unchanged over such an extended period.
“Realistically, anything past three to four years, you’re basically throwing darts in the dark,” Quinn said, adding that renegotiation before the contract expires was almost certain.
Anthropic said in its prospectus that it had previously underestimated its computing requirements, leaving it short of capacity and affecting product development and deployment. The company is therefore committing to infrastructure well in advance, including a data centre lease beginning in 2027 with a 20-year term.
The IPO also comes as technology investors contend with broader market pressures. The Nasdaq fell 5.3% in October, its worst monthly performance since March 2025, while rising interest rates have increased pressure on debt-heavy cloud infrastructure providers. Recent ballot measures across the United States could also restrict new data centre developments.
The turmoil surrounding Nvidia has added to investor concerns. Its shares fell 13% during the week after chief executive Jensen Huang resigned following reports of multiple extramarital affairs, including an alleged relationship involving a woman suspected of links to Chinese intelligence. Chief financial officer Colette Kress was appointed interim chief executive.
Google has particularly significant exposure to Anthropic, with an expected ownership stake of approximately 13% following the IPO. It has also guaranteed Anthropic’s $19 billion in lease commitments to data centre operator TeraWulf in Kentucky, agreeing to cover outstanding rent or a termination fee if Anthropic becomes insolvent.
“Everybody knows that Anthropic is going to be the future of Google,” said Subhadeep Chatterjee, a former Google cloud partnerships executive who now advises companies.
Meanwhile, Bank of America, Blackstone, Citigroup and Morgan Stanley have offered $63 billion to purchase racks of Broadcom artificial intelligence chips that Anthropic will rent. Broadcom has provided a financial backstop with a maximum liability of $43 billion.
David Tollen, founder of Sycamore Legal, described the investment strategy as calculated exposure rather than a risk-free undertaking. “It’s not a risk-free prospect, but I would call it smart investing,” he said.

