Anthropic Prepares IPO With Massive $131.7 Billion Cash

Artificial intelligence powerhouse Anthropic has officially started its path to going public by filing an S-1 registration statement with the US Securities and Exchange Commission (SEC).

The company has applied to list its shares on the Nasdaq Global Select Market under the ticker symbol ANT.

The regulatory filing offers a rare look at the financial scale required to build frontier AI models. It shows an enormous cash balance alongside massive long-term commitments for computing power.

Anthropic Holds More Cash Than Apple and Nvidia Combined

As of June 30, Anthropic reported having $131.7 billion in total liquid holdings.

This financial total breaks down into $112.4 billion in cash and cash equivalents plus $19.3 billion in short-term investments.

To highlight how large this cash reserve is, Anthropic currently holds more cash than tech giants Nvidia and Apple combined.

Data from CNBC and S&P Capital IQ shows that only two non-bank public companies in the United States hold more cash reserves: Berkshire Hathaway and Alphabet.

Understanding the $587 Billion Infrastructure Bill

While Anthropic sits on a massive war chest, training and operating AI models like Claude comes with extraordinary costs.

The S-1 filing outlines approximately $587 billion in total disclosed lease and spending commitments.

• Undiscounted Lease Obligations: $321.4 billion for datacenters and facilities that have not yet started operation.
• Guaranteed Minimum Spending: $265.6 billion in fixed commitments for raw computing power.

Anthropic's current cash reserve covers roughly 22% of this combined $587 billion commitment.

However, these lease terms run as long as 20 years, spreading payments out through 2045 rather than requiring immediate cash.

Contracts signed or updated after June 30 add another $245 billion in minimum payment obligations, though Anthropic has not yet published an updated aggregate figure.

Chip Suppliers and Take-or-Pay Clauses

Because computing hardware takes time to deploy, Anthropic usually orders compute capacity 12 to 18 months before delivery.

Its Claude AI platform relies on hardware provided by major industry partners, including Nvidia, Amazon, Google, and AMD.

As a result, shareholders in those tech firms hold indirect exposure to Anthropic's long-term spending requirements.

Crucially, Anthropic's agreements include strict take-or-pay terms.

These terms mean Anthropic must pay for the contracted computing capacity regardless of whether it actually uses it.

Financial Results: Revenue and Operating Losses

The S-1 filing also gives potential investors a clear snapshot of Anthropic's top-line performance.

In the first half of the year, Anthropic generated $16.3 billion in revenue.

The company reported an operating loss of $693 million over the same period.

At the same time, Anthropic reported an adjusted operating income of $1.6 billion, which is a internal performance metric defined by the company.

IPO Structure and Risk Factors

Reports from CNBC indicate the public offering could exceed $100 billion, potentially pushing Anthropic's total cash past $230 billion.

However, existing investors will sell shares directly into the offering. Anthropic receives no proceeds from secondary share sales by current owners.

The company intends to use its own share proceeds for general corporate purposes, including acquiring more compute infrastructure.

Full offering details are expected to be available in roughly two weeks.

In its regulatory filing, Anthropic identified key business risks. Notably, several of its main cloud providers also build competing AI models.

Additionally, total obligations exceed cash reserves by more than four times, meaning sustained revenue growth or additional capital will be necessary to meet obligations through 2045.