Anthropic may spend $750 billion on computing costs as IPO plans advance
By David Morris
Anthropic PBC could spend $750 billion on extra computing power in the coming years, according to a report by The Wall Street Journal. The company is preparing to sell shares to the public for the first time through an initial public offering (IPO).
The artificial intelligence company lost $80.7 billion in the first half of 2026, the Journal reported, citing a draft of its IPO prospectus, a document that explains the company's finances and plans to potential investors. This was much higher than the $13 billion loss it reported for the same period a year earlier, according to the Journal.
According to the report, $72 billion of the company's loss in the first half of this year came from accounting charges. These charges were linked to changes in the value of convertible notes and other financial instruments that Anthropic had issued.
Anthropic, one of the leading companies developing AI, is expected to publicly file its IPO paperwork in the coming weeks. This could lead to one of the biggest stock market listings ever. Selling shares will help the company raise more money to invest in computer chips, data centers and skilled workers so it can build new and better AI models.
The company behind Claude spent nearly $9 billion on computing power and infrastructure during the six months ending June 30, according to the Journal. It spent $2.3 billion during the same period in 2025. In just the first half of 2026, Anthropic spent more money securing computing power than it spent during all of 2025, when its spending reached $7.3 billion.
The prospectus shows that Anthropic has signed many agreements to secure computing power. According to the Journal, these agreements are worth about $450 billion over the next seven to 10 years, and most cannot be canceled. The company plans to spend $111 billion on contracts with Google Cloud, owned by Alphabet Inc.; $110 billion with Amazon Web Services, owned by Amazon.com Inc.; and $31 billion with Microsoft Corp.
Anthropic's prospectus warns that AI could cause harm. For example, its models might show behavior that appears to protect themselves from being turned off. They might also respond in ways that resemble "blackmail" to stop people from preventing them from operating, according to the report. The company also faces the risk of political opposition to building data centers in the United States, where most of its computing capacity is located, the Journal reported.
Bloomberg News has reported that Anthropic told investors it had become profitable in the third quarter when measured using adjusted figures. People familiar with the matter have said that the company was generating positive free cash flow, had a gross margin of 85% and had a positive adjusted operating margin of more than 11% as of the third quarter.
Bloomberg News has reported that Anthropic could begin officially promoting its IPO as soon as the week of November 9. This would put the company on track to begin trading before the US Thanksgiving holiday. It is expected to make its stock market debut by the end of the year at the latest.