7 figures to figure out Anthropic’s business

SRI MUPPIDI AND LAUREN HIRSCH

Anthropic, the San Francisco company that makes the Claude chatbot, fully revealed its financials for the first time Monday as it races to go public in a blockbuster listing as soon as next month. In a filing, the company provided rare insight into the business of a “frontier” artificial intelligence lab that makes cutting-edge AI models.

Here’s what stood out in Anthropic’s business.

$16.3 billion

Anthropic generated $16.3 billion in revenue for the first half of this year, up nearly 14-fold from the same period a year ago — a stunning rate of growth. The company generates revenue primarily through selling subscriptions and use of its AI models and Claude AI products.

$80.7 billion

Net loss was $80.7 billion for the first half of this year. The loss appeared large because it includes a one-time charge of $72.4 billion related to convertible notes, a type of debt that will later convert into equity, and other financial instruments.

Anthropic had raised billions of dollars from major backers like Amazon and Google in the form of convertible debt. As Anthropic’s valuation skyrocketed, the value of the convertible debt also rose. Because that debt is considered a liability, it was logged by Anthropic as a loss.

$807 billion

Anthropic has committed to various contracts for computing power and other infrastructure that it needs for building and running its AI. The contracts listed in its financial prospectus could total up to $807 billion, according to a tally by The New York Times.

Not all these contracts are the same. One deal to lease computing power from SpaceX through 2029, for example, can be scrapped starting in 2027 with a 90-day warning. Others are longer term, such as a $110 billion deal with Amazon that lasts until 2036.

For contracts that cannot be cancelled, Anthropic is on the hook to pay whether it uses the computing power or not. That could pose a major risk if its growth does not keep pace with its commitments.

$132 billion

Anthropic had about $132 billion in cash and readily accessible investments at the end of June, already an enormous well of capital it can tap to fuel its growth. The fact it is going public to raise even more money underlines how aggressively the company is spending. In fact, Anthropic has so much money that the interest it earned on its cash pile alone was worth more than $1 billion in the first half of the year.

5 customers, 25% revenue

Five customers made up roughly 25% of Anthropic’s revenue in the nine months that ended Sept 30, which means the company is dependent on just a few major clients. (Anthropic did not disclose the names of the customers.) Many customers are not locked into long-term contracts and could reduce or cancel spending anytime.

50.1%

This number refers to the proportion of voting control that will be held by Dario Amodei, Anthropic’s CEO, and the company’s six other co-founders after the public listing, which would continue the founding group’s influence in the AI lab. The structure is common with founder-led technology companies, allowing founders to run their firms with some protection from the day-to-day whims of public investors.

94 pages

Of the 335 pages of Anthropic’s filing, 94 were dedicated to explaining the company’s risk factors. Among them was how AI models could pose “catastrophic or existential risks to humanity,” which could also expose the company to product liabilities and reputational harm. Anthropic also cited “intense competition” and its inability to secure enough computing power as major risks.