Anthropic is preparing to seek a $2 trillion valuation in an initial public offering, with its roadshow targeted for the week of November 9. Morgan Stanley, Goldman Sachs and JPMorgan are leading the offering, and Nasdaq has been selected as the listing venue.
The company recorded a net loss of $80.7 billion in the first half of 2026, compared with a $13 billion loss a year earlier. Most of the latest loss came from a $72 billion non-cash accounting charge linked to the rising fair value of convertible notes.
Revenue reached $16.3 billion in the first half of 2026, up from $1.1 billion a year earlier. Third-quarter revenue then rose to $16.8 billion, a 45% sequential increase from Q2 2026.
Anthropic spent nearly $9 billion on computing capacity and infrastructure in the first half, up from $2.3 billion a year earlier. Around one-fifth of its first-half revenue came from two customers.
The larger question for prospective shareholders is the scale of the company’s future obligations. Anthropic has committed more than $600 billion to computing infrastructure and has indicated that spending could reach $750 billion in coming years, including potential deals still under discussion.
About $450 billion of that spending is tied to cloud agreements running for seven to ten years. Most of those contracts cannot be cancelled, and around 80% of the payments are due whether the capacity is used or not.
Anthropic has committed approximately $111 billion to Google Cloud, $110 billion to Amazon Web Services and $31 billion to Microsoft. It also has commitments of up to $86 billion for Nvidia chip rentals and more than $160 billion in Google chip leasing obligations through Broadcom.
Some current and prospective investors who attended Anthropic’s investor day in San Francisco, California last week are weighing a lower valuation for the company. Three people who attended said investors discussed a valuation of around $1.5 trillion.
Anthropic has told prospective investors that it was profitable on an adjusted basis.
Its prospectus also lists risks from models producing responses resembling blackmail, self-preserving behavior aimed at avoiding shutdown, and the possibility that the financial consequences of a catastrophic AI incident could exceed available insurance.
Comments (0)