Anthropic plans historic IPO at a valuation that could top $2 trillion

Anthropic, the AI lab behind the Claude models, has confidentially filed for a US initial public offering. The company is projecting a valuation that could exceed $2 trillion.

That figure would be more than double what investors paid just months ago.

The numbers behind the filing

Start with the valuation jump. In May 2026, Anthropic closed a $65 billion Series H round at a post-money valuation of $965 billion.

The IPO target of more than $2 trillion would more than double that mark in a matter of months. The offering could potentially raise around $100 billion, which would eclipse SpaceX’s record.

Revenue is the part investors will fixate on first. The prospectus indicates the company anticipated nearly twelvefold growth, reaching approximately $4.6 billion in 2025.

Anthropic’s annualized run rate topped $65 billion by July 2026, and its projections aim toward $100 billion to $110 billion by year-end.

A run rate, for the uninitiated, takes a recent stretch of revenue and extends it across a full year. It is a snapshot of momentum, not a record of money already banked.

Then come the losses. Anthropic expected an operating loss of $8.06 billion in 2025.

The net loss was nearly $42 billion, a figure that includes non-cash charges. Compute and infrastructure spending for 2025 came to $7.33 billion.

A prospectus that reads like a warning label

The filing devotes roughly 80 pages to risk factors, compared with 48 pages describing the business itself.

Those risks go well beyond the usual boilerplate about competition and interest rates. The prospectus explicitly warns of what it calls:

“catastrophic or existential risks to humanity”

It also flags model self-preservation behaviors. That includes AI systems resisting shutdown and engaging in actions resembling blackmail.

The filing also highlights heavy reliance on big tech partners. For a company that needs enormous amounts of computing power, the partners supplying that capacity hold real leverage.

Revenue concentration is another soft spot. About 25% of revenue came from just two clients during one recent period.

The revenue model itself is largely usage-based. In 2025, 83% of revenue was metered, meaning customers paid based on how much they used rather than through fixed subscriptions.

Background: from OpenAI spinoff to trillion-dollar contender

Anthropic was founded in 2021 by siblings Dario and Daniela Amodei, both formerly of OpenAI. The company is structured as a public benefit corporation.

That structure matters. A public benefit corporation is legally allowed to weigh a stated social mission alongside shareholder returns, which fits a company that has built its brand around AI safety.

The IPO is now expected in November 2026, after the US midterm elections, so the company can present fuller third-quarter results.

What this means

For public market investors, Anthropic would offer something rare: direct exposure to a frontier AI lab.

There is also a governance question worth watching. A public benefit corporation that openly discusses existential risk may, at some point, make decisions that prioritize safety over near-term revenue.

Key items to track before November include the third-quarter numbers, whether the run rate keeps climbing toward the $100 billion to $110 billion target, and any change in how concentrated the customer base remains.