Anthropic files S-1 for IPO with $1.8 trillion valuation target

The path to a massive IPO

Anthropic has officially filed its S-1 prospectus with the SEC, marking a major milestone for the AI company. This document allows the firm to begin its roadshow in 15 days and potentially list its shares on the Nasdaq under the symbol ANT by mid-November.

The company is aiming for a valuation between $1.8 trillion and $2.0 trillion. If successful, this would make Anthropic the second-largest market debut in history, trailing only SpaceX.

The filing serves as a roadmap for Wall Street to judge the company's viability. Investors are looking to see if Anthropic can generate enough profit to justify such a massive price tag while managing its heavy spending on computing power.

Strong revenue and margins

The S-1 reveals that Anthropic's business fundamentals are currently very healthy. In the first half of 2026, the company saw its revenue explode 13x, jumping from $1.3 billion to $17 billion compared to the previous year.

While the company reported a paper loss of $80 billion under GAAP rules, this was largely due to non-cash charges from convertible notes. When those are removed, Anthropic actually earned an adjusted operating income of $1.6 billion, representing a margin of over 9%.

The company also demonstrated positive operating leverage. In the second quarter of 2026, its revenues of $11.6 billion were more than double its $5.6 billion cost for the compute power needed to run its models like Claude, Fable, and Mythos.

The $441 billion asterisk

Despite the strong sales, Anthropic faces a significant financial obligation. The prospectus lists infrastructure service agreements and equipment leases totaling $630 billion with partners like Microsoft, Amazon, and SpaceX.

A critical detail is that 70% of these arrangements are non-cancellable. This means Anthropic must pay for the reserved capacity regardless of whether they actually use it, creating a fixed cost of approximately $441 billion.

Because most of these contracts run for seven years or less, the company must generate at least $60 billion in annual sales just to cover these commitments. Anthropic plans to expand its data center footprint even further, which will increase these capital outlays significantly.

Shift toward hybrid infrastructure

Anthropic is changing how it acquires the hardware needed to train and run its AI. Previously, it relied on partners like Microsoft to pay for and install the chips, which typically account for 80% of a data center's cost.

The company is now moving toward a hybrid model. In some instances, Anthropic will buy the chips itself and pay to equip dedicated centers operated by other providers. This is a much more capital-intensive strategy than what the company has done in the past.

This shift explains why the company's capital spending has been relatively low until now. As they move toward owning more of the hardware, their cash outlays for capacity are expected to explode in the coming years.

Shareholder dilution and growth

The prospectus also includes a unique provision regarding the Anthropic Public Benefit Fund. The board can issue an additional 2% of outstanding shares to this philanthropy each time the stock price rises 50% over a specific benchmark.

This means that if the company's stock price rockets, existing shareholders will see their ownership percentage diluted to support the fund. Conversely, if the stock price drops significantly, shareholders could end up with a smaller piece of the business than they started with.

Ultimately, the S-1 shows a company with a thriving core business but a massive leap-of-faith wager. Anthropic must prove that its sales can ramp up at a pace that outstrips its growing, locked-in infrastructure costs.