Anthropic's IPO prospectus reveals massive growth and complex governance

Rapid revenue scaling

Anthropic recently released its S-1 filing, which serves as the most detailed public statement to date on how a pure-play AI lab generates money. The document shows the company booked $16.29 billion in revenue during the first half of 2026. This represents a massive jump from the $1.16 billion reported during the same period in 2025.

Despite this growth, the pace of expansion appears to be cooling down. The company saw a revenue increase of approximately 144% in the second quarter compared to the first, but that growth rate slowed to roughly 47% in the third quarter. Management has not yet explained what is driving this sequential slowdown in the growth rate.

The company reported a gross margin of 85% for the first half of 2026. This margin is calculated by subtracting a cost of revenue of $2.38 billion from the total revenue. Investors are now looking to see if the company can maintain these high margins as it scales to meet broader market expectations.

The compute bottleneck

A significant portion of Anthropic's cost of revenue is tied to compute and infrastructure. The filing notes that operating margins are expected to fluctuate when the company delivers substantial incremental compute. This creates a complex picture for investors trying to predict future profitability as the company expands its capacity.

There is currently a lack of clarity regarding how much money is spent on training new models versus serving existing customers. Nick Patience of the Futurum Group told CNBC that this specific split is what will ultimately dictate future margins. Because the prospectus does not provide a breakdown, it is difficult to see how much of the forward commitment is earmarked for each purpose.

Anthropic has admitted to operating in a compute-constrained environment where demand has exceeded available supply. The filing states that the lab previously sought quantities of compute from cloud service providers that exceeded what those partners were willing to provide. The company must now figure out if it can secure enough capacity to keep up with its growing customer base.

Unique corporate governance

Anthropic is structured as a public benefit corporation, which introduces a unique governance model. The filing reveals that seven co-founders, including CEO Dario Amodei, hold special voting power. This power is concentrated in a single Class F share that represents 50.1% of the total voting power over matters brought to a shareholder vote.

The company describes this capital and corporate governance structure as having no direct precedent in the public markets. While the company plans to issue different classes of shares in the future, the current setup gives the founders significant control. This lack of control for outside investors is a major point of uncertainty that will likely impact the share price once trading begins.

Each of the six co-founders and the CEO owns over 200 million shares. If the company reaches its expected valuation of over $1 trillion, these holdings could be worth billions of dollars. This concentration of ownership and voting power creates a unique dynamic for a company entering the public markets.

A massive cash position

One of the most striking details in the S-1 filing is the sheer amount of liquidity the company holds. The AI lab currently has more cash than almost every other company in the United States. This level of capital is unprecedented for a company preparing to go public, providing a massive cushion for future operations.

This deep reservoir of cash allows Anthropic to navigate the high costs of infrastructure and research. However, the company is keeping its target IPO valuation secret for now. The final price will be determined through negotiations between the company, the selling stockholders, and the representatives involved.

Market estimates for the valuation currently range between $1 trillion and $2 trillion. If the company hits the higher end of that range, it could restore confidence in the AI sector. Conversely, a lower valuation could create a difficult environment for the company as it attempts to establish its public identity.