Who is poised to become the "pioneer large-scale model stock on the U.S. stock exchange"? The likely contender is Anthropic, the parent company of Claude, an AI large-scale model that competes alongside ChatGPT. On October 26 (Eastern Time in the U.S.), Anthropic submitted an S-1 registration statement to the U.S. Securities and Exchange Commission, seeking to list on Nasdaq under the ticker symbol "ANT," with Goldman Sachs and Morgan Stanley leading the underwriting efforts.

The specific number of shares to be issued, the price range, and the listing date remain undetermined.

This marks the first instance of a pure large-scale model company publicly filing for an IPO on the U.S. stock exchange. In June, both Anthropic and OpenAI confidentially submitted applications, which were not publicly accessible; this time, all figures have been disclosed. As for the title of the global "pioneer large-scale model stock," it was claimed by Zhipu on the Hong Kong Stock Exchange in January this year.

Previously, Wall Street estimated Anthropic's valuation at up to $2 trillion, with potential fundraising reaching $100 billion. If realized, this would surpass SpaceX's $75 billion and set a new global IPO record.

Upon reviewing the prospectus, the financial data is more candid than the rumors suggest.

$16.3 Billion in Revenue, $80.7 Billion in Losses

First, let's examine the earnings. In the first half of the year, revenue reached $16.286 billion, a 13-fold increase year-over-year.

Now, let's consider the losses. Net losses for the same period amounted to $80.748 billion, nearly double the total for the previous year.

However, approximately $72 billion of the $80.7 billion in losses stemmed from book adjustments due to the fair value increase of financial instruments such as convertible notes. Reuters pointed out that this expense does not correspond to an equal amount of operating cash outflow. The same applies to the $41.965 billion net loss in 2025—operating losses for that year were only $8.06 billion.

Focusing on the core business, here are some additional figures: operating losses narrowed from $2.135 billion to $693 million; adjusted operating profit shifted from a loss of $1.8 billion to a gain of $1.6 billion; spending on computing power and infrastructure in the first half was approximately $9 billion, compared to $2.3 billion in the same period last year.

Investing more aggressively, generating revenue more rapidly.

In the third quarter, revenue reached $16.8 billion, with a month-on-month (or quarter-over-quarter) growth of about 45%, an adjusted gross margin of 85%, an adjusted operating profit margin exceeding 11%, and positive free cash flow. These figures are based on non-GAAP standards, excluding expenses such as stock-based compensation—the company has not yet achieved GAAP net profit.

From $1.163 billion to $16.286 billion, Anthropic achieved this growth within 12 months.

However, the revenue generated is insufficient to cover the contracts signed.

As of June 30, cash and cash equivalents on hand amounted to $112.4 billion, with short-term investments of $19.3 billion.

Another set of figures lies in the "commitments" section: computing and infrastructure leases, as well as commercial agreements signed or amended after the end of 2025, involve future minimum payment obligations of approximately $417 billion; additional agreements with undetermined terms are expected to require payments of $125 billion to $150 billion. This is not even the broadest estimate—Reuters, using a ten-year horizon, puts cloud services and infrastructure commitments at least $518 billion; Bloomberg, including parts still under negotiation, estimates up to $750 billion.

Who are the competitors? Google Cloud, AWS, and Microsoft. These three entities are not only Anthropic's computing power suppliers but also its shareholders and sales channels—in 2025, 47% of its revenue was generated through Amazon and Google's cloud platforms.

In this business, the upstream suppliers, downstream channels, and shareholders are the same entities.

Anthropic states candidly in its risk disclosures: Revenue growth may be challenging to sustain, and it may not achieve or maintain profitability in the future. This is the most honest statement in the entire prospectus.

Shareholders Don't Dictate, the Market Remains Uncertain

The foundation of this company is a group of individuals who departed from OpenAI.

In 2021, Dario Amodei, then Vice President of Research at OpenAI, left with a team due to disagreements over safety and commercialization. He assumed the role of CEO, and his sister, Daniela Amodei, became president. Most of the core authors of GPT-2 and GPT-3 are part of this team.

Anthropic is registered as a Delaware Public Benefit Corporation (PBC). The founders, through Founder LLC's single Class F common stock, hold 50.1% of the voting rights for general shareholder matters; a long-term interest trust has the right to appoint a majority of the board, according to official statements.

There's another consideration for shareholders: 5% of the shares are allocated to a newly established public benefit fund, valued at approximately $100 billion based on a $2 trillion valuation; thereafter, for every 50% increase in market value, the fund receives an additional 2%. Whether this 2% comes from new or existing shares is not specified in the prospectus, and the entire arrangement still requires shareholder approval. The founders have each pledged to use 80% of their personal equity for charitable purposes.

In simpler terms: You can invest in its stock and share in its growth, but it will be challenging to alter its direction, and you'll also be contributing to its ideals.

This is not unprecedented in Silicon Valley. However, for a company on the verge of becoming the "pioneer large-scale model stock on the U.S. stock exchange," it carries significant implications—a group of individuals who left due to "safety" concerns are now going public, and this structure will enable it to proceed at its own pace, even if the market disagrees.

The issue is, the market has not yet determined its worth.

On June 12, SpaceX went public at an issue price of $135, raising $75 billion with a valuation of $1.75 trillion, still the largest global IPO to date. Four months after listing, the stock price surged to $225, then fell below $105, only returning above the issue price in early October.

This served as a preview of the "AI narrative" in the public markets.

Anthropic's valuation is also cooling down: In May, it raised $65 billion in a funding round, reaching a post-money valuation of $965 billion; a week ago, the market was still discussing a $2 trillion valuation; after the third-quarter data was disclosed, some investors adjusted their expectations to the $1.5 trillion range.

A $500 billion drop in just one week.

OpenAI has postponed its IPO to early 2027. The one currently in the spotlight is Anthropic.

However, beyond the fame, it must address not only "whether AI has a future," but also "how computing power translates into revenue," and "whether AI revenue can fill the massive hole of AI losses."