Breaking the U.S. tech IPO financing record, Anthropic raised a maximum of 90.4 billion USD, with a fully diluted valuation of 1.4 trillion USD

Author: Zhang Yaqi, Wall Street Journal

Anthropic is knocking on the capital market with the largest IPO in U.S. tech history, with a fully diluted valuation target of up to $1.4 trillion, aiming to become the first publicly listed frontier artificial intelligence laboratory.

According to the prospectus submitted to the U.S. Securities and Exchange Commission (SEC) on Tuesday, Anthropic seeks to raise up to $90.4 billion through the IPO and private placement combined. The base IPO issuance size is up to $64.4 billion, lower than the $69.3 billion shown in the previously withdrawn version; additionally, seven tech giants, including Alphabet, Nvidia, AMD, Broadcom, Micron Technology, Microsoft, and TSMC, will separately invest a total of $26 billion through private placements. The price range is set at $70 to $78 per share, with approximately 801.4 million shares to be issued, and some existing shareholders will sell about 23.6 million shares.

At the upper limit of the issuance price of $78, Anthropic's valuation is approximately $1.02 trillion, slightly higher than its $965 billion private valuation. The stock is expected to be listed on the Nasdaq Global Select Market under the ticker ANT, with Goldman Sachs and Morgan Stanley as the lead underwriters, along with 20 other banks participating.

The listing is scheduled to be priced on November 18 (Wednesday). If the overallotment option is included, the total fundraising amount will approach $100 billion, surpassing the previous record of $86.5 billion set by SpaceX. However, AI-related stocks recently experienced a sharp decline, and concerns about tightening data center regulations and rising costs have not dissipated. Anthropic has chosen to advance its listing amid pressured market sentiment, and its IPO performance will also serve as an important indicator of investor acceptance of the commercialization prospects and high valuations of AI.

IPO Structure: Massive Private Placement Locking in Tech Giants

Anthropic's issuance is divided into two parts. For the public offering, Anthropic will issue 801.4 million shares, with some existing shareholders selling 23.6 million shares, corresponding to a market value of about $1 trillion. If employee options, restricted stock, and shares reserved for the company's charitable donation matching program are included, the fully diluted valuation will reach approximately $1.4 trillion.

The private placement portion will raise an additional $26 billion, with subscribers including AMD, Broadcom, Alphabet's Google, Micron Technology, Microsoft, Nvidia, and TSMC. This list of investors overlaps significantly with Anthropic's computing power suppliers, reflecting the unique "circular capital" structure of the AI industry—tech giants providing computing power to Anthropic are also its shareholders and customers.

Goldman Sachs and Morgan Stanley serve as co-lead underwriters for this IPO, with 20 other banks participating in the underwriting.

Explosive Revenue Growth, but Returning to Losses in Q3

The prospectus publicly discloses Anthropic's complete financial data for the first time. The company's revenue for the third quarter approached $17 billion, compared to only $1.4 billion in the same period last year, showing a remarkable year-on-year growth rate. As of October 31, the company's annualized revenue has risen to approximately $81 billion, an increase of about $10 billion since the end of September. As of September 30, the number of customers with revenue exceeding $100,000 in the past 12 months has surpassed 11,900.

However, profitability remains unstable. The company's total operating expenses in the third quarter surged to $17.5 billion, a year-on-year increase of 292%, sliding back into an operating loss of $567 million from an operating profit of $974 million in the second quarter—the first profit after at least five consecutive quarters of losses. The main factors dragging down the third-quarter performance include $5 billion in short-term computing power contract costs and $2 billion in non-cash expenses related to the company's charitable donation matching program.

Additionally, pricing pressures in AI are eroding revenue growth. The prospectus shows that some new model price reductions and contract discounts offered to customers reduced third-quarter revenue growth by about $600 million. Anthropic lowered the prices of new models like Opus 5.5 launched in September, and competitor OpenAI has also recently introduced lower-priced models.

Commitment to Large Expenditures, Dependence on Computing Power Constitutes Core Risk

The cost of maintaining technological leadership is extremely high. The prospectus shows that Anthropic's total committed expenditures amount to $895 billion, primarily for computing power and chip procurement; approximately $70 billion of this will be paid within one year ending September 2027—this figure exceeds the individual market capitalization of about two-thirds of the S&P 500 index components.

Unlike tech giants with diversified revenues such as advertising, Anthropic currently relies almost entirely on AI services for revenue, needing to convince investors that a single AI business can generate profits far exceeding infrastructure costs. This logic remains to be validated over time.

In terms of technology application, Anthropic is seeking to embed Claude into professional service scenarios, covering industries such as healthcare, life sciences, law, and accounting. The documents indicate that over 90% of the code contributions in Anthropic's codebase have been completed by Claude. Novartis CEO Vas Narasimhan serves as a board member of Anthropic, indicating that its layout in the medical field is gradually deepening. On Monday, Anthropic released new versions of the Mythos and Fable models, both of which have improved speed, and the pricing of Fable has also been reduced.

Safety Risk Warnings and Governance Structure Raise Concerns

Anthropic's founding background is unique—the core team was formed by former executives from OpenAI, driven by differences in prioritizing AI safety. Nearly six years later, the company is going public as a Public Benefit Corporation, with a governance structure that includes a "Long-Term Benefit Trust," which has the power to elect and remove a majority of the company's board members, aiming to balance shareholder interests with social benefits.

However, the company explicitly warns in the prospectus that if AI technology develops without adequate safeguards and oversight, it could pose "catastrophic or existential risks" to humanity. Regulators, investors, and even Anthropic's internal employees have raised alarms about the significant risk of humanity losing control over this technology, with some estimating that the probability of AI eliminating all of humanity within the next decade exceeds 10%. CEO Dario Amodei has also called for top AI laboratories to "control the pace of advancing frontier technologies" to prioritize safety.

In contrast, competitor OpenAI has set its IPO target for 2027, with its CEO Sam Altman characterizing 2026 as an "unwise time" to go public, citing AI safety concerns. According to insiders, OpenAI is currently raising funds at a valuation of $1.25 trillion.

Founding Team and Institutional Shareholder Lineup

The seven co-founders of Anthropic each hold about 1.4% of the shares, which, at the current valuation, places them among billionaires. Even if they fulfill their commitment to donate 80% of their personal holdings to charity, this assessment still holds.

In terms of institutional shareholders, funds under asset management firms such as Fidelity Investments, Capital Group, BlackRock, as well as venture capital firms like Iconiq, Spark Capital, and Sequoia Capital, all hold shares in Anthropic before the IPO. Google, Amazon, and Microsoft serve as investors, customers, and computing power suppliers simultaneously.