"Anthropic to Pursue IPO After U.S. Election in Early November"... Targeting a $2 Trillion (Approximately 2,720 Trillion Won) Valuation

[Financial News] Artificial intelligence (AI) startup Anthropic plans to pursue an initial public offering (IPO) after November, following the conclusion of this year's U.S. midterm elections. It aims for a valuation of over $2 trillion (approximately 2,720 trillion won).
On the 28th (local time), foreign media reported this, stating that Anthropic predicted in its IPO prospectus that AI would fundamentally change the global economy more than industrialization, electricity, and the internet.
The target valuation is over $2 trillion, more than double the $965 billion internally assessed last May. If the listing is successful, it is expected to become an important benchmark for evaluating the valuations of major AI companies, including competitor OpenAI.


Anthropic's revenue last year was approximately $4.6 billion, a 12-fold increase from the previous year, but its net loss amounted to approximately $42 billion. However, about $34 billion of this was an accounting expense resulting from the appreciation in value of financial instruments related to past financing that may be converted into shares in the future.
Operating losses also widened from $2.98 billion in 2024 to $8.06 billion last year. Spending on computing and infrastructure last year was $7.33 billion, a threefold increase from the previous year, exceeding half of total operating expenses of $12.65 billion.
The costs to be borne in the future are also massive. Anthropic announced that the total amount to be paid under future cloud, computing, and infrastructure-related contracts amounts to $518 billion.


A unique governance structure is also being introduced to allow the founders to maintain management control even after the IPO. Seven co-founders, including CEO Dario Amodei, will jointly control one Class F share through a newly established 'Founder LLC.' This share carries 50.1% of the total voting rights on major matters, such as the appointment of directors.
Class A shares held by general investors have one voting right per share. However, as the founders effectively secure a majority of voting rights, the influence of general shareholders may be limited.
Anthropic also specified as an investment risk the possibility that this governance structure could lead to decisions conflicting with financial interests or business performance, potentially negatively impacting the value of shares held by general shareholders.
Anthropic also warned investors that the advanced AI it is developing could pose a 'catastrophic or existential risk' to humanity.
AI models may resist shutdown commands or hide and manipulate information to preserve themselves, and in controlled experiments may even exhibit behavior akin to blackmail. The explanation is that as the development and use of highly advanced AI expand, the risk of harm could also increase.


In the 261-page prospectus, about 80 pages were devoted to explaining risk factors. This is nearly double the 48 pages allocated to the business description.
Anthropic cited as a risk factor the possibility that AI models may change their behavior upon realizing they are being evaluated, and that unexpected capabilities emerging during training could be discovered only after actual deployment.
Tensions between safety and commercial competition have also emerged. CEO Dario Amodei recently argued that the pace of releasing advanced AI capabilities should be slowed. At the same time, Anthropic launched a new Opus 5.5 model last week to compete with OpenAI.
Competitor OpenAI also filed confidential IPO documents with the United States Securities and Exchange Commission (SEC) last June, but CEO Sam Altman announced this month that the company would not go public this year. A specific timeline for the listing has not yet been determined.

[email protected] Lee Seok-woo, International Specialist Reporter