In its latest registration statement, filed with the US Securities and Exchange Commission on 10 November 2026, Anthropic proposed selling 825 million shares at between $70 and $78 each, potentially creating an initial public offering worth as much as $64.35 billion before the underwriters purchase any additional shares. The artificial-intelligence company behind Claude is also arranging a separate $26 billion private placement with Google, Nvidia, AMD, Microsoft and other technology suppliers and partners. The proposed transactions would make Anthropic's entry into public markets one of the largest in stock-market history. Yet the remarkable financing headline obscures a more consequential set of figures in the company's own prospectus: $33.24 billion of revenue for the first nine months of 2026, nearly $901 billion in multi-year infrastructure commitments, and a proposed governance structure in which ordinary shareholders would acquire economic ownership without control over the company's most important decisions. ¹ ²
The distinction matters because Anthropic's offering is not simply a valuation contest between competing AI companies. The company is asking public investors to support a business that has expanded at extraordinary speed but whose continued growth depends on access to computing capacity on a scale rarely contemplated in the software industry. The largest names participating in its private placement are also, in different combinations, its chip suppliers, cloud providers, commercial partners and competitors. Their investments may help guarantee the resources required to build more powerful models, but they also reveal how closely the economics of frontier AI remain tied to a small number of established technology groups. At the same time, Anthropic's founders and an independent trust are seeking to preserve control through differentiated voting rights that would limit the influence of outside shareholders. The IPO will therefore test whether investors are prepared to value an AI company not only on its rapid sales growth, but on the cost and governance arrangements required to sustain it.
As of 11 November, the transaction remains a proposal, not a completed public listing. Anthropic has applied to trade on the Nasdaq Global Select Market under the symbol ANT, but the preliminary prospectus expressly states that its information remains subject to change and that the registration statement has not yet become effective. The offer price is a proposed range; the initial market valuation is not an observed trading price; and the $26 billion private placement is conditional on completion of the IPO. Those limits are essential when evaluating estimates that have placed the company's fully diluted value as high as approximately $1.4 trillion. ³
Anthropic's proposed IPO, at a glance
Who is buying shares, and who receives the money?
The company's amended prospectus identifies 801,351,354 newly issued Class A shares to be sold by Anthropic and another 23,648,646 shares to be sold by existing stockholders. The total is exactly 825 million shares, but their economic treatment differs. Anthropic receives the proceeds from the newly issued stock after underwriting expenses; the proceeds from the secondary shares belong to those selling shareholders. At the $74 midpoint of the suggested price range, the company-issued shares would represent approximately $59.3 billion of gross share sales, while the shares sold by existing holders would account for about $1.75 billion. Roughly 97.1% of the shares in the initial offering are therefore primary issuance. This is principally a capital-raising transaction for Anthropic rather than a large-scale exit by its established investors, although the final amounts will depend on the actual offer price and the exercise of any underwriters' option. ⁴
Who is selling shares in the IPO?
The underwriters may purchase up to 123.75 million additional Class A shares from Anthropic during a 30-day option period. That amount should not be silently incorporated into the initial $64.35 billion offering figure, which reflects the 825 million shares in the base transaction at the upper end of the proposed range. The distinction between primary issuance, secondary shareholder sales and additional shares available to underwriters is important for any IPO, but especially for one as large as Anthropic's: the published fundraising headline may change considerably depending on which components are counted. The company separately estimates that the IPO and concurrent placement could generate approximately $84.6 billion of net proceeds to Anthropic at the price-range midpoint, after relevant transaction expenses. That estimate concerns the company's expected cash receipts; it is not the same figure as the gross value of stock being offered to investors.
The concurrent private placement is another significant part of the transaction. Anthropic states that Google and Nvidia intend to buy $10 billion of Class N stock each, AMD $2.5 billion, and Broadcom, Micron and Microsoft $1 billion apiece, with TSMC Partners committing another $500 million. The Class N shares are generally non-voting, according to the proposed terms. The private investments would provide capital to Anthropic and deepen financial relationships with some of the companies that manufacture or supply the hardware and cloud services on which its business depends. They are not part of the 825 million registered Class A shares offered to public-market investors, and combining the two as though they were one ordinary IPO would misstate the transaction's structure. ⁵
The $26 billion concurrent placement
These relationships are potentially valuable for both sides, but they also complicate a conventional assessment of the company's independence. Nvidia designs the graphics processors that underpin much of the AI infrastructure market, Google operates a major cloud business and develops competing models, and Microsoft provides computing services while competing across enterprise software and AI. Investment from such partners may strengthen access to infrastructure and distribution, but it also gives them an additional financial interest in Anthropic's growth. The filing does not, by itself, establish that the partners' investments will be recycled into corresponding purchases from those same companies, or that ordinary commercial arrangements are improper. It does establish that Anthropic's financing and operating relationships overlap in ways investors will need to understand when assessing pricing power, bargaining leverage and future cash requirements.
The growth that makes the valuation possible
Anthropic's sales have increased at a pace almost without precedent for a company approaching a public listing. The prospectus reports that revenue increased from approximately $41 million in 2023 to $4.6 billion in 2025, before reaching approximately $33.24 billion during the first nine months of 2026. The nine-month figure was roughly thirteen times the corresponding period a year earlier, when the company reported approximately $2.55 billion. Anthropic attributes much of this expansion to enterprise demand and adoption of Claude across professional tasks, coding and other workflows. It also reported an annualised revenue run rate of about $81 billion at the end of October. That run rate is a measure extrapolated from recent activity; it should not be represented as $81 billion of recognised annual revenue or a forecast guaranteed to materialise. ⁶
Revenue on that scale demonstrates substantial commercial demand, but it does not settle the profitability question. Anthropic's reported net loss for the first nine months of 2026 was extraordinarily large, at approximately $104 billion, but much of it reflected non-cash changes in the fair value of convertible financing instruments as the company's estimated valuation increased. The prospectus describes roughly $94.3 billion in such fair-value effects. Those accounting charges should not be confused with the amount Anthropic spent in cash delivering Claude to customers. Its operating loss over the same nine-month period was much smaller, approximately $1.3 billion, and management reported positive adjusted operating income under its non-GAAP definition. Each measure answers a different question: net income reflects recognised financial and accounting changes; operating income describes the reported performance of the business before certain non-operating items; and management's adjusted figure excludes expenses under its chosen methodology. Presenting only one would produce an incomplete description of the company's economics.
The September quarter also illustrates the sensitivity of the business to computing costs. Anthropic reported revenue of approximately $16.96 billion, but its operating result weakened compared with the preceding quarter. The company identified short-term computing capacity purchased at less favourable prices and other charges among the factors affecting the period. This is consequential because frontier AI does not share all the economic characteristics of conventional software. Producing and serving increasingly complex models requires substantial computing resources, and expansion in usage can generate additional direct costs even after the software has been developed. Whether Anthropic can maintain revenue growth while reducing the cost of serving each unit of demand will be central to its financial performance as a public company.
Nearly $901 billion of future commitments
The most arresting figure in Anthropic's filing is not its suggested market capitalisation but its disclosed infrastructure and purchasing commitments. The company reports approximately $900.9 billion in aggregate committed expenditure under purchase agreements and leases entered into as of and after 30 September 2026, substantially all connected with computing and infrastructure. That sum is spread over many years and includes short-term commitments and leases that have not yet commenced. Approximately $451.6 billion is expected to be incurred by the end of 2030, including about $70.7 billion within the twelve months following 30 September 2026; substantially all the remainder falls between 2031 and 2045. The figure is therefore not an immediate cash obligation, a measure of recognised debt or proof that the full amount will be paid on the originally indicated timetable. It is a description of contracted and expected expenditure commitments that shape the company's future financial requirements. ⁷
When Anthropic expects infrastructure commitments to fall due
The breadth of those commitments is a reflection of the strategic choices now confronting AI developers. Anthropic is pursuing a mixture of computing arrangements across cloud providers, processor architectures and more directly controlled infrastructure. Securing dedicated capacity can reduce dependence on short-term purchases and potentially improve operating costs if utilisation remains high. It also creates a risk that the company will commit to substantial expenditure before it knows whether demand, technology and pricing conditions will justify that investment. A processor generation that appears economically attractive today may be overtaken by cheaper or more efficient alternatives, while competition could reduce the prices customers are prepared to pay for AI services. The investment case therefore depends not just on the volume of future usage, but on the relationship between contracted computing costs, utilisation, revenue and cash generation.
Anthropic is not approaching this period without financial resources. The prospectus reports approximately $146.6 billion of liquidity and available borrowing capacity as of 30 September, including $95.6 billion of cash and cash equivalents, $36.3 billion in short-term investments and approximately $14.8 billion available under a credit facility. The components are not interchangeable: unused credit capacity is not cash sitting on the company's balance sheet, and the infrastructure commitment total is not payable immediately. The IPO and $26 billion concurrent placement are intended to provide additional capital. Even so, the size of the long-term obligations makes future cash flow, cost discipline and continued access to financing central questions for investors. It would be misleading to claim that a $900.9 billion commitment means Anthropic is about to run out of money, but equally misleading to treat the headline revenue figures as sufficient evidence that those commitments will comfortably finance themselves.
A public company whose shareholders may not control it
Anthropic's proposed governance is as consequential as its financing plan. The company is organised as a Delaware public benefit corporation, a structure that requires its board to balance the financial interests of shareholders with the interests of people materially affected by its conduct and the public-benefit purpose specified in its governing documents. In Anthropic's case, that purpose concerns the responsible development and maintenance of advanced AI for humanity's long-term benefit. The public-benefit designation is not a guarantee of particular outcomes, and its effectiveness depends on the powers and conduct of the people making decisions. It does, however, form part of the company's legal framework and helps explain why its founders propose a voting system that differs substantially from conventional one-share, one-vote ownership.
Under the amended prospectus, a special Class F share held through an entity associated with the seven co-founders will carry 50.1% of votes on general shareholder matters, even though the special share does not represent a corresponding claim on the company's economic returns. A separate Class T share provides the Long-Term Benefit Trust with the power to elect and remove a majority of the board's authorised directors. These are distinct mechanisms: the founders' entity controls general shareholder voting, while the trust has special board-selection rights. Taken together, they mean that investors buying publicly traded Class A shares would acquire a financial interest in Anthropic without an equivalent ability to determine who governs it. ⁸
The company presents the structure as protection against pressures that could cause an AI developer to sacrifice long-term safety commitments for near-term financial performance. There is a coherent argument for continuity in a business pursuing technologies with substantial social consequences, particularly when scientific and infrastructure investments may not produce returns for many years. But concentrated control also reduces conventional shareholder accountability. Public investors may be unable to replace leadership or change strategic direction even if operating results deteriorate or governance disputes arise. The Long-Term Benefit Trust may provide a counterweight, yet that cannot be assumed from its existence alone; the quality of its appointments, independence and oversight will have to be evaluated over time. Anthropic's IPO asks investors to accept that the people who can exercise the greatest authority over the company will not necessarily own a proportional share of its economic value.
What the market will actually be asked to price
At the upper end of its price range, Anthropic's initial offering would raise $64.35 billion in gross sales before any overallotment; its separate placement would contribute another $26 billion in planned investment. The company's basic post-offering valuation is around $1 trillion at the top end of the range, while a broader fully diluted calculation that incorporates options and potential shares can produce an estimate approaching the $1.4 trillion reported by Reuters. Those values are not directly comparable unless their share-count assumptions are stated. Investors also need to consider the amount of primary capital going into the business, which existing holders are selling, and the potential future dilution from equity awards and other instruments. The number most frequently repeated in a headline is rarely the only financially meaningful one.
There is a historical context for that caution. The world's largest IPOs have produced sharply different long-term outcomes for public investors. Visa and Meta generated substantial appreciation over many years, while other headline offerings produced more modest price gains or relied heavily on distributions to reward shareholders. The experience of those businesses does not establish whether Anthropic's valuation is justified, particularly because AI model economics and infrastructure requirements differ from the payment networks, advertising businesses and industrial companies that dominate conventional comparisons. It does show why the success of a listing should not be confused with the future return earned by someone purchasing shares at the offer price or, potentially, at a first-day market premium.
For Anthropic, the central question is whether its extraordinary revenue growth can be converted into durable profits and cash flow after accounting for the infrastructure required to deliver increasingly capable models. The prospectus supplies evidence that substantial demand already exists. It also shows the cost of serving that demand, the long-term contracts management is willing to enter, and the extraordinary degree of control the founders and trust intend to retain. These are not separate stories. They are aspects of the same proposition: investors will finance a company whose technology may become central to enterprise computing, while much of its future spending and governance will be determined outside ordinary shareholder control.
The offering has not yet been priced and Anthropic is not yet a publicly traded company. Its final valuation, first-day market reception and subsequent financial performance therefore remain unknown. What the 10 November SEC filing establishes is a more useful starting point than speculation about those outcomes. It describes who will receive the proposed proceeds, who is committing additional capital, how much infrastructure expenditure management has undertaken and who will exercise the principal voting rights after the listing. The IPO may establish what investors are willing to pay for Claude's future; it will take years to determine whether that future can earn the valuation.
Reporting method: This investigation is based primarily on Anthropic's 10 November 2026 amended Form S-1, its roadshow announcement and contemporaneous Reuters reporting. Offering terms are preliminary. IPO proceeds and the share-allocation proportions are editorial arithmetic on the disclosed terms. Revenue refers to recognised sales for the specified periods; the $81 billion run rate is not full-year revenue. Infrastructure commitments are aggregate multi-year contractual and lease arrangements, not a current cash bill. The portrait cover is an AI-generated editorial illustration rather than an original photograph of the IPO or of Dario Amodei. The Raisina Post has not examined confidential contracts or interviewed Anthropic management, its underwriters or investors. This article is general information, not an investment recommendation.