Anthropic Sets IPO Price Range at $70 to $78 and Launches Roadshow Toward November 19 Nasdaq Debut

Anthropic, the AI safety company behind the Claude family of models, launched its initial public offering roadshow on Monday under the ticker ANT on the Nasdaq Global Select Market, setting a price range of $70 to $78 per share. At the midpoint, the offering values Anthropic's post-IPO outstanding shares at approximately $975 billion — placing the company just below the trillion-dollar threshold it had previously targeted in private markets, and well below the $2 trillion valuation that had circulated in pre-roadshow investor briefings.

The offering covers 825 million Class A shares: 801,351,354 newly issued by the company and 23,648,646 sold by existing shareholders. Goldman Sachs, JPMorgan Chase, and Morgan Stanley are serving as lead underwriters, joined by 19 additional banks in a 22-institution syndicate. Underwriters hold a 30-day overallotment option for up to 123,750,000 additional shares; if exercised in full, the total diluted capital raise reaches approximately $73 billion.

Anthropic is targeting the week of November 17 for pricing and November 19 or 20 for trading to begin — the week before the Thanksgiving holiday. CNBC reported Monday that the book is already covered within the $70–$78 range, suggesting baseline institutional demand is in place heading into the roadshow.

Five Share Classes Lock in Founder Control and Mission Independence

Anthropic's capital structure is unusually complex even by tech-company standards. Its amended S-1 describes five distinct share classes, each carrying different economic and governance rights, specifically designed to protect what the company describes as its AI safety mission from the near-term pressures of public ownership.

Class A shares — the ones offered in the IPO — carry one vote each and represent the standard public float. Class N shares are non-voting stock issued to strategic cloud and technology partners; they carry full economic participation but no governance power. After the IPO, Google is expected to hold approximately 1.79 billion Class N shares; Amazon, approximately 1.92 billion.

The most significant governance layer sits in Class F stock: a single share issued to Founder LLC, controlled by CEO Dario Amodei, his sister and President Daniela Amodei, and their five co-founding colleagues who left OpenAI to start Anthropic in 2021. Regardless of how many shares are outstanding — in the IPO, in later secondary offerings, or through option exercise — Class F carries a fixed 50.1% of all combined voting power across all classes. The founders do not need to retain any economic interest to maintain control. The share has no dividend rights and no liquidation preference.

Sitting above even Class F in operational terms is the Long-Term Benefit Trust: a five-trustee body with the sole authority to elect the majority of Anthropic's board. The trustees hold Class T shares, which carry no economic rights whatsoever. Because the board majority is elected by an independent trust rather than by common shareholders, Anthropic qualifies as a "controlled company" under Nasdaq listing rules and is accordingly exempt from some board independence requirements that apply to ordinary public companies.

Anthropic incorporated as a Delaware Public Benefit Corporation, a legal form that specifies the responsible development of AI for the long-term benefit of humanity as a binding corporate purpose — not advisory language. Delaware PBC directors must balance financial returns against that stated benefit, and shareholders have legal standing to challenge decisions that deviate from it. At the scale Anthropic is seeking to reach, the PBC structure has not been tested in public markets.

Revenue Growth of 14x Runs Beside an Accounting Loss That Needs Unpacking

Anthropic's H1 2026 revenue reached $16.286 billion, compared with $1.163 billion in the year-earlier period — roughly a 14-fold year-over-year increase. For Q3 2026, the company told investors revenue was approximately $17.0 billion, representing 45% sequential growth in a single quarter. Full-year 2025 revenue was $4.585 billion.

The headline net loss for H1 2026 — $80.748 billion — is almost entirely an accounting artifact rather than a cash outflow. Anthropic issued convertible notes and SAFEs to its investors before going public; under GAAP fair value accounting, as the company's valuation rose sharply from roughly $61 billion in early 2025 to nearly $1 trillion by the IPO, the market value of those instruments grew correspondingly, and the increase appears as a non-cash charge to the income statement. The actual H1 2026 non-cash fair value loss on convertible instruments was $72.388 billion. Strip that out and the operating loss for H1 2026 was $693 million — dramatic improvement from the $8.06 billion operating loss recorded across all of 2025.

Net Revenue Retention — the amount existing customers are spending in later periods relative to their initial spending — stood at approximately 900% as of September 30, 2026, an extraordinary figure even by enterprise software standards, where 110–130% is considered strong. The metric reflects Claude Code's rapid expansion within enterprise accounts, as customers who began with small API experiments have graduated to large-scale integrations. Two customers account for approximately one-fifth of Anthropic's total revenue, a concentration risk the S-1 flags explicitly and that investors will probe during the roadshow.

Approximately 80% of Anthropic's revenue is consumption-based — users pay for tokens processed — with 20% from subscriptions. This model amplifies both growth (a customer who expands usage multiplies revenue without a new sales cycle) and volatility (a customer who reduces usage immediately impacts the top line).

A $26 Billion Strategic Pool Anchors the IPO Structure

Concurrent with the public offering, Anthropic is closing a $26 billion private placement of Class N non-voting shares at the IPO price. The buyers are strategic cloud and AI infrastructure partners whose relationships with Anthropic are already embedded in its compute infrastructure: Google is committing approximately $10 billion, Nvidia approximately $10 billion, AMD approximately $2.5 billion, with Broadcom, Micron, Microsoft, and TSMC accounting for the remainder.

The structure gives Anthropic a capital base substantially beyond the IPO proceeds themselves — the combined event totals roughly $89 billion to $99 billion depending on greenshoe exercise, exceeding SpaceX's June 2026 IPO, which raised approximately $75 billion and was itself the largest IPO in history at the time. The private placement buyers receive economic participation at IPO-equivalent prices without voting rights, leaving the governance architecture intact.

Over $900 Billion in Compute Commitments Explains Why the Capital Is Needed

The combined proceeds are largely destined toward compute. Anthropic has signed take-or-pay cloud agreements that total approximately $900.9 billion, including approximately $114.5 billion with Google Cloud, $110 billion with Amazon Web Services, $44.9 billion with Microsoft Azure, $161.2 billion with Broadcom for custom chip development, and $84.5 billion with SpaceX's Starshield satellite division for inference infrastructure. Of that total, approximately $451.6 billion is expected to come due before December 31, 2030.

Take-or-pay agreements are fixed obligations: Anthropic must pay for contracted capacity whether or not it uses it. These commitments are not aspirational targets but legally binding contracts that require the revenue trajectory to continue. The S-1 identifies this as a material risk. The structure reflects a fundamental constraint of frontier AI development: training runs and inference at scale require GPU and accelerator capacity booked years in advance, and no company at Anthropic's size can operate on spot pricing.

The Valuation Question Calls for Precision

Coverage of the Anthropic IPO has used the phrase "one trillion dollar valuation" loosely, and the distinction matters for investors. At the midpoint of the $70–$78 range ($74), the post-IPO outstanding share count of approximately 13.09 billion shares gives a basic market capitalization of approximately $968 billion — just below $1 trillion. At the top of the range ($78), the basic cap reaches approximately $1.02 trillion.

The fully diluted share count, including unissued options, RSUs, and reserved employee shares, comes to approximately 18 billion shares. At $74 per share, fully diluted, that implies a market capitalization of approximately $1.33 trillion — meaningfully above the basic figure, and the number that matters most for assessing dilution risk over time.

Earlier estimates of a $2 trillion valuation that circulated in the press reflected aspirational targets from Anthropic's bankers in advance of the roadshow, not the range that actual investor book-building produced. Anthropic's bankers reportedly circulated preliminary terms showing 867 million shares at prices up to $80 before revising the terms downward to 825 million shares at $70–$78 as investor feedback came in during October. The difference between $2 trillion and $975 billion is not a rounding error; it illustrates a broader recalibration of AI valuations as companies approach public markets, where institutional investors apply more rigorous earnings-based discipline than private round participants.

SpaceX's June 2026 IPO provides a relevant precedent: the company listed at $1.77 trillion — the largest IPO valuation in history — but fell nearly 50% from its peak of $225.64 in mid-June to approximately $113 by late July 2026, before partially recovering to around $141 by November 2026 as markets repriced its terminal value after initial growth deceleration. Anthropic enters its roadshow knowing that precedent exists.

What the Roadshow Needs to Resolve

The core tension investors will probe over the two-week roadshow period is whether Anthropic's exceptional revenue growth rates are durable. Revenue growing nearly 14-fold year-over-year reflects a base effect — Anthropic's enterprise salesforce and Claude Code adoption curve are still in an early expansion phase — but it cannot compound at that rate indefinitely. The question is what the steady-state growth rate is, and what valuation multiple that rate justifies.

The governance structure — particularly the fixed 50.1% founder voting control and the Long-Term Benefit Trust board architecture — means that public shareholders have limited mechanisms to redirect the company's strategy if they disagree with management's choices. That is a deliberate feature of the PBC + multi-class structure, and investors who buy ANT shares are accepting it explicitly. It places an unusual concentration of consequential judgment in the Amodei founding team at a moment when AI safety decisions have implications well beyond any single company's balance sheet.

If Anthropic prices at or above the midpoint of its range, it will begin trading before Thanksgiving as potentially the highest-valued AI company to have ever listed on a U.S. exchange. The milestone that follows immediately is not the opening price — it is whether the quarterly revenue trajectory sustains the valuation the range implies, and whether the Long-Term Benefit Trust's governance model holds up to the scrutiny that only comes once a company is publicly accountable.