A trillion-dollar test for the AI market
Anthropic has launched the roadshow for its proposed initial public offering, putting hard numbers on a deal that Wall Street has treated for months as a referendum on the economics of frontier AI . The company says the offering consists of 825 million Class A shares: 801,351,354 shares to be sold by Anthropic itself and 23,648,646 shares to be sold by existing stockholders . The expected IPO price range is $70 to $78 per share, and Anthropic has applied to list on the Nasdaq Global Select Market under the ticker symbol ANT .
At the top of the range, the public offering would raise about $64.4 billion before fees, matching Reuters’ calculation from the latest draft prospectus and making the deal one of the most ambitious equity offerings ever attempted by a software company . The valuation implied by the current terms could reach about $1.02 trillion, Reuters reported, a level that is extraordinary in absolute terms but lower than the $1.5 trillion to $2 trillion figures previously discussed around the company . Axios also reported Tuesday that Anthropic is seeking a valuation that could reach $1.02 trillion, with the company aiming to sell just over 801 million primary shares at $70 to $78 each .
That adjustment is the central story of the roadshow. Anthropic is not merely asking investors to pay for past growth; it is asking them to fund a compute-intensive race in which revenue, infrastructure commitments, governance design and AI-risk disclosures all matter at once. The company’s decision to proceed now, after volatility in AI-linked stocks and questions about the durability of the sector’s investment cycle, turns the IPO into a live market test of whether public investors still want pure-play exposure to frontier model developers .
The structure: public shares plus a $26 billion private placement
The headline IPO is only one part of the financing package. In its amended prospectus, Anthropic discloses a concurrent private placement in which Google, Nvidia, AMD, Broadcom, Micron, Microsoft and TSMC Partners have agreed to purchase an aggregate of $26 billion of Class N common stock at the same per-share price as the IPO . The structure means that, assuming the offering closes, Anthropic would combine public-market demand with committed capital from companies that are deeply tied to the AI supply chain.
The private placement is especially important because of who is participating. At the $74 midpoint of the IPO range, Google would purchase about $10 billion of Class N shares and Nvidia would purchase another $10 billion; AMD would purchase about $2.5 billion; Broadcom, Micron and Microsoft would each purchase about $1 billion; and TSMC Partners would purchase about $500 million . Those names are not passive financial backers in any ordinary sense. They represent cloud distribution, chips, memory, networking and manufacturing capacity — the infrastructure stack Anthropic needs to keep training and serving Claude at global scale.
The private placement is scheduled to close immediately after the IPO, but it remains contingent on the completion of the offering and other conditions . Anthropic also makes clear that the Class N shares sold privately will not be registered in the public offering and will be restricted securities . In practical terms, the company is building a two-track financing: a liquid public float for Class A shareholders and a strategic non-public capital raise from suppliers and platform partners.
Anthropic will not receive proceeds from the shares sold by existing stockholders in the IPO . The company’s own proceeds, plus the private-placement proceeds, are intended for general corporate purposes, including working capital, operating expenses, capital expenditures, compute and infrastructure investments, and hiring or retention of highly skilled personnel . That use-of-proceeds language is broad, but in the context of frontier AI it points to a simple reality: the IPO is being sold as growth capital for compute, not as a conventional software-company liquidity event.
Why the valuation matters
A $1.02 trillion target would still place Anthropic among the most valuable technology companies in the world. Yet the number also signals restraint compared with the expectations that surrounded the company earlier this year. Reuters reported that some investors had questioned whether a previously discussed $1.5 trillion valuation was too high, especially given Anthropic’s infrastructure commitments and volatility in AI-related stocks . Reuters also reported that at one point investors expected the IPO to seek a valuation as high as $2 trillion .
This is why the roadshow is about price discovery as much as fundraising. If public investors accept the $70 to $78 range, the transaction would validate a trillion-dollar public valuation for a company whose business depends on rapidly scaling AI model usage. If demand is weak, the offering could force a broader reset in how the market values frontier AI labs, cloud capacity deals and the suppliers that finance the buildout.
The range also changes the optics of the offering. At $2 trillion, Anthropic would have been presented as an almost inevitable platform company. At $1.02 trillion, the pitch is still enormous, but less absolute. It gives investors a discount to the most aggressive private-market expectations while preserving the message that Anthropic belongs in the top tier of AI companies. That compromise may be the point: the company is trying to attract enough institutional demand to complete a record-scale IPO without asking the market to underwrite the most bullish version of the AI story.
Growth, compute and operating leverage
The amended filing gives investors a clearer view of the business behind the valuation. Anthropic reported $33.2 billion in revenue for the nine months ended September 30, 2026, up from $2.5 billion in the same period of 2025 . Over that same nine-month period, the company reported an operating loss of $1.3 billion, down from $5.2 billion a year earlier, and adjusted operating income of $4.1 billion, compared with an adjusted operating loss of $3.1 billion in the prior-year period .
Those figures are the core of the bullish case. Anthropic is arguing that revenue is scaling fast enough to offset at least part of the immense cost of compute. The company says trailing-12-month revenue rose approximately 1,200% from September 30, 2025 to September 30, 2026, while headcount increased approximately 170%, driving trailing-12-month revenue per employee from $1.4 million to more than $6.6 million . It also says more than 90% of contributions to its codebase in September 2026 were written by Claude, a claim meant to show that the company is using its own models to increase internal productivity .
But the filing also shows why investors are cautious. Anthropic reported a $567 million operating loss for the three months ended September 30, 2026, after generating $974 million of operating income in the prior quarter . The company attributed the swing primarily to $5 billion of short-term compute commitments that currently cost significantly more than comparable compute procured under multi-year commitments, along with $2 billion of non-cash charges related to its donation matching program .
That volatility is a reminder that frontier AI does not behave like a mature SaaS business. Demand can rise quickly, but capacity must be bought, reserved or financed ahead of time. Margins may improve when compute is efficiently contracted, then compress when demand spikes and short-term capacity is needed. The roadshow will therefore have to convince investors that Anthropic can turn model demand into durable operating leverage rather than merely cycling between revenue surges and infrastructure bottlenecks.
Governance and control remain part of the pitch
Anthropic’s filing also makes the IPO unusual from a governance perspective. The company warns that its capital structure will concentrate voting power with the Long-Term Benefit Trust and the Founder LLC, and that this voting power will not be derived from or proportionate to economic ownership . The company also says its stock structure may effectively eliminate Class A holders’ ability to influence important decisions, including a change in control .
That governance framework reflects Anthropic’s public-benefit identity and its claim that advanced AI requires long-term oversight. In the filing, Anthropic says it has contributed 846,840,960 Class N shares, representing 5% of fully diluted shares as of June 30, 2026, to the Anthropic Public Benefit Fund . It also describes a framework under which the board may issue up to an additional 2% of then-outstanding shares to the fund or another designee each time the share price appreciates 50% above the previous milestone, beginning with the IPO price .
For investors, that creates a trade-off. The company is offering exposure to one of the most important AI labs in the market, but with governance rights that are deliberately constrained. Buyers of Class A shares are being asked to accept a structure in which mission, safety and long-term benefit are embedded in the capital stack — and in which economic ownership does not automatically translate into control.
The roadshow’s real question
Anthropic’s IPO roadshow begins with unusually clear tension. The company has growth numbers strong enough to justify a blockbuster valuation, strategic investors willing to commit $26 billion alongside the IPO, and a ticker that could become the first major pure-play public benchmark for frontier AI . At the same time, the valuation has come down from the most aggressive expectations, the cost of compute remains unpredictable, and the structure gives public shareholders limited influence over the company’s direction .
That makes the $70 to $78 range more than a pricing detail. It is the market’s first formal answer to a bigger question: how much is Wall Street willing to pay today for the promise that frontier AI will become a trillion-dollar operating business, not just a trillion-dollar capital cycle? If Anthropic prices successfully near the top of the range, it will strengthen the case that public investors still believe in the AI buildout. If it struggles, the roadshow may mark the moment when AI valuations began to meet the discipline of public markets.