AI Trends

Anthropic Plunges After Strong Debut as AI Model Firms Face Public-Market Valuation Test

TMTPost · AGIUpdated Nov 20 12:52

AI summary

Anthropic debuted on Nasdaq under the ticker ANT, opening at $111.50 and closing at $97.17, up 17.07% from its $83 offering price. Drawing on its $33.2 billion in revenue and $567 million operating loss in the first three quarters, as well as approximately $900.9 billion in computing procurement and lease commitments, the article examines how public markets weigh rapid growth, long-term spending, and cash-flow risks. The IPO and concurrent private placement raised approximately $107.3 billion in total, with several companies across the computing supply chain participating as investors.

Why read this

SelectedAI score 87

The article places first-day price performance, revenue growth, and long-term computing commitments within a single framework, highlighting the public market’s valuation dilemma regarding frontier AI model companies.

Topics

Tags

Body · AI translation

On November 19 U.S. Eastern Time, the largest IPO in history officially debuted on Nasdaq. Shares trading under the ticker ANT opened at $111.50, 34.3% above the $83 offering price. They climbed as high as $115.10 intraday, briefly pushing the fully diluted valuation above $2 trillion, and fell as low as $92.72 before closing at $97.17, up 17.07%. An estimated $86 billion to $88 billion in trading volume that day, based on the average transaction price, made the newly listed stock the market leader by individual-stock turnover.

Figure 1: Key ANT price points on its first trading day (Sources: Nasdaq market data and public reports)

The broader market was itself turbulent that day. Jensen Huang had resigned as Nvidia CEO on November 12, and Nvidia had fallen for six consecutive trading sessions. The S&P 500 dropped another 2.02% that day, the Nasdaq Composite declined 2.25%, and AMD fell nearly 8%. ANT gained despite clear pressure on the broader market, demonstrating genuine investor appetite for cutting-edge AI assets. Of course, subsequent data will be needed to determine how much of that demand came from new capital and how much reflected funds reallocated after investors sold other AI stocks. The fact that ANT surrendered roughly half its gains after opening 34% higher also shows that initial enthusiasm had temporarily run ahead of itself, with the secondary market completing a spontaneous price recalibration within a few hours.

A Valuation Revised Again and Again

The $83 price was the result of repeated reductions. In August, the market had circulated a fully diluted valuation of as much as $2 trillion. When the roadshow began in November, the valuation fell to approximately $1.4 trillion based on the top of the range at $78, about 30% below the rumored peak. The final price of $83 implied a valuation of approximately $1.5 trillion.

Figure 2: Convergence of Anthropic’s target valuation (fully diluted basis; Sources: public reports and prospectus)

The reductions were based on financial data disclosed in stages during the roadshow: revenue for the first three quarters totaled $33.2 billion, up 1,206% year over year, while its annualized run rate stood at approximately $81 billion in late October. However, third-quarter sequential growth had slowed to 47% from 144% in the second quarter, and the company again recorded an operating loss of $567 million.

Another set of figures was even more striking: approximately $900.9 billion in computing-power purchase and lease commitments, of which about $70.7 billion must be paid within 12 months.

On one side is extraordinarily rapid revenue growth; on the other is an equally extraordinary level of highly certain expenditure. In essence, the transaction is a bet on uncertain continued growth financed against measurable long-term costs.

Subscription orders totaled approximately $500 billion, including about $400 billion from institutions and $100 billion from retail investors, equivalent to more than six times the size of the public offering. The final price of $83 was only 6.4% above the top of the range, acknowledging the strength of demand without exhausting the potential for a first-day gain. In hindsight, that 6.4% restraint left ample room for the stock to open 34% higher the following day.

Pricing was finalized after the market closed on the 18th, with the first trade on the 19th determined through an opening auction. After ANT rang the bell at 9:30 a.m., it was not until 11:48 a.m. that the first trade was matched at $111.50. The market waited more than two hours, and the opening transaction involved approximately 108 million shares.

From Rejection to the Opening Bell: A Five-Year Journey to Listing

Anthropic was founded in 2021 by former OpenAI research vice president Dario Amodei and others. Established around the principle of “AI safety,” it joined the industry’s top tier in 2023 with its Claude series. Over five years, its private-market valuation climbed from several billion dollars, with Amazon and Google successively making major investments. According to the prospectus, after the IPO and concurrent private placement, Amazon will hold 14.7% of the outstanding share capital and 33.9% of the Class N shares. If all convertible notes are converted into equity, those stakes will rise to 23.8% and 48.2%, respectively, subject to a contractual conversion cap.

The offering brought together nearly every major upstream computing-power supplier: the public sale of 980 million shares raised approximately $81.3 billion, while entities affiliated with AMD, Broadcom, Google, Micron, Microsoft, Nvidia and TSMC subscribed at the same price for $26 billion of nonvoting Class N common shares, including $10 billion each from Google and Nvidia. The IPO and private placement raised approximately $107.3 billion in total, surpassing SpaceX’s $75 billion offering record. Goldman Sachs and Morgan Stanley led the underwriting, with 22 banks participating.

Figure 3: Comparison of funds raised in the Anthropic and SpaceX IPOs (offering basis, excluding overallotments; Source: public disclosures)

The subscriber list reads more like an industry-chain declaration: upstream computing-power suppliers are using real money to bind themselves to a downstream model company, creating a closed loop of “investment, procurement and reinvestment.”

For the issuer, this effectively locked in a large portion of the offering and reduced its risk in advance. For subscribers, the equity stake and purchase orders provide mutual protection. The other side of the coin, however, is deep cross-ownership throughout the industry chain and an intertwining of revenue and investment. Public-market investors must decide for themselves whether this mutual support reflects genuine end demand or circular credit. It is a question the market will continue asking through quarterly data now that ANT is publicly traded.

Is $83 Expensive? Valuation Logic and Industry Benchmarks

The revenue curve provided the first pillar of support for the $83 price. Revenue rose from $41 million in 2023 to $33.2 billion in the first three quarters of 2026, with just those nine months generating more than seven times the total for all of 2025. Based on an annualized run rate of $81 billion, the valuation-to-revenue multiple was approximately 18 times, far above that of traditional software companies. The market was clearly not paying for current profits, but assigning a value to the idea of “models as infrastructure,” with leading laboratories ultimately coming to resemble cloud providers. That is also where the controversy lies: whether Anthropic can secure its platform position depends on its ability to sustain a leading edge in the next generation of models, something that financial statements cannot reveal.

Figure 4: Anthropic’s revenue growth (Source: Prospectus)

The absence of peers magnifies its significance. OpenAI is not planning an IPO until 2027 at the earliest, while peers such as xAI and Mistral remain privately held. Over the next year or two, ANT will be virtually the only way for public-market investors to gain exposure to a frontier large-model company.

There is no need to rush to label the $83 price itself as expensive or cheap. It is better viewed as the public market’s first benchmark valuation for the frontier AI industry, against which the financing and IPO pricing of peers will inevitably be compared.

Sell-side estimates corroborate this range from another angle: Anthropic’s EBITDA could reach $183 billion in 2030, implying a multiple of approximately 8 times at $83, compared with around 4 times for SpaceX on the same basis. The institution also cautioned that its approximately 13-times figure for the S&P 500 is based on the next 12 months and should not be directly compared with a forward multiple for 2030. It assigned a buy rating and a $105 price target. Looking back at the first day of trading, the intraday high of $115 exceeded that target, while the $97.17 close landed about 8% below it. In a single day, the market found a middle ground between an optimistic narrative and a cautious valuation.

Cracks can also be found in the numbers. Revenue dependence on major customers and cloud channels will need to be verified in quarterly reports. After computing costs, approximately $7.1 billion remained from third-quarter revenue on a non-GAAP basis, but the company still posted a $567 million loss on an accounting basis. Compute gross margin fell from 52% in the second quarter to 42%, making the trajectory of unit economics more important to watch than total revenue. As for the gap on the spending side, the chart makes it even clearer:

Figure 5: Comparison of compute procurement commitments and contracted customer orders (Source: Prospectus)

Approximately $900.9 billion in procurement commitments extending to around 2045 stands against only about $13.8 billion in contracted customer revenue commitments, primarily short-duration RPO. The amounts and time horizons are both mismatched. If growth continues to slow or the financing environment tightens, the company will have to reassess the balance between the pace of expansion and cash-flow security. Buying at $83 does not provide certainty, but rather a conditional growth option—the condition being that neither growth nor cash flow can lose momentum.

After the Listing: Threefold Changes for the Company, the Industry, and the Market

The most immediate change is to Anthropic itself: the listing first resolves its need for “ammunition.”

The roughly $100 billion raised will be used mainly to fulfill compute commitments and fund model research and development, easing short-term concerns about the funding chain. The company’s status has also changed: from now on, it must face the market through quarterly reports, and its $900.9 billion in procurement commitments, continuing net losses, and the pace at which stock-based compensation expenses are recognized will all be scrutinized every quarter. Private markets can tolerate a decade without profits; public markets are far less patient. For the first time, financial discipline has become a hard constraint hanging over a leading AI laboratory.

Looking beyond the company, ANT’s listing fills the most critical missing link in the AI capital market.

Over the past three years, exits from AI investments have depended almost entirely on valuations in the next private funding round. The listing opens that pathway to the public market for the first time, giving early investors and employees a genuine liquidity exit, while the 17% first-day gain brought this round of exits to a respectable conclusion. OpenAI’s listing timetable and the fundraising narratives of companies such as xAI will all take this market response into account, opening an IPO window for a group of AI companies. Talent competition and governance are also changing: monetizable equity has become a bargaining chip in the battle with OpenAI for talent, while the arrangement under which Class N shares carry no voting rights preserves the founding team’s control. But concentrated control combined with quarterly earnings pressure presents an entirely new challenge. For the first time, Dario Amodei will have to balance long-term investment against short-term returns under the gaze of public shareholders. The petition signed by more than 80 people on the eve of the listing questioning the company’s safety commitments is also a reminder that the public market will scrutinize the halo surrounding a company built on safety.

In recent months, divisions in the U.S. stock market over the AI sector have widened markedly: one camp believes capital spending on compute is creating a bubble, while the other believes revenue is materializing faster than expected. The orderly absorption of an offering worth roughly $100 billion on a day when the broader market fell more than 2% amounted to a real-world stress test of the market’s capacity to take up the shares. Given its size, ANT will most likely become a core holding for AI-themed funds, and its share-price performance will replace rumors about private-market valuations as the industry’s daily sentiment gauge.

What to Watch After the First Day

After the excitement subsides, three things merit more attention than the stock’s movement on any single day: short-term capital needs time to settle, and true price discovery must await the completion of fund flows related to lockup expirations and index inclusion—expected on December 1 for MSCI and around December 21 for the Nasdaq-100; whether the 147 million-share greenshoe option is exercised in full will determine the underwriting syndicate’s room to support the stock and the final amount raised; and whether OpenAI brings forward its listing and companies such as xAI adjust their fundraising narratives to reflect ANT’s multiple will determine whether November 19 is remembered as the beginning of the public-market era for AI assets or merely another lively interlude.

On every trading day from now on, the debate over AI valuations will play out publicly on the Nasdaq trading screen rather than solely in private-funding conference rooms. For the industry, that may be the real incremental gain delivered by this record-breaking IPO. (First published by TMTPost APP; Author | Silicon Valley Tech-news; Editor | Jiao Yan)

Source

TMTPost · AGI