AI Trends

Anthropic Surges 36.3% on First Day of Trading, with Long-Term Computing Expenditure Commitments of Approximately $900.9 Billion

EastmoneyUpdated Nov 20 01:49

AI summary

Anthropic debuted on Nasdaq on November 19. As of publication, its shares were up approximately 36.3% from the $83 offering price, giving it a publicly traded market capitalization of about $1.48 trillion. Its prospectus shows long-term computing and infrastructure expenditure commitments of approximately $900.9 billion, compared with $13.8 billion in customer payment commitments. Future growth, underlying gross margins, and competitive pressure facing closed-source models have become key areas of market focus.

Why read this

SelectedAI score 95

The article places the IPO valuation, growth requirements, and long-term computing commitments within the same framework, making it easier to understand the cash flow pressures behind Anthropic's high valuation.

Topics

Tags

Body · AI translation

On November 19, against the backdrop of volatility in U.S. AI stocks, controversy surrounding Nvidia, and a growing anti-data-center movement in the United States, Anthropic officially debuted on Nasdaq, becoming the first publicly traded pure-play AI company in the U.S. market.

As of publication, Anthropic shares were trading at $113.13, about 36.3% above the $83 offering price, giving the company a free-float market capitalization of approximately $1.48 trillion and a fully diluted valuation of about $2.04 trillion.

Image source: Google Finance

But Anthropic’s listing is about more than the company itself.

The company is deeply intertwined with a host of technology giants, including Amazon, Google, Microsoft, Nvidia, SpaceX, AMD, and Micron, and its development will shape the future of the entire U.S. AI industry.

Going forward, the market will scrutinize its $90 billion in committed computing expenditures, whether the commercial model for closed-source models can sustain rapid growth or even achieve long-term profitability, its actual gross margin, and pressure from competitors.

These factors will determine whether Anthropic’s listing marks a new beginning for the AI boom or the start of the bubble’s collapse.

Before trading began, protesters gathered outside Nasdaq. One organization said Anthropic was “inflating the AI bubble in an attempt to make a quick profit before it bursts.”

How Long Can Rapid Growth Continue? The Company May Already Be at Its Growth Peak Upon Listing

All of the top 10 companies on the U.S. Fortune list are high-value Anthropic customers. More than half of the U.S. Fortune 500 and over 30% of the Global 2000 have also become high-value customers.

The rapid growth in customer numbers and spending has also driven a marked improvement in Anthropic’s financial performance.

Anthropic’s revenue increased from $41 million in 2023 to $4.585 billion in 2025, then climbed further to $33.242 billion in the first three quarters of 2026, up 1,206% year over year.

During the same period, the company’s operating loss narrowed to $1.260 billion, while cash flow from operating activities turned positive at $6.569 billion.

The proceeds raised through Anthropic’s listing will primarily be used to supplement working capital, cover operating expenses, expand investment in computing capacity and infrastructure, and attract and retain skilled professionals. The company may also use part of the funds for future corporate acquisitions and technology investments.

Anthropic’s ownership and governance structure is equally distinctive.

Anthropic was jointly founded by seven co-founders. Dario Amodei serves as CEO, while his sister, Daniela Amodei, serves as president and chair. The remaining founders are responsible for areas including computing capacity, scientific research, technical architecture, public benefit, and interpretability research.

Based on the fully diluted IPO valuation and information in the prospectus, the seven founders collectively hold an approximately 9.3% stake, corresponding to a market value of about $139.3 billion.

Through Founder LLC, which controls 50.1% of the voting power, and their individual shareholdings, the seven founders collectively control approximately 61.3% of the voting power on matters submitted to common stockholders.

Michael Ewens, a finance professor and co-director of the Private Equity Program at Columbia Business School, believes the timing of Anthropic’s listing coincides with the point at which its operating margin turned positive. Anthropic hopes to position itself as a company that has just become profitable while continuing to grow rapidly, but companies “often go public at the peak of their growth.”

So what level of growth does Anthropic need to achieve to justify its current valuation?

Shivaram Rajgopal, a professor at Columbia Business School, estimates that if Anthropic reaches a valuation of $2 trillion, its relevant profit metrics would need to maintain a compound annual growth rate of 36% to 76% over the next five years, depending on the profit measure used, to provide investors with their expected returns.

A reporter from National Business Daily, hereinafter referred to as the NBD reporter, recalculated the figures using Rajgopal’s model. If Anthropic is to sustain a valuation of approximately $1.5 trillion, its relevant profit metrics would need to maintain a compound annual growth rate of roughly 28% to 66% over the next five years.

However, citing historical research, Rajgopal noted that between 1950 and 2015, not a single company with annual revenue exceeding $50 billion achieved compound annual profit growth of more than 35% for five consecutive years.

$90.09 Billion in Computing Commitments Tie the Fortunes of Technology Giants Together: Anthropic’s Growth Becomes a Shared Bet Across the Entire AI Supply Chain

Anthropic has established long-term investment, commercial, and even competitive relationships with numerous chip, cloud-computing, and computing-capacity giants. Anthropic’s future is intertwined with theirs.

For example, Google has not only invested in Anthropic but also provides it with TPU computing capacity through Google Cloud and sells Claude to enterprise customers via its own cloud platform. At the same time, however, Google continues to develop its Gemini family of large models, which compete directly with Claude.

Microsoft provides Anthropic with Azure computing capacity and sales channels on the one hand, while developing its own AI products and supporting other model developers on the other.

Michael Ewens said he had never before seen an industry form such close and complex relationships of shared interest in such a short period of time.

Of these, more than $900.9 billion in computing expenditure commitments extending through 2045 will directly affect the future performance of major tech giants.

The prospectus shows that, under related agreements newly signed as of and after the end of September 2026, Anthropic’s long-term purchase, lease and other expenditure commitments involving computing capacity and infrastructure totaled approximately $900.9 billion.

However, Michael Ewens described the long-term computing commitments as debt. Many of them are legally binding and require payment for years of computing capacity, and expenditure obligations of such magnitude are uncommon among newly listed companies.

Rajgopal noted that most contracts require Anthropic to pay regardless of how much computing capacity it actually uses. By contrast, customers have committed to paying Anthropic only $13.8 billion. He suggested that investors focus on one metric: revenue per dollar of computing cost.

That figure fell from $2.08 in the second quarter to $1.72 in the third quarter, below the approximately $1.82 that Rajgopal estimates is needed to maintain margins.

Oren Etzioni, head of the Allen Institute for AI in the United States, told the National Business Daily that Anthropic’s computing procurement contracts are fixed, but its revenue is not. To cover these rigid payment commitments, its revenue would need to reach roughly twice its current level. Even if growth slows for just one year, those bills will still have to be paid.

Open-source models are also becoming an important factor affecting Anthropic’s business model.

Anthropic acknowledged in its prospectus that open-source models could gain market share in certain applications, put pressure on product pricing and even prompt customers to adopt other models as substitutes.

Oren Etzioni told the National Business Daily that open-source models are already capable of meeting the needs of many use cases. Anthropic’s advantage rests solely on Claude continuing to lead in highly challenging tasks; what enterprise customers pay for is reliability, security assurances and supporting services.

Anthropic’s current model revenue is highly dependent on continuous technological iteration.

The prospectus disclosed that model versions launched during the six months ended September 2026 contributed more than 80% of the company’s third-quarter revenue.

R&D spending increased from $1.9 billion in the first half of 2025 to $9.7 billion in the first half of 2026. Rajgopal said R&D costs keep rising, while the returns from new models can dissipate entirely within a matter of months.

The company also estimated that approximately 5% to 7% of its revenue in the first three quarters of 2026 came from customers who were later found to have violated its usage policies or commercial terms. As the relevant accounts are restricted, this portion of revenue may not be sustainable.

Its true gross margin remains a mystery: Is Anthropic’s listing a new beginning for the AI boom, or the start of the bubble bursting?

Uncertainty over growth, enormous spending commitments and business risks have left Wall Street wondering: Will Anthropic’s listing reignite enthusiasm for AI investment, or mark the beginning of the bubble’s collapse?

Optimists see Anthropic’s successful offering as a positive signal in itself.

Dan Ives, partner and senior managing director at Yorkville Ives & Co., told the National Business Daily that, with the Nvidia incident casting a shadow over the entire technology sector, Anthropic’s IPO is undoubtedly a major short-term positive that will help restore investor confidence in the growth prospects of the AI revolution.

Jim Handy, a prominent U.S. semiconductor analyst and general manager of the independent market research and consulting firm Objective Analysis, also told the National Business Daily that if Anthropic successfully goes public, investors will most likely view it as a signal of confidence, potentially sparking a rebound.

Ben Preston, head of Orbis Global Equity Strategy, said the fund he manages would not participate in the IPO. In his view, paying a valuation of approximately $1.5 trillion for a company whose long-term profit model has yet to be fully validated is too expensive.

Oren Etzioni, meanwhile, told the National Business Daily that a strong first-day performance by Anthropic would demonstrate market demand for leading AI companies, but a single scarce investment target would be unlikely to lift the entire sector. He made a rough calculation: If Anthropic achieves its 2028 targets of $195 billion in revenue and a 36% profit margin, applying a multiple of 25 would imply a market capitalization of $1.75 trillion. If the revenue forecast were cut in half, its market capitalization would be less than $900 billion. Investors will next use this logic to value every AI company.

Going forward, Anthropic’s gross margin will be a key metric for the market.

Beginning in 2025, Anthropic changed the way it presents its income statement. It no longer reports operating costs and gross margin separately, instead classifying expenditures by nature, including computing capacity and infrastructure, administrative expenses, marketing and other operating expenses.

Rajgopal believes this change overstates Anthropic’s gross margin because the computing costs of serving free users are recorded as marketing expenses.

Anthropic’s true gross margin remains a mystery.

(Source: National Business Daily) Embedded advertisement in the middle of the article

Source

Eastmoney