Anthropic IPO: Claude, Explosive Growth and the Valuation Test

Executive summary

Anthropic has made its stock-market plans concrete. On October 26, 2026, the company behind Claude publicly filed its S‑1 registration statement, proposing a Nasdaq listing under ANT. A filing is not a completed IPO: the preliminary prospectus reviewed here leaves the offer price, share count and final offering valuation open.

The investment question goes beyond whether Claude succeeds. Can Anthropic turn exceptional adoption into durable, profitable growth—and how much of that future will a new shareholder already be paying for?

As of November 2, 2026. Financial figures refer to the periods disclosed in the S‑1; first-half 2026 financial statements are unaudited. All monetary amounts are in US dollars. Valuation illustrations are not confirmed IPO terms.

1. What is confirmed about the Anthropic IPO?

Anthropic confidentially submitted a draft registration statement on June 1 and publicly filed on October 26. The proposed offering includes Class A shares issued by the company and shares sold by existing shareholders. Anthropic will not receive proceeds from those existing shareholders’ sales.

That distinction matters. Primary issuance finances the business, including computing capacity, infrastructure and personnel. Secondary sales give existing owners liquidity. Investors need to examine the split, not simply the headline deal size.

Registration does not mean the securities regulator endorses the investment or guarantees completion. Effectiveness, pricing and the start of trading are separate milestones. ANT is the proposed symbol, not evidence that the stock is already trading in the public market.

2. The business model: intelligence sold by usage

Anthropic is not simply selling a chatbot. Claude reaches customers through its own applications, developer access and cloud distribution. Claude Code extends its role in software development and agentic work. Subscription plans complement consumption-based revenue.

In the first half of 2026, consumption-based revenue was approximately $13.1 billion, or 80% of total revenue. Customers predominantly pay for model usage metered in tokens. Increased usage can lift revenue rapidly; reduced usage can feed through just as directly.

This differs from a traditional software licence with almost costless incremental users. Substantial queries require computing resources. Revenue and cost per completed task therefore matter more than user counts alone. More efficient models can improve economics if they deliver the same customer value with less compute, but competitive price cuts may transfer those gains to customers.

3. The financial trajectory: extraordinary growth and real operating improvement

Reported revenue rose from $386 million in 2024 to $4.585 billion in 2025. First-half 2026 revenue reached $16.286 billion, compared with $1.163 billion a year earlier—approximately 1,300% growth.

Growth accelerated within the half: first-quarter revenue of $4.731 billion was followed by $11.555 billion in the second quarter. The first-half GAAP operating loss was $693 million, implying a calculated operating margin of approximately negative 4.3%. The second quarter alone, however, generated $974 million of GAAP operating income.

These distinctions prevent two misleading shortcuts. Anthropic cannot accurately be described simply as hopelessly unprofitable, but one profitable quarter does not establish durable profitability either. The disclosed trajectory shows operating leverage. Subsequent quarters must demonstrate how resilient it is.

Compute and infrastructure expense was $8.945 billion in the half, approximately 55% of revenue. That expense ratio is not a conventional gross margin. The prospectus presents expenses differently from many software companies, and this category is broader than the cost of serving an individual customer request.

4. Why an $80.748 billion net loss is not $80.748 billion of cash burn

The most striking headline is the $80.748 billion first-half GAAP net loss. Treating it as an equivalent cash outflow would confuse accounting earnings with cash movements.

$72.388 billion came from changes in the fair value of convertible notes, derivative liabilities and SAFEs. As the company’s valuation increases, related financing claims can be remeasured upward, generating substantial accounting losses. Those charges are different from recurring server bills or payroll.

The opposite shortcut—“it is all accounting, so it does not matter”—is also wrong. These instruments shape conversion terms, the capital structure and shareholders’ economic claims. Other charges also affected earnings. The post-IPO structure needs to be assessed using the final offering documents.

Anthropic reported $1.618 billion of adjusted operating income for the half, versus the $693 million GAAP operating loss. Its reconciliation principally adds back $2.185 billion of stock-based compensation and $126 million of non-cash charitable contributions. Equity compensation preserves cash in the short term, but it is not free to shareholders because of potential dilution.

5. Cash flow and the balance sheet: ample liquidity, not a blanket all-clear

First-half operating cash flow was positive at $3.929 billion. Subtracting $482 million of property and equipment purchases produces a simple operating-cash-flow-minus-capex measure of approximately $3.447 billion. This is a derived metric, not proof of permanently distributable cash.

Working capital, customer prepayments, supplier payment terms and the timing of compute payments can materially influence operating cash flow. A single half-year’s property and equipment purchases also do not capture every economic infrastructure commitment.

The $17.181 billion investing cash outflow should not be labelled entirely as infrastructure consumption. A major driver was purchases of short-term investments, partly offset by sales and maturities. Moving cash into financial securities is not the same as burning it in operations.

At June 30, the balance sheet showed $131.651 billion of cash, cash equivalents and short-term investments. Financing cash inflows during the half were $110.774 billion. The liquidity cushion therefore depends heavily on capital raised, not simply accumulated business profits. Liabilities, preferred claims and conversions also make a mechanical “market value minus all cash” valuation bridge unreliable without a complete capital-structure review.

6. Understanding 900% net revenue retention

Anthropic reported approximately 900% net revenue retention as of September 30. The metric compares revenue from qualifying existing customers in the latest three-month period with their revenue in the corresponding period a year earlier. Claude Pro and Max subscribers are excluded.

That signals extraordinary expansion within the measured customer cohort. It does not mean 900% more customers or a 900% probability of renewal. It is not contractually guaranteed future growth. Small initial spending levels, new use cases and early adoption can help produce unusually high expansion rates.

A durable moat would require Claude to remain embedded in important workflows, customers to keep expanding despite alternatives, and unit economics to improve simultaneously. A single benchmark win does not establish that moat.

7. Valuation: a great product can become an expensive stock

Anthropic’s May financing raised $65 billion at a $965 billion post-money valuation. That private-round valuation is neither a confirmed IPO market capitalisation nor necessarily directly comparable with the future value of listed common stock. Investor rights and the capital structure matter.

For a transparent scale comparison, multiply reported second-quarter revenue of $11.555 billion by four: $46.22 billion of annualised quarterly revenue. This is not a full-year forecast or contracted annual revenue. In a rapidly growing business it may understate the latest business scale, but it makes the denominator explicit.

Using that denominator gives the following rough equity-value-to-revenue ratios:

• $965 billion valuation reference: approximately 20.9 times.
• Hypothetical $1.5 trillion equity value: approximately 32.5 times.
• Hypothetical $2 trillion equity value: approximately 43.3 times.

The last two are sensitivity illustrations, not confirmed offering prices. These are not enterprise-value-to-revenue multiples: no adjustment has been made for cash, liabilities or conversions.

The arithmetic highlights the expectations hurdle. A $1.5 trillion equity value would require $75 billion of annual revenue to represent a still-high 20-times revenue multiple; $2 trillion would require $100 billion. These are calculations, not revenue forecasts. Revenue alone also does not deliver shareholder returns: ultimately, the business needs cash flow per share.

A defensible per-share price target cannot be extracted from this preliminary prospectus while key offering parameters remain blank. Attaching an exact number would add presentation polish, not analytical credibility.

8. Competition and partnerships: distribution benefits with competing interests

Anthropic competes with other frontier-model developers, large technology platforms and open-weight alternatives. Customers can allocate tasks across providers. Switching is not always technically simple, but a model provider does not automatically enjoy the entrenched switching costs of a deeply integrated legacy system.

Cloud partnerships provide distribution and access to compute. The same partners can also be investors, suppliers and competitors. Investors should separate genuine end-customer demand from partnership economics: who pays, who takes a distribution share, and which costs are contractually fixed?

The prospectus identifies a crucial mismatch. Much of customer revenue depends on consumption, while compute contracts can involve multi-year minimum purchases or take-or-pay obligations. The cost base does not necessarily shrink when customers become more cautious.

9. Governance: owning shares does not mean controlling the company

Anthropic is a public benefit corporation. It must balance financial interests with its specified public-benefit purpose and other affected interests. Through Class T shares, the Long-Term Benefit Trust generally has authority to elect and remove directors who will constitute a majority of the authorised board.

Separately, a Class F share held through the co-founders’ entity is intended to carry 50.1% of the general voting power, with special treatment for Class T directors. These mechanisms should not be added together as if they were identical voting rights; they govern different decisions.

New Class A investors will have limited influence. This arrangement may protect long-term research and safety decisions from quarterly pressure. It can also support decisions that do not maximise public shareholders’ near-term returns. That belongs in the valuation discussion, not merely the footnotes.

10. The main risks

Technology and pricing: Model advantages can narrow. Lower prices and cheaper open alternatives may reduce revenue per task and pressure margins.

Capacity and fixed commitments: Insufficient compute can constrain growth; excess contracted capacity can become a burden if demand disappoints. Efficiency improvements do not automatically solve this timing problem.

Legal exposure and safety: Copyright, data use, privacy, cyberattacks and harmful autonomous-system behaviour can generate costs, liability and operating restrictions. The 2025 accounts already included a $1.5 billion Bartz settlement expense. That demonstrates the economic relevance of legal risks; it does not establish that the same charge will recur.

Dilution and share supply: Further financing, employee awards, conversions and sales by existing owners influence value per share. Specific post-IPO selling restrictions matter more than assuming a standard lock-up period.

Valuation and market conditions: Strong business growth can coexist with a falling share price if investors accept lower multiples or previously expected even better results.

11. Bull, base and bear cases—without invented probabilities

Bull: Claude becomes a core enterprise work layer. Existing customers expand usage, model efficiency improves economics, and Anthropic maintains a frontier technical position. Revenue growth and cash flow per share support a sustained valuation premium.

Base: The market keeps growing, but extraordinary early growth normalises. Competition requires continuous investment and pricing discipline. Shareholder returns depend increasingly on entry valuation, dilution and margin progress rather than enthusiasm for the product.

Bear: Revenue growth slows faster than infrastructure commitments can adjust. Customers allocate more usage to cheaper alternatives, and further financing becomes necessary despite large historical funding inflows. Multiple compression adds to the pressure on equity value.

No numerical probabilities or price targets are attached. A sufficiently robust forecast path and the final share structure are not yet established in the material reviewed.

12. What investors should check next

• Final price range, issue price and fully diluted share count after conversions.
• Primary versus secondary shares and actual net proceeds received by Anthropic.
• Newer quarterly revenue, GAAP operating performance and non-GAAP reconciliations.
• Operating cash flow, working-capital effects, capital expenditure and separately disclosed infrastructure commitments.
• Customer concentration, direct demand and existing-customer expansion trends.
• Cost per completed task, pricing power and reliability in enterprise deployment.
• Stock-based compensation, other capital rights and the precise schedule permitting insider sales.

SmartDeer’s assessment

Anthropic’s filing presents more than an AI narrative: exceptional revenue growth and visible operating improvement are real features of the disclosed figures. Financing structure, compute economics and governance are also more complicated than a quick interaction with a chatbot might suggest.

The relevant question is not whether Claude is impressive. It is whether the price paid for future cash flow per share leaves sufficient room for growth, margins or dilution to develop less favourably than hoped. A compelling company is not a compelling stock at every price.

Not investment advice. This article is for information and analysis only. Scenarios and valuation calculations are not forecasts or promises of returns.