Anthropic Stock Analysis 2026: Revenue, Valuation and the Risks Behind the IPO

The headline number in Anthropic’s prospectus is its revenue growth. The more important investment question is what it costs to deliver that growth—and what rights investors receive in return. Anthropic’s proposed IPO brings both issues into view for the first time through detailed public financial disclosures. This article examines the business rather than the mechanics of subscribing to its shares.
Anthropic stock: what investors are buying
Anthropic, PBC develops AI models and services marketed under the Claude brand. Revenue comes largely from model usage by customers and a smaller subscription stream. Investors considering Class A shares are assessing a capital-intensive technology business, not a conventional software company whose sales automatically translate into high margins.
The numbers in the S-1: growth and cost
Metric (USD millions) | FY 2025 | 9m 2025 | 9m 2026 |
Revenue | 4,585 | 2,545 | 33,242 |
Compute & infrastructure | 7,326 | 4,381 | 18,808 |
Operating loss | (8,060) | (5,224) | (1,260) |
Consumption-based revenue | 3,796 | 2,125 | 26,306 |
Subscription revenue | 789 | 420 | 6,936 |
The 2026 figures cover the nine months ended 30 September, while 2025 full-year figures cover 12 months. They should not be compared as if they describe identical periods. More useful is the disclosed comparison of the first nine months of 2025 and 2026: revenue rose from $2.545 billion to $33.242 billion, while compute and infrastructure expenses rose from $4.381 billion to $18.808 billion.
Compute costs are a business variable, not a footnote
For the first nine months of 2026, compute and infrastructure spending amounted to about 57% of reported revenue ($18.808 billion divided by $33.242 billion, rounded). In the comparable nine months of 2025, the amount exceeded revenue. This is an arithmetic ratio of disclosed line items, not a gross-margin measure or a forecast of future spending.
This distinction matters because AI model developers may face large upfront infrastructure commitments, expensive inference demand and competitive pressure to cut prices. Improved model efficiency can support margins, but more customers can also require more capacity. Both forces can operate simultaneously.
Why operating loss and net loss must not be confused
Anthropic reported an operating loss of about $1.260 billion for the first nine months of 2026, versus about $5.224 billion for the comparable period of 2025. The filing separately reports other items affecting net results. It would be misleading to present headline net loss alone as proof that customers or core operations are unprofitable. Readers should examine the full financial statements, stock-based compensation and transaction-related items.
What could the IPO valuation mean?
The preliminary US range of $70–$78 is a proposed per-share offer price, not a direct statement about annual earnings. An equity valuation requires a consistent share count, including expected conversions, share classes and potential dilution. The prospectus contains these assumptions; taking the share price and multiplying it by an outdated private-company share count can produce a misleading number.
For investors, the useful question is not simply whether an AI IPO is “expensive”. It is which level of future revenue growth, operating margin and capital efficiency would be needed to support the market value once a final offer price and fully diluted capitalisation are known.
For the offering timetable and access details, see our Anthropic IPO guide for investors in Germany
A valuation checklist rather than a price target
Factor | Question to examine | Why it matters |
Growth | Can enterprise demand stay strong? | Revenue assumptions |
Compute efficiency | Can cost per unit of output decline? | Operating margins |
Dilution | What is the fully diluted share count? | Per-share valuation |
Governance | What rights attach to Class A? | Shareholder influence |
Cash and capital | How are proceeds invested? | Funding needs and returns |
These are analytical questions, not an undisclosed valuation model. No share-price target is calculated here because the final IPO price, trading price and updated dilution assumptions had not been confirmed as of 13 November 2026.
How Anthropic makes money—and where margins can change
Usage-based revenue reflects customer consumption of models; subscription revenue provides another channel.The prospectus reports $26.306 billion in consumption-based revenue and $6.936 billion in subscription-based revenue in the nine months ended 30 September 2026. This implies roughly four-fifths of reported revenue was consumption-based over that period.
A high usage-based share can be attractive when customer adoption accelerates. Yet it also raises analytical questions about customer concentration, pricing per unit of compute, usage retention, and whether costs rise with demand. Enterprise buyers may reduce usage or switch models if comparable capabilities become cheaper.
Public-benefit status and shareholder rights
Anthropic is organised as a Delaware public benefit corporation, with a stated objective involving responsible advanced AI development. Theproposed governance includes several share classes and mechanisms involving a Long-Term Benefit Trust and founder-related voting arrangements. Investors should not assume Class A shareholders exercise the same control as founders or special classes on every matter.
This arrangement could protect long-term priorities from short-term pressure, while also limiting how public shareholders can influence certain decisions. Neither outcome is automatically good or bad for returns. The key issue is the exact rights attached to the security being bought.
Risks that matter beyond the IPO headlines
Competition and customer pricing
Model quality, reliability, distribution and cost are all competitive factors. New models may narrow differentiation quickly, and clients can negotiate pricing as the market becomes more crowded.
Infrastructure and supply dependence
Compute capacity, chips, networking, energy and cloud arrangements can materially affect service economics. Strategic supply relationships do not guarantee cheaper infrastructure forever or sustained profitability.
Regulatory, safety and legal exposure
AI safety decisions and regulatory requirements can affect development timelines, spending and market access. The company’s stated mission and governance model should be read alongside its risk disclosures rather than treated as a valuation premium.
Dilution and share supply
IPO issuance, concurrent private placement and future releases of previously restricted shares can change the share count or tradable supply. The filing’s lock-up provisions contain exceptions and different expiry dates. Investors should consult the final prospectus rather than relying on a simplistic six-month assumption.
What should investors watch after listing?
Post-listing updates to enterprise demand, compute expenses, operating margin, share-based compensation and share-class disclosures will matter more to a long-term appraisal than the first day’s trading spike or drop. A comparison with listed AI infrastructure companies also requires care: a model developer and a chip supplier operate different businesses and have different cost structures.
Sources and methodology
European retail prospectus dated 11 Nov 2026
Financial figures are reported in USD millions unless stated otherwise; calculations are labelled as editorial arithmetic and are not forecasts. Research cut-off 13 November 2026.
Is Anthropic profitable?
Its S-1 discloses an operating loss for the first nine months of 2026. Profitability must be assessed from the relevant accounting measure and period; an adjusted non-GAAP figure is not interchangeable with GAAP profit.
What was Anthropic revenue in 2026?
The prospectus reports $33.242 billion of revenue for the nine months ended 30 September 2026, not for the full calendar year.
Does the IPO make Anthropic comparable to NVIDIA?
No. Anthropic sells AI models and services; NVIDIA supplies accelerated-computing platforms and technology. Demand for AI may affect both, but their economics and stock-price drivers differ.
Disclaimer: The content provided is for informational purposes only and does not constitute investment advice, a personal recommendation, or an invitation or inducement to buy or sell financial instruments. It has been prepared without consideration of your personal investment objectives or financial situation. No representation or warranty is made regarding the accuracy or completeness of the information. References to past performance or future projections are not reliable indicators of future performance.





