Anthropic is reportedly preparing to tell potential IPO investors that its total addressable market exceeds $30 trillion. The claim was first reported by The Wall Street Journal and then summarised by Reuters. It is a striking number, but its meaning is narrower than a headline may suggest.
A total addressable market is not present revenue. It is not a valuation. It is not a prediction that a company will capture a stated share of an economy. It is an estimate of the annual revenue opportunity available if a product or service achieved 100% of the relevant market. In this case, Reuters says Anthropic is considering the full scope of work that could be completed with AI models.
That is a big framing decision. It treats the opportunity as far broader than a single software market, a subscription category or a traditional enterprise-technology segment. It also makes the figure much harder to interpret without understanding the assumptions beneath it.
What a $30 trillion TAM does and does not say
The reported $30 trillion figure is a statement about the outer boundary of an opportunity. It says nothing, on its own, about how much of that opportunity is reachable, how quickly it might be reached, which competitors share it, or what a company can charge for a given capability.
Alex Brunicki, co-founder and general partner at Backed VC, described the logic to Fortune: “With things like Claude and the way it writes code, you could argue it’s replacing the work that humans do end-to-end, and so the TAM for those products is essentially the labor market for that work output.”
Brunicki is describing an interpretation of potential opportunity, not Anthropic’s realised revenue or a settled valuation method. A large addressable-market estimate depends on assumptions about what models can do, the sectors that will adopt them, the alternatives available and the value a provider can retain. The Wall Street Journal also notes the uncertainty in estimates that anticipate rapid AI-driven industry change.
The reported investor narrative is built around future scale
The TAM report sits alongside other source-attributed financial projections. Reuters reported on 15 August that two people familiar with Anthropic’s financials said the company was projecting 2028 revenue of roughly $190–200 billion. Reuters also reported that investment bankers and investors were using enterprise-value-to-revenue multiples based on forecasts as they considered the company’s prospective IPO valuation.
Reuters described looking two years ahead as less typical, though it said the approach reflected the speed of Anthropic’s reported expansion and the difficulty of valuing an AI business still investing heavily in computing capacity, model training, inference and hiring. Anthropic did not immediately respond to Reuters’s request for comment.
The distinction between fact, report and forecast matters here.
• The reported TAM is an estimate of a theoretical annual opportunity at full market capture.
• The reported 2028 revenue range is a private-company projection attributed by Reuters to people familiar with its financials.
• The possible IPO valuation is an expectation attributed to sources, not an announced market price.
• The business outlook depends on future adoption, costs, margins, competition and infrastructure investment.
CNBC reported that Anthropic had confidentially filed for a US IPO in June and was holding confidential “test-the-water” meetings with bankers and investors. The outlet said sources expected the company could float at about a $2 trillion valuation. CNBC also reported that Anthropic declined to comment. That means the figure should be read as a source-reported investor expectation, not an agreed valuation or an official company target.
Reported growth does not turn a TAM into revenue
Reuters’s reporting gives some indication of why future revenue features prominently in the investor story. It says Anthropic’s revenue run rate was about $9 billion at the end of 2025 before rising to more than $47 billion by May, a figure the company had publicised. Reuters also reported that the company projected at least $10.9 billion of second-quarter 2026 revenue and was on track for a first quarterly operating profit of $559 million.
These are specific figures reported by Reuters, with different levels of attribution and different periods. A run rate is not a full-year audited revenue figure. A quarterly forecast is not a completed result. A projection for 2028 is not an outcome. Keeping those distinctions clear is especially important when numbers are being used to explain a possible public-market valuation.
The same applies to comparison with SpaceX. The Wall Street Journal reported that SpaceX cited a $28.5 trillion addressable market in its May filing, with much of that estimate linked to AI technology. This supports a comparison of reported market estimates. It does not establish that either company will earn an amount approaching the size of its reported TAM.
David Merkel, principal at Aleph Investments, put the core investor question to Reuters: “Does it [AI] really produce so much additional productivity?” He added that these were questions that needed to be asked when considering how to price or buy the company. That is not a judgement that AI cannot create value. It is a reminder that the size of an opportunity and the ability to convert it into durable revenue are separate questions.
Enterprise use is central to Anthropic’s own positioning
Anthropic has emphasised enterprise use as an important part of its business. In a January CNBC interview, chief executive Dario Amodei said that roughly 80% of the company’s business was enterprise and 20% was consumer.
Amodei said: “Since the beginning, Anthropic has thought in terms of safety and reliability of AI systems, and one of the things we realised is that that was very synergistic with working with enterprises as compared to consumers.”
This executive statement, reported by CNBC, is not an independently audited segmentation disclosure. It does help explain why the reported market is framed through the value of work models it may support, rather than a count of consumer subscriptions.
Reuters says Anthropic’s reported TAM considers the full scope of work that AI models could complete. That framing reaches into the value created by knowledge work across many sectors. It is a strategic narrative about the possible scale of AI, not a simple calculation of an existing product category.
The connection between safety, reliability and enterprise adoption is also worth treating carefully. Amodei’s statement describes Anthropic’s rationale for working with enterprises. It does not establish that every enterprise will adopt a given model, that all use cases will be suitable, or that the company will maintain a particular market share. Those are future outcomes, not facts contained in the interview or the TAM report.
The implications for a reader of AI-market headlines
Numbers at this scale can seem either persuasive or absurd, depending on the reaction they provoke. Neither response is sufficient. A better approach is to identify the type of claim and then ask what evidence would be required to assess it.
For a TAM, the first question is the market boundary. Is the market defined as a current product category, a group of buyer budgets, a set of tasks, or the economic value of work? The broader the boundary, the further the estimate moves from a near-term sales outlook.
The second question is the route from theoretical opportunity to revenue. A market may be large while the proportion captured by any one company remains small. Adoption can be constrained by data, cost, governance, integration, competition, trust, skills and customer priorities. Those are not reasons to predict failure. They are the variables that a TAM does not settle.
The third question is time. A revenue run rate refers to a recent pace of business. A forecast refers to a future period. A valuation reflects what investors are prepared to pay under a particular set of assumptions. Conflating the three creates a story that sounds definite but is not.
Takeaway
Anthropic’s reported $30 trillion TAM is an important signal of the scale of its prospective investor narrative. It is not a reason to treat $30 trillion as expected sales, an announced valuation or a measure of a buyer’s likely return from an AI tool.
Keep four distinctions in view:
• TAM is a potential opportunity, not revenue. It describes a theoretical annual market at full capture.
• Private-company forecasts are projections. The reported 2028 revenue range comes from sources familiar with the company’s financials, not a public audited statement.
• A source-reported valuation is not a market price. The possible $2 trillion IPO valuation reported by CNBC remains an expectation cited to sources.
• Vendor scale does not decide a specific use case. Product fit, security, commercial terms, cost and adoption must be considered separately from a headline market estimate.
The largest number in the story is not necessarily the most useful one. The more useful question is which assumptions turn the number into a claim about a real market, a real product and a real period of time.
Big AI numbers are easy to repeat and harder to interpret. Subscribe to the Project Flux newsletter for the context behind the next one.
Links and Stuff
All content reflects our personal views and is not intended as professional advice or to represent any organisation.
1