Musings on Markets is the personal blog of Aswath Damodaran, professor of finance at NYU Stern and widely known as the "Dean of Valuation." Running since 2008, it publishes hands-on intrinsic-value teardowns of headline companies (SpaceX, Tesla, Nvidia) using his narrative-and-numbers DCF framework, plus periodic market-wide reviews.
In his October 2026 analysis, Aswath Damodaran classifies Anthropic and OpenAI as high-growth, high-loss, negative free cash flow young companies, arguing their trillion-dollar valuations rely on 'messy metrics' like ARR rather than intrinsic valuation. He finds Anthropic's claimed Q2 2026 profitability unreliable due to adjusted earnings that add back legitimate expenses. Anthropic's ARR is $67B (expected to exceed $100B by year-end), while OpenAI's ARR dropped from $70B to $50B after switching from gross to net revenue reporting. Damodaran warns that markets may shift to forward revenues (e.g., 2030) to justify high multiples, but investors might mistakenly compare these forward multiples to current market multiples.
In his October 2026 analysis, Aswath Damodaran classifies Anthropic and OpenAI as high-growth, high-loss, negative free cash flow young growth-stage companies based on operating metrics (revenue growth rate, profit margins, free cash flow) rather than years since founding or industry categorization. He argues that their trillion-dollar valuations rely primarily on "messy metrics" (such as ARR) rather than traditional intrinsic valuation.
Both Anthropic and OpenAI are in a high-growth, high-loss, negative free cash flow young growth stage, despite their revenue already reaching tens of billions of dollars.
During a San Francisco event on October 15, 2026, the author noted that Anthropic disclosed third-quarter 2026 revenue of $16.8 billion (up 45% quarter-over-quarter) in its pre-IPO meetings, achieved adjusted profitability, and had an annual recurring revenue (ARR) of $67 billion, expected to exceed $100 billion by year-end. The author argued that young companies often rely on “messy metrics” (such as ARR, eyeballs, or user counts) rather than traditional valuation metrics (earnings, book value, cash flow) for pricing, and explained why pricing replaces valuation.
The author proposed three methods for determining a company’s stage:
| Dimension | Method | Limitation |
|---|---|---|
| Age | Classify by years of operation | Infrastructure industries may take decades; tech companies can go from startup to maturity in a few years |
| Industry Classification | Treat the entire industry as being in a specific lifecycle stage | Within tech, there are mature/declining companies; within utilities, there are companies growing due to the data center boom |
| Operating Metrics | Revenue growth rate, profit margin, free cash flow, failure risk | Most accurate; can identify misclassified companies |
Common operating characteristics of young companies: High revenue growth (often from a small base, but with exceptions), operating losses (business model not yet formed, fixed costs overwhelming revenue), negative free cash flow (reinvesting to support growth), and high failure risk (requiring external capital to survive).
The author’s original words: “if a company has revenue growth in single digits, stable margins and positive free cash flows to the firm, it is a mature company, no matter how it views itself.” In other words: “If a company’s revenue growth is in the single digits, its margins are stable, and its free cash flow to the firm is positive, then it is a mature company, no matter how it sees itself.”
As of October 2026, the key operating metrics for the two companies:
| Metric | Anthropic | OpenAI |
|---|---|---|
| Revenue (ARR basis) | $67 billion (expected to exceed $100 billion by year-end) | Specific figure not disclosed |
| Revenue Growth Rate | Extremely high (45% quarter-over-quarter) | Extremely high |
| Operating Profit | Adjusted profitability, but the author considers it unreliable | Loss-making |
| Free Cash Flow | Deeply negative | Deeply negative |
| Valuation Rumors | Total market cap over $1 trillion | Total market cap over $1 trillion |
The author noted that both companies have already reached a large revenue scale (“growing at scale”), but their business models are still in flux (usage-based vs. subscription pricing, high-end vs. mass-market positioning). Regarding Anthropic’s claimed profitability, the author’s original words: “I know that Anthropic claimed to be profitable in the second quarter of 2026, but I don't think that even Anthropic believes this to be true, since these are adjusted profits and the adjustments represent a patchwork of add-backs of legitimate expenses.” In other words: “I know Anthropic claimed to be profitable in the second quarter of 2026, but I don’t think even Anthropic itself believes this to be true, because these are adjusted profits, and the adjustments are a patchwork of add-backs for various legitimate expenses.”
The author distinguishes between two approaches:
The author believes that those who do not accept the efficient market hypothesis have two paths: the investor path (valuation → buy undervalued/sell overvalued → wait for price to revert to value) and the trader path (abandon intrinsic value judgment → use charts and other technical tools to buy low and sell high). Young companies tend to rely on pricing rather than valuation because their cash flows are unstable, traditional valuation metrics are ineffective, and they can only depend on “messy metrics” like ARR or user counts.
The author explicitly states that the adjusted profitability of Anthropic and OpenAI is unreliable, their free cash flow is deeply negative, and whether they can generate positive cash flow in the future depends entirely on how the AI business evolves. Readers should note that this is an academic perspective; the author does not offer buy or sell recommendations but emphasizes that understanding a company’s lifecycle stage is crucial for choosing between valuation and pricing methods.
The author points out that while investors tend to use pricing (comparables) instead of valuation (discounted cash flow) to assess young companies, the pricing process itself relies on implicit assumptions about growth, risk, and profitability, and faces challenges such as data scarcity and difficulty in finding comparable companies.
The author first explains why valuing young companies is more complex than valuing mature ones. Valuation relies on financial statements, market size, and industry statistics to forecast future cash flows, growth, and risk. For young companies like Anthropic and OpenAI, their financial statements are full of "noise" due to high growth and unformed business models, historical data is limited, and the markets they operate in are still rapidly evolving, making the valuation task exceptionally difficult. The author states, "valuing young companies will always be more complicated than valuing more mature businesses."
Consequently, many investors turn to pricing, which involves referencing the transaction prices of similar companies. Using the recent funding rounds of Anthropic and OpenAI as examples, the author illustrates the market's "pricing" of them:
| Company | Most Recent Funding Round Date | Amount Raised | Post-Money Valuation |
|---|---|---|---|
| Anthropic | May 2026 | $65 billion | $965 billion |
| OpenAI | March 2026 | $122 billion | $852 billion |
| OpenAI (Pending) | Pre-IPO | $30 billion | ~$1.4 trillion |
However, the author emphasizes that venture capital (VC) pricing is inherently "forward-looking pricing," not based on intrinsic value. For companies like Anthropic and OpenAI approaching an IPO, the pricing process is "both more compressed and even less informative." Investors are effectively pricing based on expectations of the IPO price, discounting it to reflect uncertainty. For example, when Anthropic raised funds in May 2026, the market's expected IPO price was $1.15 trillion, while the post-money valuation of $965 billion represented a discount of approximately $150 billion.
The author breaks down the pricing process into three steps: standardizing prices, finding comparable companies, and controlling for differences, noting that each step faces unique difficulties when evaluating Anthropic and OpenAI due to their lifecycle stage.
The author elaborates on the three steps of pricing:
1. Standardizing Prices: Dividing the stock price by a common variable (e.g., revenue, profit, book value) yields multiples like the P/E ratio or EV/Sales, enabling cross-company comparisons.
2. Finding Comparable Companies: Identifying companies similar to the one being priced that already have market prices, typically categorized by industry, geography, and size.
3. Controlling for Differences: Even when using the same multiple, different companies will have different multiple levels due to variations in growth, risk, etc. These differences must be accounted for during pricing.
The author points out that for companies like Anthropic and OpenAI, all three steps face challenges. First, because they are not yet profitable or have unstable earnings, commonly used profit-based multiples like P/E are unusable, narrowing the choice to revenue-based multiples (e.g., EV/Sales) or more "messy" metrics. Second, finding truly "comparable" companies is extremely difficult, as they are unique in technology, business model, and growth potential. Finally, controlling for differences requires making numerous implicit assumptions about future growth and profitability, which essentially brings the analysis back to the core judgments needed for valuation. In the next section, the author will specifically explain why using multiples like Annual Recurring Revenue (ARR) or user counts inherently involves implicit assumptions about future profitability, growth, and reinvestment.
The author notes that early-stage loss-making companies like Anthropic and OpenAI can only be priced based on revenue multiples, but the calculation of Annual Recurring Revenue (ARR) is not standardized and can be easily manipulated. The author explains that when a company is loss-making and has negative free cash flow, revenue is the only viable metric, hence "desperation drives you up the income statement until you find a positive number." This forces the market to focus on revenue, and due to extremely rapid growth, it shifts to annualized revenue run rates (ARR). The author warns verbatim: "the perils with annualized revenue rates (ARR) is that they are not standardized and can be easily gamed." He cites an example where OpenAI recently shifted its reporting basis from gross revenue to net revenue, causing its ARR to drop from $70 billion to $50 billion, while Anthropic still reports gross revenue, treating cloud provider payments as expenses. Furthermore, ARR can be calculated using data from the last day, last week, or last month. Although both companies currently use monthly annualization, there is no binding standard.
The author argues that as the pricing debate intensifies, the market may shift toward forward revenue or even forward earnings, stretching the forecast period to make current high multiples appear reasonable. Anthropic has already set a target of over $100 billion in ARR by the end of 2026, opening the door to using forward revenue. The author predicts that investors and analysts may look even further to 2030 or 2036 revenue, because "scaling prices today to forward revenues five or ten years out will make the pricing multiple look lower, and thus easier to defend." More extreme scenarios involve shifting to expected operating profit or net profit margins for 2031 or 2036—for example, claiming that Anthropic's current market cap is only 25 times its expected 2036 earnings. However, the author warns: "Investors use frames of what comprises normal or high, based upon their acquaintance with a multiple, and many will (mistakenly) compare these forwards earnings multiples to current earnings multiples in the market, leading to incorrect judgments of whether something is expensive or cheap."
The author points out that linking pricing to user numbers or compute power has inherent logic, but such multiples are only applicable to companies within the same industry and lack a historical frame of reference. The logic is that more users or compute power are leading indicators of future revenue and earnings. The author recalls the precedent from the internet era when Mary Meeker used website traffic and eyeballs to price companies. For AI companies, market cap or enterprise value can be linked to compute power (measured in gigawatts) or spending on compute capacity (compute commitments). However, the author notes the downside: "The downside of using metrics that are specific to a business (users, eyeballs or compute power) is that you can compute that multiple only for the companies in this business, restricting comparisons, and you have no perspective on what comprises high, low or typical for that multiple."
The pricing comparison table provided by the author shows that based on different metrics, Anthropic and OpenAI each appear cheaper in some respects, but both may be overvalued overall. Under assumed market caps of $1.5 trillion and $2 trillion, based on latest ARR or 2026 ARR, Anthropic appears cheaper than OpenAI; but based on 2030 expected revenue, compute cost/capacity, and user numbers, OpenAI appears cheaper. The author notes caveats for each multiple: Anthropic uses gross revenue while OpenAI uses net revenue; Anthropic's users are more enterprise-oriented and usage-based, so its per-user pricing should be higher. But the author emphasizes: "it is entirely possible that both Anthropic and OpenAI may be overpriced, and by comparing the two, you are just finding the less overpriced entity."
The author argues that Anthropic and OpenAI face the most difficult pricing scenario—very few companies in the industry and almost none are publicly listed, resulting in a lack of a reliable peer group. Pricing relies on a peer group, but competitors in the LLM space are either small parts of large tech companies (e.g., Gemini is a small part of Alphabet, xAI is part of SpaceX, Muse is overshadowed by Meta's advertising business) or are themselves unlisted. The author categorizes pricing scenarios into four levels of difficulty, from "many listed companies" to "very few companies and almost none listed," with Anthropic and OpenAI falling into the last category. When Anthropic's IPO roadshow begins, the lack of a peer group will become an obstacle. The author notes that the narrowest pricing game is to pair the two against each other, but at least until OpenAI goes public, the comparison will be between Anthropic's public pricing and OpenAI's private pricing, with SpaceX potentially added as a noisy peer.
The author argues that regardless of which pricing metric is used (e.g., price-to-sales), one must understand the transmission chain linking that metric to the company’s intrinsic value—growth, profitability, and risk, all three are indispensable. A bullish or bearish narrative on Anthropic that seizes only one link is incomplete.
The author expands the pricing comparison to a broader universe. If Anthropic approaches a $1.5 trillion market cap, it would be benchmarked against the Mag Seven (the seven mega-cap tech giants) and SpaceX—companies with market caps near or well above $1 trillion. The author’s original words:
> "The argument will be that investors who seek out large market cap companies will compare them to each other, even though Tesla, Apple and Anthropic are very different companies in terms of business model and where they are in the corporate life cycle."
Meaning: "The argument will be that investors seeking large-cap companies will compare them against one another, even though Tesla, Apple, and Anthropic are very different companies in terms of business model and corporate lifecycle stage."
Pricing comparison results (based on $1.5 trillion and $2 trillion target valuations):
| Comparison Company | Meaning for Anthropic Bulls | Meaning for Anthropic Bears |
|---|---|---|
| SpaceX | Anthropic is cheaper than SpaceX | The AI boom has inflated both companies; SpaceX’s high price does not imply fairness |
| Mag Seven | Anthropic’s high valuation can be justified by higher expected growth | The vast gap in revenue and profitability (not shown in the table, but strengthens the Mag Seven case) cannot be bridged by hope and potential alone |
The author points out that regardless of which pricing metric is chosen (e.g., P/E), one must consider how that metric connects to company value and which variables need to be controlled for in pricing. Using P/E as an example, the chain linking current earnings to overall value has three links:
1. Earnings growth: Higher growth → Higher value
2. Growth efficiency: Higher return on equity (ROE) → Same earnings generate more cash flow
3. Risk: Lower risk → Lower discount rate → Higher equity value
Since P/E is inapplicable to Anthropic and OpenAI, the author shifts to enterprise value/revenue (using ARR or trailing twelve-month revenue). Under this metric, Anthropic’s high valuation requires three conditions:
| Pricing Determinant | Specific Content | Requirement for Anthropic |
|---|---|---|
| Revenue growth | Not just large in magnitude, but also fast | Achieve "sky-high" revenue growth in early years |
| Profitability | Path from current losses to profitability | Operating margins surge to 30% or higher, relying on favorable unit economics and scale effects |
| Risk | Operational risk, financial risk (compute commitments), regulatory risk | Risk profile converges rapidly to average levels; regulatory risks fade and business model solidifies |
The author judges: "While you may find this story to be a reach, it is a plausible one"—meaning: "While you may find this story a stretch, it is plausible."
The author uses the parable of the blind men and the elephant to summarize: Both Anthropic’s bulls and bears tend to grasp only part of the company’s story.
The bull’s complete narrative:
The bear’s complete narrative:
The author explicitly declines to pick a side: "It is because I am investor at heart and feel more comfortable making my decision based upon an intrinsic valuation of Anthropic"—meaning: "Because I am an investor at heart and feel more comfortable making my decision based on an intrinsic valuation of Anthropic." This will have to wait until the full prospectus is filed (expected next week).
| Ticker | Direction | Author's One-Sentence View | Key Data |
|---|---|---|---|
| Anthropic | Hold & Observe | The author believes its valuation relies on "messy metrics," adjusted earnings are unreliable, and free cash flow is deeply negative, but the high-valuation narrative (high revenue growth + high margins) is logically "reasonable." | Q3 2026 revenue of $16.8 billion (+45% QoQ); ARR of $67 billion (expected to exceed $100 billion by year-end); post-money valuation of $965 billion in a May 2026 financing round; rumored total market cap exceeding $1 trillion. |
| OpenAI | Hold & Observe | The author believes it is also in a young stage of high growth, high losses, and negative free cash flow, facing the same valuation dilemma as Anthropic. Additionally, the shift in ARR metric from gross revenue to net revenue has caused the figure to decline. | Post-money valuation of $852 billion in a March 2026 financing round; pre-IPO pending financing valuation of approximately $1.4 trillion; ARR dropped from $70 billion (gross revenue basis) to $50 billion (net revenue basis). |
| SpaceX | Not Explicitly Stated | The author uses it as a "noisy" peer in the valuation comparison between Anthropic and OpenAI, but believes the AI boom may have inflated valuations for both companies. | Market cap close to or well above $1 trillion (benchmarked against the Magnificent Seven and Anthropic). |