Anthropic IPO Pricing vs Valuation: ARR Multiples and Market Risks

This article delves into the imminent IPO of Anthropic, contrasting the concepts of value and price, and examining the pricing metrics used for young, high-growth companies like Anthropic and OpenAI.

The Anthropic IPO and the Shift to Annualized Revenue Run Rate (ARR)

Anthropic's pre-IPO meeting on October 15, 2026, revealed its annualized revenue run rate (ARR) was approaching $100 billion by the end of 2026. ARR is a metric adopted by companies like OpenAI and Anthropic, which annualizes recent revenue (e.g., from the last week or month) rather than relying on the past 12 months' revenue. This approach is favored by young, rapidly growing companies whose trailing 12-month revenues would not accurately reflect their current trajectory.

Pricing vs. Valuing Young Companies

Young companies are more likely to be "priced" than "valued." * Valuation requires understanding the business, projecting future revenue growth and cash flows, and assessing risk to arrive at an intrinsic value (e.g., through discounted cash flow). * Pricing involves looking at what others are paying for similar companies.

While ideally, these two approaches should converge in an efficient market, they often yield divergent numbers, especially for young companies.

Challenges in Valuing Young Companies

Valuing a company typically relies on three "crutches": 1. Past Financial Statements: For mature companies like ExxonMobil, historical data provides insights into business models and profitability. For young companies (2-3 years old), this history is minimal, and financial statements often show growing revenues but significant losses. 2. Market Size and Competition: Established companies operate in defined markets with clear competitive landscapes. Young companies often operate in evolving, undefined markets, making it difficult to estimate total addressable market (TAM) or future competitive advantages. 3. Industry Averages: Industry benchmarks for margins, turnover ratios, and cost of capital are readily available for established sectors. For nascent industries or unique business models, these averages are scarce or non-existent.

For companies like Anthropic and OpenAI, these challenges are amplified: * Limited Financial History: Only 2-3 years of data, with growing revenues but messy income statements. * Uncertain Future Market: The total addressable market for AI is still debated, with estimates ranging into trillions, but the competitive landscape and sources of sustainable advantage are unclear. * Risk Assessment: Lack of financial history makes it difficult to assess cyclicality or overall risk. * Compute Cost Commitments: Both companies have significant contractual commitments for computing power, which, while a potential strength, also represents a substantial financial risk for money-losing entities. * Corporate Governance: Anthropic's plan for two classes of shares (high and low voting rights) adds complexity to valuation.

Given these uncertainties, most people default to pricing young companies.

Venture Capital (VC) Pricing

VC "valuation" is often a form of "forward pricing." It involves: 1. Projecting a Future Metric: Arbitrarily selecting a future point (e.g., 3-5 years) and estimating a key metric (e.g., revenues). 2. Applying a Multiple: Using multiples from publicly traded companies (e.g., 5-7 times revenues for MAG 7 companies) to arrive at an "exit year pricing." 3. Discounting Back: Discounting this future price back to the present using a high, arbitrary discount rate (e.g., 30-70%), which serves more as a bargaining tool than a true reflection of risk or return.

For Anthropic and OpenAI, VC pricing is even more immediate, projecting to the imminent IPO date and discounting for the short waiting period and potential risks. For example, Anthropic's May 2026 pricing of $965 billion was based on an expected IPO pricing of $1.15 trillion, discounted by an arbitrary $150 billion.

The Pricing Process: Multiples, Peer Groups, and Storytelling

The generic pricing process involves three steps: 1. Choose a Multiple: * Numerator: A market value (market cap, enterprise value). * Denominator (Scalar): A common, positive metric across companies. For young companies, this often means climbing up the income statement to revenues (as earnings are often negative). For pre-revenue companies, it might be users, subscribers, or eyeballs. The timing of this scalar (last fiscal year, next year, or a distant future year) also needs to be chosen. 2. Find a Peer Group: Identify similar companies (by sector, geography, size) to compare pricing multiples. This is subjective and can be messy, especially for unique companies or those in nascent industries. 3. Control for Differences and Tell a Story: Compare multiples across the peer group and adjust for differences in growth, risk, and profitability.

Pricing Metrics Across the Life Cycle

• Startups (Pre-revenue): Multiples based on drivers of revenue (subscribers, users, eyeballs, total addressable market).
• Young Growth Companies: Primarily revenue multiples (e.g., EV/Sales, EV/ARR).
• High Growth Companies: Revenue multiples, with some emerging earnings multiples.
• Mature Growth Companies: PEG ratios, forward P/E ratios.
• Mature Stable Companies: A wide range of multiples (P/E, EV/EBITDA, etc.).
• Declining Companies: Book value multiples (Price/Book).

Peer groups also shift across the life cycle. Young companies are compared to other young companies, not mature ones, to avoid making them appear excessively expensive.

Anthropic and OpenAI: Potential Pricing Metrics

Given their characteristics, Anthropic and OpenAI will likely be priced using revenue multiples or other non-traditional metrics.

• Annualized Revenue Run Rate (ARR):

  • Anthropic (Sept 2026): $65-67 billion ARR.
  • OpenAI (Sept 2026): $50 billion ARR.
  • If Anthropic IPOs at $1.5 trillion, its multiple would be ~23x ARR.
  • A key challenge is that Anthropic reports gross revenues, while OpenAI reports net revenues, making direct comparisons difficult.
  • Using forward ARR (e.g., end of 2026 estimates: Anthropic $110B, OpenAI $70B) would lower the multiple (e.g., Anthropic at $1.5T would be 13.6x 2026 ARR).
  • Even more aggressive forward projections (e.g., 2030 revenues of $300B) would make the multiple appear even lower (e.g., 5x revenues at $1.5T). Bulls will likely use these far-forward numbers to justify high valuations.

• Compute Cost/Capacity:

  • Compute power is a scarce resource for AI companies.
  • Companies with more compute power lined up could be priced higher.
  • Anthropic's planned spending to accumulate compute power is over $600 billion, with over $400 billion in contractual commitments.
  • This could lead to metrics like "enterprise value to compute cost commitment" or "market cap to gigawatt compute capacity."

• Active Users:

  • Early on, OpenAI had a significant lead in active users (1.2 billion vs. Anthropic's 250 million in Sept 2026).
  • Dividing market cap by active users would make OpenAI appear cheaper.
  • However, Anthropic has shifted to an enterprise-focused, usage-based model, generating significantly more revenue per user ($6,000/user vs. OpenAI's $1,250/user). This allows Anthropic to argue its users are more valuable.

These divergent metrics will be used by bulls and bears to support their respective arguments about whether Anthropic is cheap or expensive.

Connecting Pricing Metrics to Value

While pricing avoids explicit valuation, the chosen metrics implicitly rely on underlying value drivers. * Price-Earnings (P/E) Ratio: Connects to value through expected earnings growth, return on equity (efficiency of reinvestment), and risk (cost of equity). * Revenue Multiples (EV/Sales, EV/ARR): Connects to value through: * Revenue Growth: Higher growth leads to higher future revenues. * Profit Margins: Revenues must translate into profits (unit economics, economies of scale). * Reinvestment Efficiency: How much capital is needed to generate revenue growth (e.g., compute costs). * Risk: Higher risk lowers value.

Therefore, to justify a $1.5-$2 trillion valuation for Anthropic, one needs a compelling story about: 1. High Revenue Growth: Not just high, but "through the stratosphere," quickly turning tens of billions into hundreds of billions. 2. Strong Profitability: Surging operating margins (30-40%) driven by favorable unit economics and economies of scale (compute costs growing slower than revenues). 3. Manageable Risk: Operating, financial (compute commitments), and regulatory risks quickly converging to average levels.

The "Blind Men and the Elephant" Analogy

When Anthropic goes public, analysts will present fragmented views, much like the blind men describing an elephant based on the part they touch. * Bulls will focus on the vast total addressable market for AI, Anthropic's high gross margins (e.g., 80%), and projected future revenues to justify high multiples. * Bears will highlight the massive compute commitments, potential regulatory pushback, unreliability of AI products, and corporate governance concerns, arguing that current revenue multiples are unsustainable.

The investor's job is to synthesize these pieces into a complete picture and decide whether the story of high growth, surging margins, and manageable risk is plausible enough to justify the price. The speaker, however, prefers intrinsic valuation over pricing due to the implicit assumptions and biases inherent in the latter, and plans to conduct a detailed intrinsic valuation of Anthropic once the full prospectus is released.