Is Anthropic Worth Two Trillion Dollars?

Episode Summary

Executive Summary: The episode argues that Anthropic’s rumored $2T IPO valuation is hard to justify by conventional finance, especially as high rates, slower IPO markets, and falling AI model prices collide with circular financing across the AI ecosystem. It contrasts the speculative pricing of AI labs with NVIDIA’s comparatively cheaper earnings multiple, suggesting markets reward promises more than profits.

Main Topics: Anthropic’s proposed IPO valuation (Priority: 5/5): The host questions how Anthropic could justify a $2 trillion listing price, noting the company has weak comparables, uncertain cash-flow forecasting, and a valuation that implies extraordinary future growth rather than current economics. Interest rates and the IPO market (Priority: 5/5): Rising Treasury yields and a hawkish Fed make future profits worth less today, while IPO activity has cooled despite record-equity markets, with several large listings cancelled or delayed. TAM inflation and speculative AI narratives (Priority: 5/5): The episode criticizes total addressable market claims that balloon from tens of trillions to nearly the entire global economy, arguing that these figures are used to rationalize extreme valuations without proving capture is possible. Circular financing in the AI stack (Priority: 5/5): The speaker describes a loop where SoftBank, OpenAI, NVIDIA, cloud platforms, and data-center projects fund and validate one another’s valuations, creating a self-reinforcing but fragile ecosystem. Why NVIDIA looks cheap (Priority: 4/5): Despite being the main beneficiary of AI capex, NVIDIA trades at a relatively modest multiple because investors fear peak-cycle margins, customer in-house chip development, and a slowdown in AI spending. Price compression and model commoditization (Priority: 4/5): AI model prices are falling rapidly even as underlying infrastructure costs rise, making it harder for frontier labs to preserve margins or defend premium pricing before a public listing. Valuation risk vs. technological progress (Priority: 4/5): The episode does not argue AI is useless; it argues that a transformative technology can still be a bad investment if entry prices are extreme and future profitability is uncertain.

Key Arguments: A $2T valuation implies roughly 31x revenue, which is difficult to justify even under extremely generous assumptions. Discounted cash-flow math becomes unfavorable when interest rates are high, because future profits are worth less today. Anthropic’s valuation is mostly a growth bet, not a present-day earnings bet. TAM figures in AI have inflated rapidly and are being used rhetorically to support huge valuations. The AI industry is increasingly circular: companies invest in, rent from, guarantee, and buy each other’s products and debt. NVIDIA appears cheap because investors trust its profits more than the labs’ speculative future cash flows. AI model prices are falling quickly, which benefits users but pressures labs’ revenue models. Public-market scrutiny may expose how much of AI’s current boom depends on private valuations and narrative rather than cash generation.

Data Points: Anthropic target valuation: $2 trillion - Reportedly sought IPO valuation Anthropic revenue run-rate: around $65 billion/year - Unofficial August figure cited, with caution about accuracy 10-year Treasury yield: 5.23% - Reached amid hot economy and sticky inflation Fed rate move: Raised rates earlier this month - Used to frame tighter financing conditions Recent IPO activity: 3 IPOs since Labor Day - Evidence of a slowing IPO window Large listings under issue price: 5 of the 10 largest listings this year trading below offer price - Shows weak aftermarket performance Anthropic valuation/revenue multiple: about 31x revenue - Implied by a $2T price tag Perpetual revenue present value under generous assumptions: about $1.27 trillion - Even assuming zero costs, zero taxes, and full payout at Treasury discount rates SpaceX enterprise apps TAM claim: $22.7 trillion - Cited as a huge AI-related market estimate Anthropic filing market size rumor: $30 trillion - Reportedly may be cited in IPO filing Morgan Stanley generative AI estimate: $60 trillion - Lex notes this would be about half of global annual output Uber TAM at IPO: $12.3 trillion - Example of huge TAM vs limited current revenue WeWork TAM: $3 trillion - Example of TAM inflation ending badly Anthropic internal extreme scenario: Over $10 trillion added to US GDP by 2030 - Then translated into a much larger equity-value implication SB Energy IPO valuation: around $50 billion - Despite not yet operating a single data center SB Energy funding need: $174 billion - Prospectus says required to build promised projects New debt yield: 9.75% - SB Energy debt priced at junk levels OpenAI expected cash burn: almost $280 billion by end of 2030 - Reported forecast of future spending NVIDIA sales growth: about $27 billion to an estimated $410 billion - Roughly over four years NVIDIA valuation multiple: under 17x expected next-year profits - Described as near its cheapest in over a decade NVIDIA gross margin: 75% last quarter - Expected by analysts to slip below 72% by year-end AI model cost decline: about 13-fold per year since 2023 - Epoch AI estimate of falling cost for a given performance level Weekly AI usage growth on OpenRouter: about 25,000% since start of last year - Shows adoption surge

Pivotal Quotes: "What were you thinking?" — Scott McNeely: His 2002 critique of paying 10x revenue for Sun Microsystems at the dot-com peak, used as a warning about Anthropic’s valuation "you don't need a Ferrari to go pick up your groceries" — Eric Gleiman: RAMP co-CEO describing how companies are routing tasks to cheaper AI models instead of using frontier models for everything "show me the numbers, not just the words" — Aswath Damodaran: His reaction to AI valuation narratives and the need for a prospectus to test assumptions

Implications: The AI boom may be real, but public-market pricing could punish labs if growth, margins, or TAM claims disappoint. Investors should distinguish transformative technology from inflated entry prices, especially amid rising rates and increasingly circular financing.