Speaker
Aswath Damodaran
Appearances over time
1 episodes
Episodes
1
Podcasts
Quotes & moments
Damodaran's intrinsic valuation of SpaceX came in at approximately $1.3 trillion, well below the market price of $2.7 trillion at time of recording.
The Starlink connectivity business accounts for roughly 60–70% of SpaceX's total revenues, making it the core commercial engine of the company.
SpaceX's prospectus claims a total addressable market of $26 trillion for AI — the largest TAM Damodaran has ever seen cited in a company filing.
The high-end AI model Claude Fable reportedly costs around $6,000 per hour to use, and Anthropic does not profit from it because the delivery costs are comparably high.
Starlink's competitive moat rests on having approximately 10,000 satellites in space, giving it coverage that no other satellite-based internet provider can match.
Elon Musk has publicly stated a target of $1 trillion in revenues for SpaceX by 2030, a figure that is only achievable if the AI business scales dramatically.
Unlike the dot-com boom which was equity-funded, the AI CapEx cycle is significantly funded by debt from private capital, meaning a correction could trigger defaults and societal spillover.
During the dot-com bust, equity investors bore losses of 60–90%, but those losses were contained to shareholders; by contrast, AI's debt-funded structure could create wider economic pain.
The Magnificent Seven tech companies are fundamentally changing their character, moving from low-CapEx, high-margin businesses to capital-intensive infrastructure operators driven by AI investment.
If the big AI market stories come true (e.g., $10–25 trillion TAMs), half of all white-collar workers could lose their jobs, creating severe societal consequences.
Damodaran asserts that the average active value investor in the 20th century actually underperformed a simple value index fund, undermining the legend that traditional value investing consistently added alpha.
Approximately 55–60% of investment capital now flows into index funds and ETFs, which value investors blame for distorting price-to-fundamentals relationships.
Damodaran disclosed he personally owns five of the Magnificent Seven tech stocks and has held Amazon since 1997, illustrating his long-term exposure to these now capital-intensive businesses.
Damodaran received a legal settlement notice revealing that Anthropic had used 12 of his books without permission to train its AI models — a sign that training data costs will rise as IP owners assert rights.
Damodaran's intrinsic valuation of SpaceX lands at roughly $1.3 trillion — less than half the market price of $2.7 trillion. The gap is not about doubting SpaceX's engineering brilliance or market position; it's about what the price already assumes about AI unit economics, gross margins, and future growth that haven't materialized yet.
SpaceX's prospectus claims a $26 trillion AI total addressable market — the largest Damodaran has ever seen. But a big market with poor unit economics and massive reinvestment needs can destroy value rather than create it. Getting to revenue and profit from a large TAM requires a chain of assumptions most analysts never complete.
The dot-com bust was painful but contained — equity investors lost 60–90% and that was it. The AI CapEx cycle is different: it's the largest infrastructure buildout Damodaran has ever seen, and it's substantially funded by private debt rather than equity. When the correction comes, defaults will spill pain into the broader economy, not just shareholders.
SpaceX claims it will win a dominant share of the AI market. At the same time, it generates nearly $2 billion renting data center capacity to Google and Anthropic — its biggest AI competitors. Damodaran frames this as a fundamental strategic contradiction: SpaceX needs to pick whether it's an AI competitor or an AI infrastructure landlord, because you can't credibly be both.
AI can only justify $10–25 trillion market valuations if it replaces people, not just assists them. If those stories come true, half of all white-collar workers lose their jobs. Damodaran draws a sharp contrast with the 1990s factory closures: the same people who dismissed displaced steelworkers with 'learn to code' are now the target. This time, the advice is 'learn to plumb.'
Investors who missed Amazon in 1999 are now haunted by that regret (ROMO — Regret Over Missing Out). Combine that with FOMO and a high-profile IPO drought, and you get irrational capital flowing into SpaceX at prices that no fundamental story fully supports. Damodaran coins 'ROMO' to name the psychological force that turns hindsight bias into investment mistakes.
SpaceX started as a space launch company, but the real commercial breakthrough was Starlink: broadband internet from 10,000 satellites in orbit. Because SpaceX launches satellites cheaper than anyone else, Starlink has coverage competitors can't match — and it now generates 60–70% of SpaceX's revenues. Without Starlink, SpaceX would still be a niche business.
Claude Fable reportedly cost $6,000 per hour to use — and Anthropic still lost money on it. The costs come from data centers, power, and water that don't benefit from traditional economies of scale. Until someone solves the unit economics of high-end AI, the entire LLM industry is competing for a market where profits may structurally not exist.
Damodaran's diagnosis of value investing's decline is that it became a religion: rigid rules that prevent nuance, rituals like reading Security Analysis and making the Omaha pilgrimage, and a righteousness that blames passive investing rather than accepting responsibility for underperformance. The result is a style that ChatGPT can now replicate in seconds — and that still refuses to acknowledge intangible or growth assets.
Value investing can evolve, but it needs to abandon three habits. First, drop 'I will never buy Tesla or SpaceX' — any company is worth buying at the right price. Second, stop hunting for accounting conspiracies in footnotes while missing the forest for the trees. Third, accept that book value is an obsolete proxy for company worth and learn to value intangible assets and future growth properly.
For companies like SpaceX where the outcome distribution is enormous, point-estimate valuations create false precision and invite overconfidence. Damodaran argues that turning inputs into probability distributions shows investors both the estimate AND how wrong they could be — and provides an honest framework for disagreeing with others who have a different but equally legitimate story.
The Magnificent Seven have never built 10-year infrastructure assets before. They grew rich by scaling with almost no reinvestment. Now they're building massive data centers that take a decade to depreciate but could be obsolete in five years. They're playing a game they don't know how to play — and Damodaran is watching their earnings reports very differently as a result.
Software companies enjoy near-zero marginal costs at scale — the next unit of output costs almost nothing. AI may be fundamentally different: like Spotify, where every stream requires a new payment, LLMs may face persistent per-unit costs from power, water, data, and compute that never scale away. If true, the margin expansion the market is pricing in for AI may never arrive.
Analysis
What they talk about
- Business 79%
- Technology 21%
Connections
Shows they appear on and people they share episodes with. Drag to explore.
