SpaceX's initial public offering prospectus values the company above $1.5 trillion by pointing to a $28.5 trillion addressable market for space, connectivity and artificial intelligence services, with AI alone accounting for $26.5 trillion of that figure. Two months later Anthropic, in comments reported by The Wall Street Journal, set its own AI total addressable market at $30 trillion, a number that assumes the company could theoretically capture 100% of all work performable by AI models.
The numbers behind the narrative
Anthropic reported $11.6 billion in revenue for the second quarter of 2026, more than double the same period a year earlier, and earned what the Journal described as a small operating profit. Spacex's prospectus does not disclose comparable quarterly revenue. Both companies' TAM estimates rely on the same theoretical ceiling: 100% market share of an expanding AI economy, with no timeline attached.
History suggests skepticism
Previous IPO candidates have used similarly expansive market claims. Uber told investors in 2019 that its addressable market reached $6 trillion; last year the ride-hailing company generated $52 billion in revenue, roughly nine-tenths of one percent of that target. Rivian cited a $9 trillion opportunity in 2021 and booked less than $5.4 billion last year, under one-tenth of one percent. Instacart's 2023 grocery TAM of $1.1 trillion proved the closest, with 2025 revenue of $3.7 billion reaching about 3.4% of the estimate.
The structural problem
The Journal characterizes TAM estimates as especially squishy because companies rarely define their components or specify when they might be reached, making verification impossible before the offering prices. When multiple firms, SpaceX, Anthropic, OpenAI, Microsoft, Meta, DeepSeek, all target the same $30 trillion theoretical maximum, the arithmetic dictates that none can capture the whole. Each will take a sliver, and the aggregate of those slivers will fall well short of the marketed number.