Aswath Damodaran: Big Tech Has No Idea How AI Pays Off

Aswath Damodaran: Big Tech Has No Idea How AI Pays Off – Prof G Markets

Market views

Bullish

AAPL— Apple is taking a cautious wait-and-see approach to AI CapEx rather than building massive infrastructure first, which may prove strategically superior. Apple maintains high returns on invested capital (70-80%+) unlike peers who have become capital-intensive manufacturing-like businesses.
AMZN— Amazon has experience managing negative cash flow cycles and returning to profitability, making it best-equipped among Big Tech to navigate the AI investment phase. Amazon's history of transparency about long-term investments has earned market trust.

Bearish

META— Meta's free cash flow collapsed 91% due to massive AI CapEx without clear business model articulation. Management is unable to explain ROI on AI investments and lacks transparency, having learned nothing from the metaverse fiasco.
GOOGL— Google's free cash flow went negative for the first time due to massive AI CapEx. The company is transforming from a high-margin tech company to a capital-intensive manufacturing-like business without demonstrating economies of scale or clear unit economics.
MSFT— Microsoft is heavily exposed to OpenAI, with Wells Fargo estimating 74% of Microsoft's AI revenue comes from OpenAI. This creates dangerous concentration risk in an unprofitable partner, representing intra-company revenue rather than end-user demand.
OPENAI— OpenAI lacks viable unit economics with high delivery costs and no visible path to profitability. The company is highly unprofitable and dependent on continuous funding from Big Tech partners who are themselves questioning AI ROI.
ANTHROPIC— Anthropic represents 73-74% of Amazon and Microsoft's AI revenues respectively, but is highly unprofitable with questionable unit economics. The company cannot survive on its own cash flows and faces existential risk if AI investment cycle corrects.
AILABS— AI frontier labs face a demand crisis rather than supply scarcity, with Meta and Musk trying to lease out compute capacity. The shift from compute scarcity narrative to excess capacity signals overestimation of AI demand and potential market correction ahead.
AIMODELS— Chinese AI models have rapidly captured over 50% token consumption share (from 8% in January 2025), engaging in AI dumping similar to historical steel market tactics. This threatens US AI companies' pricing power and business models, potentially doing to Silicon Valley what Japan did to Detroit.
DATACENTER— AI data center infrastructure lacks economies of scale, with marginal costs not decreasing despite massive investment. The physical infrastructure is too expensive and not easily scalable, creating a structural problem for AI profitability.
TECH— Big Tech AI investments mirror the 1999 dot-com bubble with sequential segment failures (B2C to B2B to infrastructure). The cycle shows eerie parallels to 2000 implosion, with no major CapEx write-offs yet but correction catalysts building.

Not investment advice.

Mentioned

Market views