AI profit gaps expose reliance on investor funding, warns Apollo economist

Torsten Slok, chief economist at Apollo, analyzed AI companies across four categories—models and applications, cloud and compute, energy and grid, and silicon and equipment—using data from Pitchbook and Bloomberg. He reported that silicon and equipment firms, such as Nvidia and AMD, achieve a 41% operating margin, whereas model and application firms like Anthropic register a negative 59% margin. Slok argues that this gap exists because investor capital, not end-user sales, is financing the losses of the upstream segment.

He warns that a slowdown in AI financing could threaten the industry's stability, as the profitable downstream segment depends on continued capital inflows. The Bank of International Settlements and Bank of America have echoed concerns about hyperscalers' rising debt and potential pullbacks. Ed Zitron further highlighted Oracle’s $23.7 billion negative cash flow and $130 billion debt tied to a $300 billion OpenAI deal, underscoring the broader risk of an AI spending bubble.