Top economist warns AI profits depend on investor cash, not customer demand

On Friday, Apollo Chief Economist Torsten Slok published a blog post warning that the AI boom's profitability structure is fundamentally misaligned with how capital is flowing through the market. Using data from Pitchbook and Bloomberg for companies including OpenAI, Anthropic, Microsoft, Amazon, Constellation Energy, Nvidia, AMD, and Micron, Slok broke AI companies into four categories: models and applications, cloud and compute, energy and grid, and silicon and equipment. His calculation revealed a stark inversion: silicon and equipment makers have the highest profit margin at 41%, while models and applications companies—the creators of AI technology itself—operate at a -59% margin.

Slok attributes this disparity to a fundamental mismatch in how the boom is funded. "AI boom's profits are currently being funded by investors rather than earned from customers," he stated. "The upstream margins are real, but they are paid for out of capital raised by the layer losing money, not out of cash generated by end demand." This creates a precarious dependency: the profitable chipmakers depend on the unprofitable AI model and application makers to continually raise capital and spend it on hardware. If that capital supply dries up, the entire chain risks collapse.

Other major financial institutions have raised similar alarms. The Bank of International Settlements noted in its annual report (June) that AI investment from the five major hyperscalers is outpacing earnings and free cash flow, forcing these companies to issue debt. A Bank of America analysis from November found that in 2025 alone, those five hyperscalers issued $121 billion in debt—four times their average annual debt issuance over the previous five years. Goldman Sachs now projects AI investments to swell beyond $1 trillion in 2026, yet the technology has produced little evidence of widespread productivity gains or profit margin growth outside the Magnificent Seven.

Tech writer Ed Zitron highlighted the concrete stakes in a Substack post in June, pointing to Oracle as a cautionary example. Oracle has a negative cash flow of $23.7 billion as of the end of fiscal 2026 and nearly $130 billion in outstanding debt, plus $260 billion in lease commitments for AI infrastructure projects that have yet to begin. All of this is tied to a $300 billion deal signed with OpenAI in September. "Oracle's existence—and Larry Ellison's personal wealth—hinges on whether OpenAI can make good on its promise to spend $300bn in compute," Zitron wrote. He warned that the real danger is not Oracle failing alone, but other hyperscalers pulling back: "If Microsoft, Google, Amazon and Meta decide that it's time to stop spending $30 billion or more a quarter on GPUs, RAM, storage, and data center construction, that'll tear a hole in the side of what people assume is a permanent supercycle." Slok's central question to the market is whether AI's end customers will show returns fast enough to sustain this level of spending—a bet that so far remains unproven.