The article argues that the current "AI demand bubble" is being artificially inflated, primarily by massive investments and revenue flows between major tech companies and two key AI labs: OpenAI and Anthropic. Hyperscalers like Amazon, Google, and Microsoft are reporting significant cloud revenue growth, which they attribute to AI. However, the author contends that this growth is largely dependent on the substantial compute spending of OpenAI and Anthropic, rather than a diverse customer base.

These AI labs, described as "unprofitable, unsustainable," rely on billions of dollars in funding from the very hyperscalers they ostensibly serve. For instance, Google has invested up to $30 billion in Anthropic, and Amazon has reportedly committed $50 billion to OpenAI. This creates a circular financing system where hyperscalers fund AI labs, which then spend that money on compute from the hyperscalers, artificially boosting revenue figures.

Analysts estimate that 73% of Amazon's AI revenues by 2026-2027 will come from OpenAI and Anthropic. Similarly, these two companies are projected to account for over 48% of Google Cloud's revenues by 2027. Without their substantial compute spend, the author suggests, the AI businesses of these tech giants would be significantly diminished, questioning the true viability of the current AI boom.