The booming artificial intelligence sector is revealing a unique financial dynamic: the very cloud providers investing massively in AI infrastructure are increasingly dependent on AI laboratories as their primary customers. This creates a powerful, self-sustaining cycle of spending, revenue, and reinvestment within the AI economy.
The Intertwined Financial Loop
According to a recent report by Jefferies’ Global Head of Equity Strategy Christopher Wood, the relationship between hyperscale cloud providers—Microsoft, Amazon, and Google—and frontier AI labs like OpenAI and Anthropic has become significantly intertwined. These tech giants are pouring enormous capital into building the computational backbone required for advanced AI, only for the leading AI companies to become their most significant consumers of these cloud services.
In the second quarter of 2026, the combined cloud revenues for Microsoft, Google, and Amazon hit an impressive $126 billion, marking a 38% year-on-year increase. However, a substantial and growing portion of this revenue is now directly attributable to AI customers.
AI Labs' Growing Cloud Dependence
- Amazon Web Services (AWS): OpenAI and Anthropic together accounted for approximately 6% of AWS’s total cloud revenue in Q2 2026, including non-AI services. This share is projected to rise to 12-18% in Q3.
- Google Cloud Platform (GCP): The two leading AI labs were estimated to represent roughly 45% of Google Cloud Platform’s revenue in Q2 2026.
- Microsoft Azure: For Microsoft’s Azure, OpenAI and Anthropic constituted about 25% of its Q2 2026 revenue.
The $2.34 Trillion Backlog and Future Risks
The scale of this interdependence is further highlighted by future contracted revenues. The major hyperscalers' combined remaining performance obligations (RPOs) reached a staggering $2.34 trillion at the end of Q2 2026, a massive 186% jump from $816 billion just a year prior. Jefferies estimates that OpenAI and Anthropic alone account for approximately 45-50% of Amazon’s RPO backlog, 40% of Google’s, and 30-40% of Microsoft’s.
This means a significant portion of the future cloud revenue justifying today's colossal AI infrastructure investments is derived from these very AI companies. While this financial model is not inherently unsustainable—AI labs are generating rapidly increasing revenues, with OpenAI and Anthropic's combined annualized revenue run rate estimated at around $115 billion—it introduces a new layer of risk.
If AI labs continue their rapid growth, hyperscalers can justify escalating infrastructure investments. However, a slowdown in AI demand could exert pressure from both sides: AI companies might reduce their infrastructure consumption, leaving hyperscalers with immense capital costs and potentially underutilized resources.