On August 30, various international media sources reported that British banking giant Barclays had highlighted in an AI industry research report a striking trend: for every $100 in revenue generated by AI model companies, a substantial $35 to $40 is funneled to the three leading cloud service providers—Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform (GCP)—primarily as fees for inference computing power. From this revenue stream, cloud service providers can secure operating profits ranging from approximately $10 to $20, boasting impressive operating profit margins of 35% to 45%.

In tandem with this, the profit margins of AI labs' paid inference services have witnessed a remarkable surge. They escalated from the low double digits in 2025 to a robust 50% to 65% or even surpassing that in 2026, with adjusted gross margins experiencing a year-on-year increase of 30 to 50 percentage points.

Barclays analysts posit that the actual profit margins may exceed the report's projections. However, they also caution that as competition among cutting-edge models intensifies and the supply of computing power expands, profit margins are anticipated to gradually diminish.