Cloud providers capture up to 40% of AI model revenue

A research report from Barclays reveals that cloud giants—including Amazon AWS, Microsoft Azure, and Google Cloud—capture approximately $35 to $40 for every $100 earned by AI model companies. These providers convert that revenue share into operating profits of $10 to $20, maintaining margins between 35% and 45%.

The report highlights a strategic shift in the AI industry from focusing on model training costs to the economics of inference. As training costs are projected to drop from 96% of expenditures in 2024 to 30% by 2028, the profitability of running models (inference) is expected to surge. Paid inference margins are projected to climb from low double digits to between 50% and 65% by 2026.

Global AI lab revenue is forecasted to grow from $7 billion in 2024 to $137 billion in 2026, potentially reaching $690 billion by 2028. While cloud providers currently hold significant leverage, Barclays notes that this influence may diminish after 2028 as AI laboratories begin deploying their own proprietary compute infrastructure.