Barclays Report Reveals AI Profit Shift Toward Cloud Giants as Inference Computing Dominates

On August 28, 2026, Barclays released a comprehensive research report on the unit economics of the AI industry, shedding light on how value and profits are distributed across the AI value chain. The findings reveal a striking trend: cloud service providers are capturing an outsized share of revenue, while AI model developers face mounting cost pressures.
According to the report, for every $100 in revenue generated by AI model companies, approximately $35 to $40 flows directly to cloud infrastructure providers—namely Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform (GCP)—as fees for inference computing. Inference, the process of running trained AI models to make predictions or generate outputs, has become the dominant cost center for AI firms as they scale their operations. This shift marks a significant transition from the earlier "training-driven" phase of AI development, where costs were concentrated in model training, to an "inference-driven" era, where ongoing operational expenses take center stage.
The report highlights that cloud providers are not merely passing through costs but are reaping substantial profits from this arrangement. For every $100 in AI model company revenue, cloud giants secure $10 to $20 in operating profit, translating to impressive operating margins of 35% to 45%. These figures underscore the pricing power and scalability of cloud infrastructure businesses, which benefit from economies of scale and the critical nature of their offerings. As AI applications proliferate across industries, the demand for inference computing continues to surge, further entrenching the role of cloud providers as indispensable partners—and formidable profit-takers—in the AI ecosystem.
Barclays analysts note that this dynamic has profound implications for the strategic priorities of AI model companies. With a significant portion of their revenue funneling toward cloud costs, these firms must carefully manage their infrastructure expenditures while continuing to innovate. The report suggests that some AI companies are exploring alternative strategies, such as developing in-house inference capabilities or partnering with specialized hardware providers, to reduce their reliance on the major cloud platforms. However, the high barriers to entry—including capital requirements, technical expertise, and the need for massive computational resources—make such diversification challenging for all but the largest players.
Looking ahead, the report projects that cloud services' share of AI-related revenue will decline to 73% by 2028, as the landscape evolves. This anticipated shift is driven by the rise of specialized AI infrastructure, including dedicated inference chips, edge computing solutions, and purpose-built data centers. These emerging alternatives promise greater efficiency and lower costs for AI workloads, potentially reshaping the competitive dynamics of the industry. Barclays analysts emphasize that while the major cloud providers currently hold a dominant position, the long-term trajectory is far from settled. The proliferation of specialized infrastructure could empower a broader range of players—from semiconductor companies to telecommunications firms—to enter the market and challenge the status quo.
The report also highlights regional variations in the adoption of AI infrastructure, with North America leading in cloud-based AI spending, followed by Europe and Asia-Pacific. Emerging markets, while smaller in absolute terms, are showing rapid growth as AI adoption expands into new sectors such as healthcare, finance, and manufacturing. This global expansion is expected to further intensify competition among infrastructure providers and drive innovation in cost-efficient solutions.
Barclays concludes that the AI industry is at a pivotal juncture. The current revenue-sharing model, which heavily favors cloud providers, may not be sustainable in the long run as competitive pressures mount and technological advancements continue. For AI model companies, achieving profitability will require a delicate balance between leveraging cloud infrastructure for scalability and investing in alternative solutions to regain control over their cost structures. For cloud providers, maintaining their profit margins will depend on their ability to innovate, differentiate their offerings, and adapt to the rapidly changing needs of AI workloads. As the industry evolves, the winners will be those who can navigate this complex landscape with agility and foresight.