Amazon Web Services delivered a division-record 34.6% operating margin in its third quarter, up from 32.1% in the year-ago period. Revenue for the segment rose 20% year-over-year, driven by increased demand for AI model training and inference workloads. AWS also announced a backlog of contracted but unrecognized revenue totaling $200 billion, a 25% increase from the prior quarter’s $160 billion, signaling sustained customer commitment to multi-year cloud engagements.

In early January, Amazon rolled out Alexa+ Web, a browser-based AI assistant integrated at no additional cost into its Prime membership. The feature enables subscribers to draft emails, summarize documents and plan schedules directly within desktop browsers. Internal estimates suggest that the service could reduce Prime churn by up to 15%, as members increasingly use voice-driven workflows for daily tasks. Executives have indicated that successful adoption metrics will support a planned Prime fee adjustment later this year.

Amazon’s capital expenditures surpassed $75 billion in the latest fiscal year, with the majority allocated to AWS data centers and custom AI chips. The company reported that 60% of all AI inference requests are now processed on its proprietary Trainium and Inferentia silicon, reducing average per-query costs by 30% compared with third-party alternatives. Analysts project that these efficiency gains will allow AWS to maintain operating margins above 30% even as global cloud capacity scales to meet projected demand of over $1 trillion by 2028.