Amazon's 39% AWS Operating Margin the Single Best Reason to Buy the Stock in September

Amazon Web Services generated 64% year-over-year operating-profit growth in Q2 2026 and a 39% operating margin, contributing roughly 60% of Amazon operating profit despite representing 21% of sales. AWS revenue is growing 37%, while management says compute capacity will remain insufficient to meet demand through 2026 and likely 2027. The article argues that Amazon's heavy data-center investment and AWS's higher-margin mix should allow companywide profit growth to outpace revenue growth over the next several years.

Analysis

The investable question is not whether AWS is profitable, but whether incremental AI capacity earns above its fully loaded cost of capital once depreciation, power, networking and customer incentives normalize. A sustained mix shift toward higher-value inference workloads could support further consolidated EBIT estimate revisions over the next 1-3 quarters; conversely, the current margin signal may partly reflect revenue recognition arriving before the associated depreciation curve. The key underwriting metric is AWS growth relative to capex growth and depreciation, not the reported segment margin in isolation.

AMZN's capacity constraint creates a near-term scarcity premium, but it also raises execution risk: delayed data-center energization, GPU/accelerator supply bottlenecks, or slower enterprise AI monetization could leave capital deployed ahead of utilization. The second-order beneficiary is NVDA if cloud capacity additions remain accelerator-intensive, though Amazon's Trainium adoption is a meaningful risk to that linkage over 6-18 months. MSFT and GOOGL are the more relevant competitive read-throughs: if their cloud growth and AI margin commentary fail to corroborate broad demand strength, the market should treat an AWS-specific acceleration as potentially customer- or accounting-mix-driven rather than secular.

Consensus appears to be extrapolating operating leverage while underweighting the lag between infrastructure spend and depreciation. AMZN can outperform if AWS backlog converts without a material rise in incentives, but valuation support becomes fragile if management signals another capex step-up without raising revenue or free-cash-flow conversion expectations. Watch AWS growth, consolidated depreciation growth, capex guidance, and any disclosed utilization or power-availability commentary at the next earnings release; a slowdown in AWS growth without a capex reduction would falsify the bullish operating-leverage thesis quickly.

Key Decisions for Investors

• Maintain a measured long AMZN into the next earnings report only if implied upside from consensus EBIT revisions exceeds the post-results volatility premium; target a 6-12 month horizon, with risk defined by AWS growth decelerating while capex and depreciation accelerate.
• Use a relative-value expression: long AMZN / short a basket of MSFT and GOOGL in equal beta-adjusted weights only if AWS growth re-accelerates while peers show stable-to-slower cloud growth. This isolates evidence of AWS-specific share gains; exit if peer cloud results validate equally strong industry demand, which removes the differentiation thesis.
• Treat NVDA as a conditional read-through rather than a direct AWS proxy. Add only on independent confirmation that Amazon's accelerator procurement is rising faster than internal-silicon substitution; Trainium adoption, lower GPU content per workload, or weaker hyperscaler capex guidance would invalidate the linkage.
• Set an alert for the next AMZN capex/depreciation outlook: a material capex increase without a corresponding AWS revenue or backlog upgrade is a signal to reduce AMZN exposure, as free-cash-flow compression can outweigh reported EBIT momentum over the following 1-3 quarters.