Jeff Bezos Built Amazon Into a $1.8 Trillion Company. Here's How Andy Jassy Is Making It Worth Even More.

Amazon's AWS revenue grew 36.8% year over year to $42.2 billion in Q2 2026, while operating income rose 63.6% to $16.6 billion and represented 60.5% of total company operating profit. The article argues that generative-AI-driven cloud demand and AWS's 28% market share support Amazon's plan to lift 2026 capital expenditures to $220 billion from $131.8 billion in 2025. Amazon's P/E has declined to 20, below its five-year median of 50 and the S&P 500's 26, supporting a constructive long-term investment view.

Analysis

The relevant debate is not AWS demand but incremental return on the capex ramp. At the stated run-rate, Amazon must convert a materially larger share of infrastructure spend into revenue-bearing capacity without extending depreciation and power costs faster than cloud gross profit. The stock’s lower earnings multiple may therefore reflect a cash-conversion discount rather than a durable rerating opportunity; operating-profit growth can coexist with weaker free cash flow for the next 12-18 months.

Competitive dynamics favor the hyperscalers with proprietary silicon, utilization scale, and enterprise distribution. AMZN’s Trainium/Inferentia adoption is the key upside variable: successful workload migration would improve AI-unit economics and reduce NVIDIA exposure, pressuring NVDA’s premium accelerator margins at the edge. Conversely, Microsoft’s enterprise bundling and Google’s AI-model differentiation can force AWS to use pricing and credits to defend workloads, limiting the operating leverage implied by headline cloud growth.

Near term, this is unlikely to be a new information catalyst: the article is promotional and supplies no backlog, utilization, depreciation, or return-on-invested-capital evidence. Over 1-3 months, focus on AWS growth versus capex guidance and management commentary on power availability; over 6-18 months, the investment case turns on whether cloud revenue growth remains above the rate at which depreciation enters the income statement. The contrarian view is that consensus may be underestimating both the FCF trough and the risk that AI capacity becomes commoditized before it is fully utilized.

Key Decisions for Investors

• Maintain a modest AMZN long only if AWS revenue growth stays above 30% and consolidated capex guidance does not rise further at the next earnings report; target a 10-15% relative return over 6-12 months, but cut if AWS growth falls below 25% or management signals weaker utilization/longer paybacks.
• Express the competitive-risk view as long AMZN / short MSFT only after quarterly results show AWS growth accelerating faster than Azure; use a 3-6 month horizon and a tight 8-10% relative-stop, since Azure’s enterprise distribution can reverse the spread quickly.
• Do not buy AMZN solely on headline P/E. Set an alert for quarterly free-cash-flow inflection, cloud backlog/RPO growth, and depreciation growth; absent evidence that returns on new capacity are holding, treat post-results weakness as a watch item rather than an automatic dip-buy.
• For AI hardware exposure, monitor disclosed Trainium/Inferentia customer wins and AWS accelerator mix. Meaningful adoption is a 6-18 month negative read-through for NVDA pricing power; without independently disclosed utilization or customer migration data, avoid initiating an NVDA short.