Amazon's AWS Backlog Just Hit $496 Billion: Is the Stock a Screaming Buy Under $270?

AWS backlog rose to $496B in Q2, implying nearly ~12 quarters (~3 years) to churn through revenue at the current $42.2B quarterly run rate. Amazon plans $220B of capital expenditures this year, largely for data centers, and CEO Andy Jassy indicated it may not fulfill cloud demand in 2026 (and likely 2027), with 2028 demand already appearing. AWS generated ~60% of Amazon operating income on ~21% of sales, reinforcing AWS as the core profit engine as AI-driven compute demand accelerates.

Analysis

The key market read-through is not just that AWS demand is strong; it is that supply is the binding constraint. That shifts the debate from “how fast can AWS grow?” to “who captures the overflow while Amazon races to add capacity?” In the near term, that is a relative tailwind for MSFT and GOOGL if enterprise workloads are forced to dual-source or migrate to available capacity, while infrastructure vendors tied to hyperscale buildouts — NVDA, ANET, VRT, DLR, EQIX — should keep seeing budget priority.

For AMZN, the backlog is bullish for revenue durability but not automatically for near-term earnings quality. Heavy capex can suppress FCF and keep reported margins noisy for several quarters, especially if power, land, and equipment bottlenecks delay monetization. The market may be underestimating the second-order effect: a constrained AWS can support pricing discipline and reduce churn, but it also caps upside if customers move incremental AI workloads elsewhere first.

The contrarian risk is that backlog headlines overstate hard demand. Some portion may be reservations rather than fully fungible revenue, so the falsifier is any sign that AWS growth decelerates even as capex stays elevated, or that backlog starts falling without an improvement in capacity-to-revenue conversion. Over 6-18 months, the real bull case is not multiple expansion on “AI hype,” but sustained utilization that converts capex into structurally higher operating income.

At current levels, the best expression is relative: long AMZN on pullbacks versus a basket of slower-growth megacap software, or long NVDA / ANET on the buildout cycle while watching for margin pressure at AMZN if capex outruns monetization. If AWS commentary on the next two quarters implies demand is still unmet into 2027, the trade remains intact; if Amazon begins discussing normalization in fulfillment timelines, the scarcity premium in the stock should compress.

Key Decisions for Investors

• Long AMZN on weakness only if the market overreacts to capex intensity; view as a 6-12 month compounder, but require evidence that AWS operating margin holds while revenue accelerates. Falsifier: AWS growth slows or capex guidance rises without corresponding backlog conversion.
• Pair trade: long NVDA / short AMZN into any post-earnings valuation compression in AMZN. Thesis: the shovel-makers monetize hyperscaler spend immediately, while AMZN bears execution and depreciation risk over 2-4 quarters.
• Watchlist relative value: long MSFT vs short AMZN if AWS capacity constraints are causing enterprise spillover to Azure. This is a 1-3 month catalyst trade if cloud checks show customer workload re-routing.
• Long infrastructure beneficiaries basket (ANET, VRT, DLR/EQIX) on any confirmation that hyperscaler capex stays >$200B. Risk/reward is favorable if power, networking, and colocation remain the bottlenecks rather than semiconductor supply.
• No standalone options trade unless implied vol is unusually cheap: the signal is structural, not event-driven. Use stock or basket positioning rather than short-dated calls if entering now.