Amazon Expects AWS Annual Revenue to Hit $600 Billion in 10 Years. The Stock Looks Like a Dirt Cheap Buy

AWS revenue guidance was raised to as much as $600 billion annually by 2036 (previously $300 billion), driven by AI demand, positioning AWS as a major growth engine for Amazon. Amazon's latest quarter grew 14% year-over-year; the stock trades at ~29x trailing earnings and ~26x forward P/E, supporting a bullish long-term growth case tied to AI, robotics and custom chips.

Analysis

AWS becoming the dominant runtime for large models creates a two-tier cloud market: hyperscalers that control optimized stacks (infrastructure + model ops) will capture high-margin recurring revenue, while mid-tier clouds and on-prem vendors face accelerating commoditization and margin erosion over 3-7 years. That means disproportionate capex and supply-chain demand for accelerators and custom silicon — a structural multi-year tailwind for firms able to scale fabs and platform partnerships, and a headwind for general-purpose CPU incumbents unless they pivot to accelerator ecosystems.

At the margin, AWS scale shifts bargaining power upstream: large enterprise buyers will pay for co-location and specialized services (fine-tuning, data pipelines, inference-at-scale) but will also pressure ASPs through long-term committed usage contracts, compressing spot and smaller-customer pricing. Regulatory and geopolitical shocks (export controls, Taiwan tensions, or energy price spikes) remain the largest single-event risks to the capex+GPU supply chain and could flip the narrative within quarters, not years.

Near-term catalysts to monitor are monetary-pressure on enterprise IT spend and hyperscaler pricing actions over the next 6-12 months, plus AWS-specific product cadence (train/infer SKUs) that will reveal margin mix. The consensus currently understates the speed at which model optimization software (quantization, sparsity) could reduce GPU-hours per inference by 20–40% within 18 months — a deflationary force that could blunt hardware revenue upside even as cloud services grow.

Finally, the asymmetric payoff is in differentiated optionality: owning exposure to the cloud-native value chain (platform + accelerator beneficiaries) while shorting legacy, scale-sensitive incumbents that lack a clear path to capture AI-specific infrastructure economics is the clearest way to express conviction without betting solely on headline revenue numbers.

Key Decisions for Investors

• AMZN long via 18–30 month call spread: buy AMZN 18–30 month 10–20% OTM calls and finance with nearer-term 0–12 month calls to create a calendar/call-spread. Size 1–2% NAV. R/R: asymmetric upside if AWS captures enterprise model hosting; capped downside limited to premium paid if retail/adjacent segments disappoint.
• NVDA targeted 3–9 month call spread to play GPU demand for inference: buy 3–9 month slightly OTM call spread (cost-defined) sized 0.5–1% NAV and hedge with a small put position to limit delta into a broad market selloff. R/R: high probability of >2x if hyperscaler procurement accelerates; tail risk is rapid model optimization reducing GPU-hours.
• INTC short or buy 9–18 month puts (or sell near-term strength): express downside from failure to transition to accelerator-led data centers. Size modestly (0.5–1% NAV) and hedge with a concentrated long in NVDA to form a dispersion pair. R/R: asymmetric if server CPU share collapses vs accelerators; risk if Intel announces a credible accelerator pivot or wins large design wins.
• Relative pair: long AMZN / short WMT (12–24 months): overweight AWS-driven secular growth vs retail margin pressure as a low-correlation hedge to consumer cyclicality. Size 1–2% NAV. R/R: captures divergence between cloud-driven valuation re-rating and retail's slower secular trajectory; risk is a surprise retail recovery or macro shock compressing multiples across both.