CEO Andy Jassy’s 2025 Letter to Shareholders

Amazon reported 2025 revenue of $717B (+12% YoY) with AWS at $129B (+20% YoY) and operating income of $80B (+17% YoY); Free Cash Flow fell from $38B to $11B driven by a $50.7B YoY increase in purchases of property & equipment. Management is accelerating heavy capex (~$200B referenced for 2026) largely backed by customer commitments (including an OpenAI commitment >$100B), while AWS AI revenue run rate exceeded $15B in Q1 2026 and the chips business has a >$20B run rate (potential standalone ~$50B). Strategic operational bets—1M+ fulfillment robots, $4B rural delivery network, Amazon Leo (200+ satellites; mid-2026 commercial launch; Delta contract for 500 planes in 2028), Same-Day perishables and micro-fulfillment/Prime Air expansion—support long-term growth and ROIC despite near-term FCF headwinds.

Analysis

Amazon’s strategy to internalize key infrastructure (chips, edge connectivity, robotics, and last-mile fabrics) is shifting the competitive frontier from software/features to capital- and scale-arbitrage. That creates a two-part win: higher long-run operating leverage for Amazon as proprietary silicon and owned connectivity compress external vendor margins, and an emerging product line (racks/chip stacks + services) that can monetize that advantage outside the company. The clearest second-order consequence is price and capacity pressure in the inference GPU market — hyperscalers and large model buyers now have a credible alternative path that reduces their dependence on incumbent chip vendors.

Logistics moves (micro-fulfillment, same‑day + drones) amplify platform effects that go beyond grocery: by owning more of the fulfillment stack Amazon raises the fixed-cost barrier for delivery marketplaces and third-party grocers, shrinking unit economics for independents. Meanwhile, robotics and Leo-like connectivity create optionality to sell industrial-grade solutions and connectivity as a service — a latent revenue stream that also insulates Amazon against retail cyclicality. Expect supplier dynamics to bifurcate: foundry/advanced-node capacity and data-center power/equipment suppliers become strategic chokepoints, while commodity fulfillment OEMs face margin compression.

Key risks are concentrated-customer exposure for large capex commitments, generational execution risk on custom silicon yields, and regulatory pushback as cloud+chips+retail vertically integrate. Short-term FCF strain is the predictable outcome of front-loaded infrastructure builds but is reversible once monetization scales — the investment payoff is crystallized over multiple years, not quarters. Watch three triggers: large model customers multi-clouding away, a material Trainium yield shortfall, or an antitrust probe that forces carve-outs; each could materially rerate expectations.

Given the asymmetric payoff, the optimal play tilts long Amazon’s ecosystem while hedging vendor/adjacent exposure. Liquidity windows tied to quarterly capex disclosures and AI contract announcements will create repeatable entry points over the next 6–24 months.

Key Decisions for Investors

• Go long AMZN (core position, 12–24 months): accumulate stock on FCF-driven pullbacks sized to 3–5% portfolio. Use 2028 LEAP calls (delta ~0.6) to express convex upside while capping downside; target 40–60% upside, stop-loss at -30% from entry to respect execution risk.
• Pair trade (12–24 months): long AMZN / short NVDA in a 2:1 notional ratio. Rationale: capture Amazon’s ROIC re-rate from proprietary silicon and services while hedging market-wide AI cyclicality. Keep position small (net 2–3% portfolio) and monitor monthly GPU spot prices as a trigger to rebalance.
• Short INTC (6–12 months): tactical short of core server-CPU exposure as ARM-based Graviton adoption accelerates. Size as a high-conviction small position (1–2% portfolio); set a tightening stop at +20% to guard vs cyclical rebounds.
• Short DASH (3–9 months): tactical short or buy put spread against DoorDash to capture margin pressure in urban grocery/last-mile as ultra-fast micro-fulfillment scales. Target 30–50% return if Amazon accelerates national rollout; cap loss at 25%.
• Event hedge: buy a cheap AMZN downside put spread across next two earnings windows to protect the long exposures from an unexpected regulatory or major-customer reversal. Keep notional ~0.5–1% of portfolio as insurance.