SpaceX vs Amazon: Which Is the Better Stock to Own Over the Next 10 Years

The article favors Amazon as the stronger long-term investment, citing its durable cloud model, multiyear AI customer contracts, custom-chip strategy, and $200 billion of quarterly revenue. SpaceX leads in space infrastructure, with more than 11,000 operational Starlink satellites versus Amazon's roughly 400, and projects $100 billion in ARR by year-end after reporting $7.8 billion in Q2 2026 revenue. However, SpaceX's projected growth carries substantial capex and profitability risk, while its short-term cloud rental economics may be less sustainable than Amazon's.

Analysis

The investable distinction is duration and capital intensity, not exposure to AI or LEO. AMZN can fund infrastructure from retail and AWS cash generation while monetizing customers through a broad software, database, security, and managed-services stack; incremental accelerator capacity therefore has a credible path from utilization to recurring gross-profit expansion. SPCX's apparent near-term compute returns should be discounted until contract duration, customer concentration, power costs, and committed capex are independently disclosed: short-duration scarcity rents can reverse quickly as hyperscaler and neocloud capacity comes online over the next 12-24 months.

AMZN's custom-silicon adoption is the more important medium-term swing factor than its satellite initiative. Higher Trainium/Inferentia mix can improve AWS contribution margins and reduce dependence on NVDA supply, but aggressive pricing needed to win AI workloads could delay that benefit; watch AWS revenue growth versus segment operating-margin progression through the next two earnings cycles. NVDA remains a volume beneficiary under either outcome, although any meaningful shift to proprietary accelerators would pressure the market's assumption that all AI infrastructure spend converts one-for-one into premium GPU demand.

The satellite-related partnerships are strategically useful but unlikely to be material to AMZN, T, AAPL, or GSAT valuations in the next 1-3 quarters. The non-obvious risk is that terrestrial cloud economics subsidize connectivity expansion, making standalone satellite broadband providers structurally less able to earn their cost of capital; spectrum ownership only acquires major value if device integration, regulatory permissions, and a scalable wholesale pricing model all develop. Consensus may be overpaying for SPCX scarcity and underappreciating AMZN's option value from bundling connectivity, cloud, and enterprise distribution, but this remains a 6-18 month thesis rather than a near-term earnings driver.

Key Decisions for Investors

• Initiate/maintain long AMZN versus short SPCX as a 6-12 month relative-value position, sized modestly given SPCX's momentum and limited public-history risk. Thesis requires AWS growth to remain resilient while AWS margins stabilize or expand; cover the short if disclosed SPCX contracted backlog and free-cash-flow conversion materially exceed expectations or if AMZN guides to sustained AWS margin compression.
• Add AMZN on post-earnings weakness only if AWS growth decelerates less than 300 bps sequentially and management confirms rising proprietary-chip usage; target relative outperformance versus the Nasdaq over 12 months. The key falsifier is a second consecutive quarter of margin erosion driven by AI pricing or capex without corresponding backlog conversion.
• Do not treat T, GSAT, or AAPL as primary expressions of satellite-connectivity upside yet. Create an alert for disclosed wholesale subscriber economics, handset activation data, and regulatory approvals; without those inputs, the likely financial contribution is too small to justify a directional position.
• Maintain NVDA exposure but hedge concentrated AI-infrastructure books with a 6-12 month NVDA underweight versus AMZN if hyperscaler disclosures show accelerating custom-ASIC mix. Reverse that hedge if cloud customers prioritize time-to-deployment over cost and NVDA supply constraints persist into 2027.