Amazon Has No Dividend. Here's Why Long-Term Investors Should Own It Anyway.

Amazon has returned 66,570% over 25 years and 520% over the past decade, far exceeding Coca-Cola's 181% total return, as management prioritized reinvestment over dividends. AWS grew revenue 37% year over year in Q2 2026, its fastest pace in 18 quarters, while generating a 39% operating margin and 60% of Amazon's total operating income. Amazon expects 2026 capex of $220 billion, up 67% from 2025, to expand AI and cloud capacity against a $496 billion AWS backlog; the article argues this investment strategy makes a dividend unlikely.

Analysis

The relevant AMZN question is not dividend policy but whether incremental AI infrastructure earns above its rising cost of capital. A $220B annual capex run-rate creates a material free-cash-flow air pocket over the next 2-4 quarters even if AWS demand remains strong; valuation support therefore depends on backlog converting into revenue and, more importantly, on stable cloud operating margins. The market is likely to reward evidence of utilization and accelerating AWS revenue more than further backlog growth.

Competitive dynamics are mixed. AMZN's scale lowers unit costs for customers deploying inference workloads, pressuring smaller cloud platforms and enterprise hardware buyers that cannot match financing capacity; this is directionally negative for IBM and ORCL's lower-end cloud share, though ORCL remains a beneficiary where customers prioritize dedicated capacity. NVDA remains a near-term beneficiary of hyperscaler spend, but AMZN's Trainium adoption is the key second-order watch: meaningful internal silicon substitution would reduce the duration of NVIDIA's accelerator pricing power rather than immediately reduce units.

Consensus may be underweight the execution risk embedded in the capex step-up. If demand is supply-constrained, capacity additions can sustain a multiyear growth cycle; if enterprise AI projects remain experimental, depreciation and power/network costs arrive before revenue, compressing consolidated margins and forcing a lower FCF multiple. KO is not a meaningful direct hedge or read-through: its relative underperformance is an allocation narrative, not a tradable causal link.

Over 1-3 months, AWS growth, margin, capex guidance, and disclosed Trainium/customer-adoption metrics are the catalysts. Over 6-18 months, the investable issue is whether AWS incremental margins hold near current levels while capex-to-revenue intensity declines. A material AWS growth deceleration, weaker margin despite utilization commentary, or another upward capex reset falsifies the bullish capital-allocation thesis.

Key Decisions for Investors

• Maintain/enter AMZN long only on post-earnings confirmation that AWS growth accelerates and segment margin is stable sequentially; target a 6-12 month rerating from AI revenue conversion, but size below a core megacap position given near-term FCF dilution risk.
• Use a 1-3 month AMZN/NVDA pair: long AMZN, short NVDA in equal beta-adjusted dollars if Trainium adoption or AWS custom-silicon disclosure improves. The thesis is that AMZN captures cloud-service economics while NVDA faces duration risk; exit if AWS growth slows or NVIDIA maintains accelerating data-center revenue and gross margin.
• Do not buy AMZN solely on the dividend-versus-growth framing. Set an earnings alert for capex materially above plan without corresponding AWS revenue acceleration or stable margins; that combination warrants reducing exposure because the market will reprice AMZN on FCF yield rather than revenue growth.
• Watch ORCL as a competitive confirmation signal over the next two earnings cycles: sustained cloud infrastructure growth and booked-capacity conversion would validate broad AI demand, while a miss alongside AMZN capex escalation would increase the probability of hyperscaler overbuild.