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

Amazon is projected to spend $220 billion on capital expenditures in 2026, up 67% from 2025, primarily to expand AWS compute capacity for strong AI demand. AWS revenue grew 37% year over year in Q2 2026, its fastest pace in 18 quarters, while its $496 billion backlog and 39% operating margin support continued investment rather than shareholder dividends. The article argues Amazon's reinvestment strategy has driven superior long-term returns, with the stock up 520% over 10 years versus Coca-Cola's 181% total return.

Analysis

The investable issue is not Amazon's lack of a dividend but whether incremental AI infrastructure spending earns above its cost of capital before depreciation and financing costs reset. AWS growth can support a higher earnings multiple only if utilization ramps quickly; otherwise, the near-term result is lower consolidated free-cash-flow conversion and pressure on retail-funded profitability. The reported backlog is directionally supportive but is not equivalent to recognized revenue, and investors should focus on quarterly capex-to-revenue, AWS operating-margin trajectory, and remaining-performance-obligation conversion.

Competitive dynamics favor suppliers with scarce AI infrastructure components—NVDA, AVGO, ANET and VRT—during the build phase, while AWS's aggressive capacity additions raise pricing and margin pressure for MSFT, GOOGL and ORCL over the next 6-18 months. The less obvious risk is that hyperscaler capacity races eventually shift bargaining power to large model developers and enterprise customers, compressing cloud returns even as AI workloads grow. Amazon's logistics and advertising cash generation make it better positioned than smaller cloud challengers to tolerate that outcome, but not immune to a capex-driven valuation derating.

Consensus appears to treat all AI capex as evidence of demand strength. A more useful 1-3 month catalyst path is evidence that capacity is monetizing: accelerating AWS growth without sequential margin erosion, improving operating cash flow after lease obligations, and management maintaining full-year capital-spend discipline. A reversal would be signaled by AWS margin falling below the prior-year range, weaker backlog conversion, or a material reduction in 2027 infrastructure commitments; those would turn the narrative from growth investment to overbuild.

Key Decisions for Investors

• Maintain a core long AMZN only on evidence of AWS revenue growth holding above 30% with stable-to-improving segment margin over the next two earnings reports; target a 12-month 15-20% upside from AI-driven earnings revisions, but reduce if AWS margin declines by more than 300 bps year over year or capex rises without operating-cash-flow acceleration.
• Express the near-term AI buildout through a pair: long ANET or VRT versus short a basket of cloud-software names with high AI expectations but limited infrastructure revenue exposure. Hold 3-6 months; this captures deployment spending while limiting broad AI-beta risk. Exit if hyperscaler capex guidance is cut or AI cluster networking lead times normalize materially.
• For a cleaner competitive relative-value position, favor AMZN over ORCL over 6-12 months: Amazon has more internally funded capacity and diversified profit pools, whereas Oracle's AI-cloud upside is more dependent on sustained external infrastructure commitments. Falsify on Oracle demonstrating sustained cloud-margin expansion and backlog conversion while AWS growth decelerates below the mid-20% range.
• Do not treat KO as a direct hedge for AMZN's non-dividend policy. If rates rise and capital-return demand rotates toward defensives, use XLP rather than KO single-stock exposure; the relevant AMZN risk is long-duration multiple compression, best monitored through real yields and hyperscaler free-cash-flow revisions.