Here's how much Amazon stock is down since Bezos dumped $346m in AMZN shares

Amazon shares have fallen approximately 14% since Jeff Bezos sold $346 million of stock on August 3, following a rally to record highs, although the stock remains up 18.5% over six months. Investor concerns center on the FTC and 22-state lawsuit over ad-auction practices, roughly $169 billion of trailing-12-month property and equipment spending tied largely to AI and cloud infrastructure, and resulting negative free cash flow. Amazon disputes the advertising allegations and maintains that its AI infrastructure investment will strengthen long-term growth; some analysts still see up to 35% upside over the next year.

Analysis

AMZN’s issue is not AI demand but the conversion of AI infrastructure into incremental AWS operating income before depreciation, power, and financing costs reset the company’s free-cash-flow profile. The market is likely to value AWS less on revenue growth and more on capital efficiency over the next two quarters; a sustained rise in capex intensity without commensurate AWS margin expansion would compress the premium multiple relative to MSFT and GOOGL. Retail and advertising cash generation can absorb substantial investment, but that cross-subsidy lowers the visibility of segment-level returns and raises the hurdle for positive estimate revisions.

The advertising litigation is more consequential as a behavioral-remedy risk than as a near-term damages event. Auction-rule changes, greater pricing transparency, or limits on self-preferencing could reduce the opacity that supports high-margin retail-media yield; even modest ad-margin pressure matters because advertising helps fund lower-margin commerce and logistics. Second-order beneficiaries would include GOOGL and META if advertisers gain more ability to reallocate budgets, while WMT and TTD could benefit from a relative improvement in retail-media pricing transparency.

Consensus may be treating the drawdown as an insider-sale signal, when the investable question is whether management can demonstrate a declining capex-to-AWS-revenue ratio by the next two earnings reports. The more contrarian outcome is that capacity scarcity preserves AWS pricing and utilization, allowing operating leverage to appear abruptly once deployment peaks; that would make current concern about negative FCF a buying opportunity. Conversely, a weak AWS backlog/RPO disclosure, lower cloud-margin guidance, or an FTC remedy that reaches auction design would validate a longer 6-18 month de-rating rather than a routine 1-3 month correction.

Key Decisions for Investors

• Maintain a tactical underweight in AMZN versus MSFT through the next earnings print; use a long MSFT / short AMZN pair sized to neutralize mega-cap beta. The thesis is AWS capital-efficiency uncertainty versus MSFT’s clearer monetization path; cover if AMZN guides to sequential AWS margin expansion and capex growth decelerates materially.
• For long-only exposure, do not add aggressively until management discloses evidence of utilization and return on new infrastructure: AWS growth acceleration plus stable-to-higher segment margin over two consecutive reports. A break above the prior high on that confirmation would support a 6-12 month long; absent it, estimate risk remains asymmetric to the downside.
• Monitor AMZN advertising revenue growth and any FTC motion or remedy proposal as the highest-value legal catalysts over the next 1-3 months. If disclosed remedies require auction redesign or advertiser fee/pricing changes, add to the AMZN underweight and consider long TTD as a second-order beneficiary; avoid the trade if the case remains limited to monetary penalties without operational constraints.
• Set a capex watch trigger rather than buying put premium: if annualized property-and-equipment spending continues rising while AWS operating margin falls year over year, expect 6-18 month multiple compression and reassess for an outright short. The missing data are contracted AI capacity, utilization, and depreciation cadence; without them, options positioning is not sufficiently informed.