What Amazon must do to get its stumbling stock back to record highs and beyond

Morgan Stanley added Amazon to its top picks and raised its price target to $350 from $335, implying 33% upside from Monday’s close, while Amazon shares had fallen more than 7% from their Aug. 3 record close of $284. The bullish case hinges on AWS revenue growth and profitability, Bedrock adoption, Amazon’s ability to compete in agentic AI, and robotics-driven savings; Morgan Stanley estimates high-40s% incremental AWS EBIT margins in 2027–28, versus FactSet consensus of 37.5% for AWS’s overall 2027 operating margin. Amazon’s Q3 results are due after the close on Oct. 29, with investors focused on AI monetization and whether capital-expenditure growth is beginning to moderate.

Analysis

The key risk is not whether AI demand exists, but whether AWS converts committed capacity into attractive incremental returns after power, networking, and depreciation. Long-lived servers and multi-year customer contracts support the buildout thesis, but they do not eliminate GPU obsolescence, utilization, or pricing risk; an accounting-life assumption is not proof of durable economic value. At the Oct. 29 report, AWS operating profit and capex trajectory matter more than headline revenue. Strong growth paired with rising investment and weaker margins could still disappoint.

Second-order, model-agnostic Bedrock can monetize inference regardless of which model wins, while Amazon’s custom silicon creates a partial substitution risk for NVIDIA over time. Conversely, if agentic shopping intermediaries capture discovery and transaction data, Amazon’s marketplace and retail-media economics may face pressure even if order volume holds. Blocking outside agents may protect control near term but risks pushing consumer habit formation toward Walmart, Best Buy, or Wayfair. Warehouse automation is a credible cost lever, but the benefit is a multi-year execution story, not an immediate earnings offset.

Consensus may be over-weighting AWS growth and under-weighting the quality and timing of its returns. The near-term setup is binary around earnings; over 1–3 months, guidance and utilization evidence should drive revisions. Over 6–18 months, power availability, chip refresh cycles, and robotics deployment determine whether the investment thesis compounds or impairments/cost overruns emerge.

Key Decisions for Investors

• Avoid chasing AMZN ahead of Oct. 29; consider a defined-risk 6–10-week call spread only if AWS growth is accompanied by stable/improving operating profitability and capex growth moderates. This caps premium risk while retaining upside to a positive earnings reset.
• Treat AWS margin and capex commentary as the event trigger, not revenue alone. Reassess or exit the bullish setup if AWS profit growth lags sales materially, management raises investment without evidence of contracted utilization, or guidance signals weaker returns on deployed capacity.
• Track power and data-center capacity constraints, depreciation-life assumptions, and customer contract duration/utilization in subsequent quarters. Evidence of accelerated hardware refreshes, lower utilization, or impairment risk would falsify the long-life asset thesis.
• Monitor marketplace traffic and retail-media monetization as agentic shopping adoption develops. If Amazon maintains transaction economics while opening controlled integrations, that reduces the risk; sustained share or engagement loss to Walmart, Best Buy, or Wayfair would undermine the retail offset.