Amazon shares have room to run as AWS AI backlog hits $35B, UBS says

UBS says Amazon is underappreciated as AI commitments build in its cloud division, highlighting roughly $35 billion of Bedrock backlog as of the end of Q1 2026. The bank sees that backlog as about 300% year-over-year growth and reiterated a Buy rating with a $333 price target. The note is constructive for Amazon shares, but it is analyst-driven rather than a fresh company disclosure.

Analysis

The key mispricing is not that Amazon has AI exposure, but that the market still treats its cloud AI layer as optionality rather than a durable annuity. A backlog of this scale implies visible demand that can support multiyear capacity expansion, and the second-order winner is likely Amazon’s infrastructure ecosystem: networking, power, chips, and data-center buildout vendors should see tighter utilization and better pricing power as AWS prioritizes inference and model-hosting workloads over lower-margin compute.

The competitive takeaway is more interesting than the headline suggests. If Bedrock demand is truly compounding at this pace, AWS can close part of the narrative gap vs. peers by converting AI usage into enterprise workflow lock-in, which raises switching costs and lowers churn. That creates a flywheel where more enterprise data and model deployment on AWS improves product stickiness, even if near-term margins are pressured by capex intensity and accelerated depreciation.

The main risk is that backlog quality may be overstated: AI commitments can be lumpy, pilot-heavy, and subject to renegotiation if customers fail to monetize usage. Over the next 1-3 quarters, the stock can rerate on evidence of conversion rates and margin discipline; over 12-24 months, the bigger risk is supply-side bottlenecks in power, GPUs, and data-center delivery limiting revenue realization even if demand stays strong.

Consensus may still be underpricing the duration of this cycle. The market is focused on near-term margin dilution from AI investment, but the more important variable is whether Amazon can turn capex into multi-year contracted demand at scale. If that conversion is visible in coming quarters, the multiple can expand before earnings catch up, making this a story where the rerating likely leads the fundamentals by several months.

Key Decisions for Investors

• Initiate/ add to AMZN on pullbacks over the next 1-3 weeks; use a 6-12 month horizon and underwrite to multiple expansion driven by AI backlog visibility rather than near-term margin expansion.
• Pair trade: long AMZN / short a slower-growing cloud incumbent over 3-6 months if you want exposure to AI demand conversion with less downside from capex skepticism; the spread should benefit if enterprise AI spend proves stickier than expected.
• Buy AMZN call spreads 6-9 months out to express upside without paying full premium for a rerating; target a scenario where the market starts capitalizing AI backlog as recurring revenue.
• If you want a second-order expression, go long select data-center/power infrastructure beneficiaries on any confirmation of AWS capacity buildout; the trade works best if AWS stays supply-constrained for 2-4 quarters.