Amazon hasn't been this cheap since 2008. Here's how the stock could rally 46% from here.

Jefferies argues Amazon could rally ~46% from current levels despite $200B in planned AI spending that has weighed on the stock. Shares are down 11% in 2026 and are at valuation levels not seen since 2008, but the analyst cites rising data-center capacity and strong retail performance as evidence fundamentals remain intact. Brent Thill calls the market negativity "extreme" and describes the shares as "mispriced, not broken."

Analysis

Think of the current weakness as a timing mismatch between near-term margin optics and multi-year infrastructure lead-taking. Heavy investment in specialized AI infra will create a stepped-up replacement cycle for GPUs/CPUs, storage and networking over 12–36 months, boosting revenue for equipment suppliers and colo landlords while compressing the buyer’s free cash flow in the interim. That dynamic also raises variable opex (power, cooling, fiber) and lifts adjacent capex in logistics and fulfillment as compute-enabled commerce features (real-time personalization, visual search) increase throughput and returns per customer over the medium term.

Leading indicators to watch are order pacing and vendor backlog for datacenter components (quarterly spending at LRCX/AMAT, server ODM bookings), colo utilization rates, and AWS product telemetry (new AI-managed services monetization, inference v. training mix). Near-term catalysts that would re-rate the stock are visible margin improvement or accelerating ARR from AI services within 2–4 quarters; conversely, missed efficiency targets or sustained unit-cost increases for inference would force multiple compression over 6–12 months. Regulatory or partner friction (enterprise reluctance to port models, antitrust interventions) represent multi-year tail risks that would materially slow payback on this investment.

Consensus underestimates the optionality: front-loaded capex can erect durable moat advantages if Amazon converts that infrastructure into differentiated, high-margin managed AI services and sticky enterprise contracts. That path is binary — if monetization works, upside is large; if it doesn’t, the market is already pricing some of that downside. Trade books should express asymmetric exposure: capture upside from mean-reversion in sentiment and structural cloud dominance while limiting blunt exposure to execution or macro shocks.

Key Decisions for Investors

• Overweight AMZN equity (size 1.5–2% net portfolio) with a 12–24 month horizon; pair with a low-cost hedge by buying 18-month AMZN puts 20–25% OTM to cap tail loss. R/R: asymmetric upside (30–40%+) if AI services monetize; limited downside to hedge cost if execution fails.
• Buy suppliers of datacenter buildout — LRCX and AMAT (equal-weighted, tactical 3–9 month trade). Rationale: vendor backlog leads revenue in quarters; target 15–25% upside if server refresh cadence accelerates; risk: enterprise capex pullback could erase gains.
• Relative-value pair: long AMZN / short NVDA (small size, 6–12 months) to capture mean reversion between cloud platform value capture and semiconductor multiple rerating. R/R: target 20% relative return; cut losses if pair moves >15% against position.
• Options alternative for asymmetric upside: buy a 12–18 month AMZN call spread (buy longer-dated call / sell nearer-dated call) to reduce cost basis and target ~2–4x return if cloud AI monetization accelerates within 12 months. Keep position small to manage vega and idiosyncratic earnings risk.