Amazon’s stock could soar 50%, as this analyst says investors are looking at AI spending all wrong

Amazon plans to deploy roughly $200 billion in AI capital expenditures this year; BNP Paribas analyst Nick Jones argues these aggressive investments are necessary to meet accelerating cloud demand and says shares could rise ~50%. Some investors are alarmed by large up-front spending on data centers, but the analyst views the capex as strategic for AWS growth and potential long-term upside to company fundamentals.

Analysis

Large-scale infrastructure builds create convex optionality: near-term FCF will be depressed while capacity is absorbed, but once utilization crosses inflection (corporate AI adoption + native AWS services), incremental margin on hosted AI workloads can be 2-3x normal cloud margins because of specialized pricing and sustained GPU utilization. That path implies a multi-quarter valley in profit metrics that markets penalize, even as the long-term structural revenue stream (model hosting, fine-tuning, inference APIs) becomes stickier than commodity IaaS.

Second-order winners include GPU and networking OEMs (pricing power for high-end accelerators and top-of-rack switches), colo and power providers that can monetize constrained capacity, and software vendors that sell migration/tuning services; losers include smaller cloud players who lack scale to amortize dark-room GPU clusters and enterprises that cannot afford elastic pricing and revert to hybrid/on-prem tactics. Supply-chain effects — longer lead times for accelerators and chassis, and upward pressure on specialized labor — will raise marginal project economics for the largest buyer, widening scale advantages.

Key risks and timing: in days–weeks, earnings commentary and capex cadence can swing sentiment; in 3–12 months utilization metrics (instance mix, GPU hours sold, pricing per inference) will determine whether the investment turns into durable revenue streams; in 1–3 years, commoditization of large-model inference or a successful verticalized on-prem alternative could compress returns. Watch for unit economics disclosures ($/GPU-hour, utilization) and partner chip roadmap shifts as potential reversal catalysts.

Key Decisions for Investors

• Long AMZN via 12–18 month OTM call positions (size 2–4% portfolio): buy 30–40% OTM calls to capture a >50% upside thesis if utilization and AI revenue ramp; max loss = premium, target 2–4x on a successful adoption runway.
• Paired trade: long AMZN / short MSFT (equal notional, 6–12 months): directional play on AWS regaining pricing leverage in AI workloads; reward = 10–30% relative outperformance if AWS converts capacity to higher-margin AI services, risk = both fall on macro or MSFT’s enterprise AI wins.
• Play the supply chain: buy NVDA or ARIA/SMCI-style exposure (3–9 month calls or equity) to benefit from ongoing accelerator/network hardware tightness; target 20–40% return on persistent GPU demand, tail risk = sudden easing of supply or competitive silicon ramp.
• Short-dated hedge: buy AMZN 0–6 month put spreads ahead of quarterly capex/usage updates (small position): protects against downside from a negative utilization surprise with limited cost, capping short-term drawdown while maintaining long optionality.