Amazon's AWS expansion could drive potential revenue upside, analysts say

Bank of America analysts say AWS’s aggressive capacity build—an estimated ~15 gigawatts added by 2027 after doubling power capacity since 2022 and adding 3.9 GW in 2025—could support AWS revenue growth of roughly 27% in 2026 and 28% in 2027. Assuming stable revenue per gigawatt, BofA models $164 billion in AWS revenue in 2026 and $209 billion in 2027, implying ~2% and ~5% upside to consensus, and keeps a Buy rating with a $275 target (current ~$209). The note flags sector risks from an estimated $1.2 trillion hyperscaler capex cycle that could raise input costs and create overcapacity and pricing pressure, but cites new AI deals, Trainium adoption and government/OpenAI agreements as demand catalysts.

Analysis

Market structure: AWS’s planned ~15 GW incremental capacity by 2027 and Bank of America’s $164bn/$209bn revenue scenario imply Amazon (AMZN) is positioned to take share in AI infra and gov cloud work—direct winners include AMZN, OpenAI (counterparties) and suppliers of custom chips/servers; losers are smaller cloud specialists and legacy on‑prem vendors facing pricing pressure. A 27–28% AWS growth cadence in ’26/’27 versus Street suggests AWS pricing power could improve short‑term, but industry capex (~$1.2T) points to longer-term downward pressure on yields if utilization lags.

Risk assessment: Tail risks include an antitrust or national security restriction on large AI contracts, a disorderly overcapacity-driven price war, or a Trainium performance setback; any of these would compress multiples >20% fast. Time windows: immediate (days) for sentiment moves, short (3–6 months) for re‑rating as Q1/2 disclosures hit, long (12–24 months) for capacity/utilization to manifest; hidden dependencies include grid capacity, transformer lead times, and foundry/GPU supply bottlenecks that can flip revenue/GW math.

Trade implications: Tactical: establish a 2–3% long position in AMZN equity (target $275 in 6–12 months, stop-loss 12% below entry) and complement with a 9–12 month call spread (buy Jan‑27 230C / sell Jan‑27 320C, size 0.5–1% notional) to lever upside at defined cost. Relative value: pair trade long AMZN vs short SNOW (equal dollar 1–2% each, horizon 3–9 months) to capture scale benefits versus higher‑multiple data incumbents. Commodities/utilities: add 1–2% exposure to copper (FCX) or grid/renewables developers to hedge rising power/copper demand from data‑center builds.

Contrarian angles: Consensus understates both downside from capex‑driven input inflation and upside from proprietary Trainium LLM acceleration—if Amazon’s LLM traction improves revenue/GW, upside could exceed BofA’s 5% 2027 delta. Historical parallels (2010s hyperscaler builds) show consolidation favored lowest‑cost operators; unintended consequences include local permitting chokepoints and higher utility tariffs that could bifurcate winners/losers more than current models assume.

Key Decisions for Investors

• Establish a 2–3% position long AMZN equity within 1–2 weeks; target price $275 within 6–12 months, set stop‑loss at 12% below entry to limit drawdown if capex concerns re‑rate the name.
• Buy a 9–12 month AMZN call spread to lever upside with defined cost: buy Jan‑27 230C / sell Jan‑27 320C (size 0.5–1% of portfolio).
• Initiate a relative value pair: long AMZN (1–2%) and short SNOW (1–2%) equal dollar notional, horizon 3–9 months to exploit scale/valuation asymmetry if pricing competition tightens.
• Add 1–2% exposure to copper/renewables beneficiaries (e.g., FCX or utility/renewables developers) to hedge rising data‑center power and conductor demand; trim if copper rallies >25% or grid permitting improves materially.