Prediction: 2 Stocks That'll Be Worth More Than Amazon 3 Years From Now

Amazon remains the world’s fifth-largest company with a $2.55 trillion market cap and 2025 revenue expected to rise 12% to $715 billion (2026 +11%), but its share price has lagged (up ~6% last year). Taiwan Semiconductor (market cap $1.77T) reported 2025 revenue +36% to $122.4B and EPS +51% to $10.65, forecasts a 25% revenue CAGR through 2029 and — under a 30% EPS CAGR assumption to 2028 — could reach $23.40 EPS implying a $608 stock price via a 26x multiple (≈76% upside). Broadcom (market cap $1.67T) posted fiscal 2025 revenue +24% to $64B and EPS +40% to $6.82, carries a $162B backlog ($73B AI) with consensus suggesting ~51% sales growth this year; using a 35x forward multiple and projected $17.89 EPS implies a $626 price (~78% upside), supporting the view that TSMC and Broadcom could outgrow or overtake Amazon’s market cap within three years.

Analysis

Market structure: The winners are foundry/AI-infrastructure suppliers – TSM (TSM) and Broadcom (AVGO) – plus equipment names (ASML, LRCX) and hyperscalers that buy specialized accelerators; the loser, on a relative basis, is AMZN’s growth multiple as its sheer size caps % growth. TSMC’s guidance of ~25% CAGR to 2029 and Broadcom’s $162bn backlog ( ~$73bn AI) indicate demand > available near-term capacity, enabling price increases and margin expansion for suppliers. Cross-asset: sustained tech outperformance should be mildly risk-on, pressuring 10y yields +10–30bp in the near term, widening USD strength vs TWD/EMFX if Taiwan risk premiums rise, and lifting specialty gases/metals prices used in fabs.

Risk assessment: Tail risks include Taiwan–China escalation or tightened US export controls that could cut TSM shipments >20% in weeks, and a sudden AI demand re-rating if model training economics change (e.g., breakthrough efficiency reducing accelerator spend). Immediate (days) risk: earnings/macro headlines; short-term (months): backlog conversion and capex cadence; long-term (years): potential overbuild leading to 2027–2029 margin compression. Hidden dependencies: foundry capacity lead times (18–36 months), customer concentration (hyperscalers account for majority of AI demand), and EUV supply bottlenecks.

Trade implications: Core tactical: overweight TSM and AVGO vs AMZN — these are growth/leverage plays on AI hardware. Implement 12–36 month directional exposure via LEAPS or call spreads to capture the 60–80% upside case implied by the article, size at 2–4% portfolio each, and hedge geopolitics with small put protection. Rotate 3–6% from consumer discretionary into semiconductors and capital equipment; use options to buy upside and sell near-term overpriced IV via calendar spreads around earnings.

Contrarian angles: Consensus underweights geopolitical and cycle risks; valuations (>30x forward for AVGO) price near-perfect execution and backlog conversion—a 20% miss in revenue growth would compress multiples 15–30%. Historical parallels: 2017 GPU froth then 2018 correction shows hardware cycles can reverse fast if end-market economics shift. Unintended consequence: aggressive fab pricing could spur hyperscalers to vertically integrate or favor alternative architectures, capping long-term pricing power.

Key Decisions for Investors

• Establish a 2.5% portfolio long position in TSM (TSM) and a separate 2.5% in Broadcom (AVGO) using LEAPS (12–24 months) or buy-and-hold equity; target a 12–36 month horizon and take profits if either reaches +50–75% or if the company misses consensus revenue by >15% on two consecutive quarters.
• Initiate a pair trade: long 2% TSM / short 1.5% AMZN (AMZN) to express AI-foundry outperformance vs large-cap maturity; tighten the short if AMZN shows AWS margin expansion >200bp or raises guidance above consensus.
• Purchase 9–15 month call spreads on AVGO and TSM sized at 1% each (buy ATM or +15% OTM calls, sell +30–40% OTM calls) to capture upside while limiting premium, and simultaneously buy 6–9 month puts on TSM sized at 0.5% as geopolitical insurance (trigger add hedge if Taiwan risk premium widens >100bp).
• Reallocate 3–6% from consumer/retail exposure into semiconductor capital equipment (ASML, LRCX) and logistics/chips supply-chain names over the next 3 months; increase if TSMC utilization stays >90% and Broadcom backlog conversion >50% in the next two quarters.
• Monitor three catalysts within 30–90 days before scaling positions: TSMC monthly utilization/data points, Broadcom quarterly backlog conversion (AI revenue growth), and any new US/Taiwan export-control announcements; if two of three fall short of consensus by >10–15%, cut combined semiconductor exposure by 50% within one week.