Breaking Down Mag 7 Earnings: Good or Bad?

Amazon missed December-quarter EPS but reported strong core performance, with AWS revenue up 24% in 2025 Q4 (accelerating from 17–20% earlier in the year) and a backlog up 40% to $244 billion; however management unveiled a $200 billion 2026 capex plan (versus $132B in 2025 and $83B in 2024) that likely exceeds operating cash flow and sparked investor concern. The market reaction reflects elevated AI-driven capex worries rather than the EPS miss, while the Magnificent Seven cohort is still set to deliver aggregate Q4 earnings growth of ~+24.2% on +18.9% revenue growth (Nvidia est. EPS +70.8%, revs +66.7%), underscoring divergent investor focus between near-term cash flow impact and broader earnings momentum.

Analysis

Market structure: Amazon’s $200bn 2026 capex plan (vs $132bn in 2025) reallocates demand toward hyperscale datacenter builders, semiconductor suppliers (NVDA, ASML class beneficiaries) and power/utility capex; beneficiaries include NVDA (consensus Q4 EPS +70.8%) and GOOGL (Google Cloud +48% y/y), while legacy software/cloud operators with slower growth (MSFT relative lag) face pricing pressure and margin re-steals. The Mag-7 concentration (26.6% of S&P earnings, 33.5% market cap) amplifies index skew — positive earnings beats lift narrow cap-weighted indices while raising systemic risk if one large name stumbles.

Risk assessment: Tail risks include capex overspend causing 2026 FCF <0 for AMZN, potential credit-rating actions, and AI-product regulatory/efficacy shocks that could re-rate multiples; probability moderate but impact high. Time horizons split: immediate (days–weeks) — volatility and put/call skew elevated into NVDA/AMZN prints; short-term (months) — reallocation of capex beneficiaries; long-term (years) — durable moat consolidation for firms that operationalize AI efficiently. Hidden dependencies include grid/power constraints, chip lead times and spot commodity inflation (copper, electricity) that can push costs 10–20% higher for datacenter builds.

Trade implications: Favor NVDA asymmetric earnings exposure via defined-risk call spreads around Feb 25 (size 0.5–1% portfolio risk) and add 2–3% long GOOGL stock exposure over 1–12 months to capture cloud share gains. Underweight AMZN by 1–2% vs benchmark and hedge remaining exposure with OTM puts (target hedge to cover 25–50% of position) until management quantifies funding; consider pair trade long GOOGL / short MSFT (1% each) on relative cloud momentum. Fixed income: expect modest widening of IG tech credit spreads (AMZN directionally) — buy 2–3yr protection if FCF falls below capex the next two quarters.

Contrarian angle: The market may be over-penalizing AMZN’s long-term optionality — AWS backlog $244bn and accelerating revenue (+24% Q4) suggest downside is capped; buy-the-dip trigger: add to AMZN if shares fall an additional 12–15% or if management discloses capex funding that keeps 2026 capex ≤110% of operating cash flow. Historical parallel: prior cloud capex cycles (2016–2018) compressed near-term FCF then expanded margins and share gains; risk is stranded, specialized AI datacenters if models/architectures shift rapidly.

Key Decisions for Investors

• Establish a 2–3% long position in GOOGL over the next 2–6 weeks to capture Google Cloud momentum (Q4 cloud rev +48%); set a 12% stop-loss and a 30% 12-month target.
• Allocate 0.5–1% portfolio risk to NVDA defined-risk bullish call spreads expiring 4–6 weeks after Feb 25 earnings (near-the-money to 10–15% OTM) to play expected +66–71% rev/EPS growth while limiting IV risk.
• Reduce AMZN exposure by 50% relative to benchmark weight immediately; hedge remaining exposure with AMZN 6–12 month OTM puts sized to protect 25–50% of position until management clarifies 2026 capex funding (watch for debt issuance or divestiture announcements within 90 days).
• Initiate a 1% long GOOGL / 1% short MSFT pair trade on a 6–9 month horizon to express relative cloud-share reallocation; rebalance if MSFT reports sequential cloud revenue acceleration (>+5% q/q growth) or GOOGL misses cloud growth guide by >500bp.