Nvidia Stock Was Squarely in Rally Mode on Friday. Here's Why.

Amazon reported Q4 net sales of $213.4 billion (up 14% YoY; +12% in constant currency) and diluted EPS of $1.95 (+4%), with AWS revenue rising 24% YoY to $35.6 billion and an AWS backlog of $244 billion (up 40% YoY, +22% sequential). CEO Andy Jassy disclosed plans for a record $200 billion of 2026 capex (a 51% increase), predominantly for AWS, and noted continued reliance on Nvidia GPUs (including deployment of Blackwell GPUs, Spectrum-X switches and NVLink), signaling substantial incremental GPU demand and providing a near-term tailwind for Nvidia shares, which jumped roughly 7-8% intraday.

Analysis

Market structure: Amazon’s $200B 2026 capex guidance (↑51%) materially tightens demand for datacenter GPUs and interconnects; if 10–20% of that is HW for AI servers ($20–40B) and Nvidia captures 30–50% share, that implies $6–20B incremental addressable spend into NVDA suppliers over 12–24 months. Winners: NVDA, Mellanox/Spectrum-X partners, TSMC; losers: legacy CPU vendors (INTC) and GPU secondaries if pricing falls later. Supply/demand: six-year-old A100s still in use signals persistent supply tightness near-term and strong pricing power for Nvidia for at least 2–4 quarters.

Risk assessment: Tail risks include tightened export controls or US/China policy changes that could cut sales (>20% downside for NVDA APAC revenue), abrupt AWS internalization (Graviton/Trainium displacement over 24–36 months), and TSMC capacity misses causing NVDA delivery delays. Immediate horizon (days–weeks): volatility around earnings/guide; short-term (3–9 months): order flow and inventory digestion; long-term (2–4 years): architecture shifts to custom silicon could erode share. Hidden dependency: AWS backlog ($244B) is lumpy and monetization timing will determine actual 2026 vendor demand.

Trade implications: Establish concentrated, sized bets: NVDA long exposure for asymmetric beta, paired with hedges and selective AMZN exposure to capture cloud spending. Preferred trade: long NVDA (2–3% portfolio) + long AMZN (1–2%) for 6–12 months; pair-short INTC (1% notional) to neutralize macro beta. Use options to control risk: buy 6–9 month NVDA call spreads for directional exposure and buy protective puts if exposures >3%.

Contrarian angles: Consensus prices near-term NVDA upside but underestimates AWS’s multi-year push to own silicon — downside risk over 24–36 months if Trainium/Graviton take 20–40% of AI workload. Reaction may be partially overdone: near-term pricing power could compress as second-source accelerators scale, recalling the 2017 crypto GPU cycle where demand collapsed and prices halved within 12 months. Monitor TSMC utilization, AWS quarterly capex cadence, and Nvidia Blackwell shipment cadence as decisive data points.

Key Decisions for Investors

• Establish a 2–3% long position in NVDA with a 6–12 month horizon; add on pullbacks of 8–12%, take profits if position is up 30–50%, and set a hard stop-loss at 18% below average entry or hedge with 3–6 month puts (strike ~10–15% OTM).
• Initiate a 1–2% long position in AMZN to capture AWS monetization of capex; trim to zero if AWS revenue growth falls below 15% y/y for two consecutive quarters or if backlog growth stalls sequentially.
• Construct a relative-value pair: long NVDA (1.5%) / short INTC (1.0%) equal notional for 6–12 months to express GPU vs legacy CPU divergence; unwind if INTC secures material hyperscaler contracts or NVDA misses guidance by >10%.
• Use options for controlled exposure: buy a 6–9 month NVDA call spread (debit) sized to 0.5–1.0% portfolio risk to express upside, and buy 3–6 month protective puts if net NVDA exposure exceeds 3%; alternatively sell short-dated covered calls to monetize near-term vol if long.