Better Semiconductor Stock During the Crash: Nvidia or AMD?

Nvidia reported $68.1B revenue in Q4 FY2026 (ended Jan. 25, 2026), up 70% YoY with a 75% gross margin; AMD reported $10.3B in Q4 FY2025 (ended Dec. 27, 2025), up 34% YoY with a 54% margin. AMD’s CPU market share rose to 29.2% in Q4 2025, it has recent partnerships with OpenAI and Meta, and a trailing-12-month PEG below 0.5; AMD stock is up ~88% over the last year vs Nvidia’s ~46%. The author favors Nvidia for superior revenue growth and GPU/CUDA dominance but notes owning both names to capture different secular strengths.

Analysis

Nvidia’s ecosystem advantage is less about a single product cycle and more about structural control points: software APIs, validated stack partnerships, and preferential access to constrained foundry and HBM supply. That creates recurring revenue optionality (software/stack lock-in) and a two-sided scarcity effect — when hyperscalers ramp, upstream suppliers capture pricing power, which in turn limits fast-follower scale-up for competitors.

AMD’s recent wins look like the opening moves of a different path to market power: platform breadth rather than monopoly on a single workload. If AMD converts more hyperscaler partnerships into co-designed, integrated CPU+accelerator deployments, the marginal economics of datacenter purchases shift — procurement can select for integrated TCO (power, rack density, software integration) rather than peak FLOPS, compressing Nvidia’s premium over time.

Key catalysts and risks span time horizons: near-term earnings and supply cadence (weeks–months) will drive volatility; medium-term (6–18 months) product wins and hyperscaler design decisions determine share shifts; long-term (2–5 years) software portability and in-house accelerators from hyperscalers could structurally dampen GPU pricing power. The consensus underestimates how quickly a co-designed platform (CPU+accelerator) can change procurement economics; conversely it overestimates the speed at which foundry and memory supply can expand to meet AI demand without generating margin tailwinds for incumbents.

Key Decisions for Investors

• Core long NVDA (12-month horizon): buy a modest sized long-dated call spread (buy 12–15 month calls, sell higher strike) to capture continued AI capex — target asymmetric payoff (~2–3x upside vs full premium downside). Use tranche entries around earnings and any supply-driven sell-offs; set stop-loss at -30% of option premium.
• Relative-value pair (6–12 months): equal-dollar long NVDA / short AMD to profit from continued ecosystem premium while hedging market beta. Size such that short AMD exposure caps portfolio volatility; target 20–30% relative return if Nvidia re-rates versus AMD; unwind if AMD beats re-rating catalysts (partnership rollouts) in two successive quarters.
• Convex AMD call position (12 months, small allocation): buy OTM calls on AMD as a low-cost asymmetric play on platform re-rating from hyperscaler deals — expect >3x payoff if AMD secures multiple co-design deployments; risk is total premium loss, so limit to <2% of tech allocation.
• Tactical hedge via INTC or cloud accelerators (3–9 months): buy skewed puts on Intel or long short-dated puts on names likely to be displacement victims if integrated AMD deployments accelerate. Use these as hedges against a fast shift to CPU+accelerator procurement that would disproportionally hurt incumbents reliant on legacy datacenter cycles.