Better Super Semiconductor Stock: Advanced Micro Devices (AMD) or Broadcom?

AMD and Broadcom are both benefiting from surging AI hardware demand, with AMD's data center revenue up 57% to a record $5.8 billion in Q1 2026 and Broadcom's AI products up 106% to $8.4 billion in fiscal Q1 2026. Management expects further acceleration, including AMD's potential 80% AI revenue growth in 2027+ and Broadcom's AI growth rate rising to 143% in Q2. The article argues Broadcom is cheaper on valuation at 56.9x P/E versus AMD at 97.5x, but AMD may offer greater upside if it gains more AI market share.

Analysis

The important second-order read-through is that AI capex is no longer just an Nvidia story; it is broadening into a multi-vendor stack where custom silicon and networking can compound faster than the headline GPU franchise. That matters because the spend profile is shifting from one-off accelerator purchases toward multi-year platform commitments, which improves revenue visibility for AVGO and, to a lesser extent, AMD. The beneficiaries are the chip designers with system-level control, while the pressure point is anyone exposed to generic server content or merchant silicon with less differentiation.

AMD’s setup is the higher-beta expression: smaller base, larger optionality, but also more execution fragility. If its next-gen platform lands with real customer conversion, the upside is not linear — it can re-rate from “alternative supplier” to “credible second-source ecosystem,” which would force hyperscalers to diversify procurement and soften Nvidia’s pricing power. The risk is that any performance or yield slippage in the new generation immediately defers revenue into 2027, making the valuation look stretched on a near-term multiple basis.

Broadcom is the cleaner monetization vehicle because custom accelerators and networking reinforce each other: more clusters deployed means more switches, optics, and rack-level complexity, which widens the moat and reduces the probability of pure-component substitution. The market may still be underestimating how much of AI infrastructure spend is migrating from accelerators to interconnect as model training scales; that supports AVGO’s mix and margin durability. The main contrarian risk is that consensus is extrapolating AI growth rates too smoothly — a digestion quarter or customer capex pause could hit AMD harder, while AVGO should still hold up better due to contract structure and broader attachment revenue.

The broader loser set is less about named competitors and more about any supplier dependent on undifferentiated AI server buildouts. If hyperscalers decide custom silicon is the economic answer, merchant GPU vendors, lower-end networking, and generic ODMs could see margin compression as procurement becomes more vertically integrated. Over the next 6-12 months, the trade is less about absolute demand and more about who captures the economic rent from AI deployment.

Key Decisions for Investors

• Long AVGO vs. AMD on a 3-6 month horizon: the pair favors AVGO’s lower valuation, stronger customer diversification, and network attach; target 15-20% relative outperformance with lower earnings-date volatility.
• Buy AMD only on pullbacks or post-launch confirmation: use a staged entry into the next platform ramp, with upside tied to 2027 revenue inflection but downside from any slip in customer adoption or product timing.
• Initiate AVGO Jan-2027 call spreads to express continued AI infrastructure compounding with defined downside; attractive if AI revenue growth remains above 100% through the next two quarters.
• Consider a short basket of generic AI-infrastructure adjacencies versus long AVGO as a hedge against capex digestion; thesis is that differentiated silicon and switching retain spend even if server orders slow.
• Do not chase NVDA as the default AI beta here; relative value now favors second-order infrastructure winners where incremental AI spend still has room to re-rate economics.