Wall Street Is Wrong About These 2 Chip Stocks

AMD reported Q2 2026 revenue growth of 50% to $11.5 billion and non-GAAP EPS growth of 245% to $1.66, exceeding consensus, while data-center revenue surged 107% to $6.7 billion. Broadcom posted Q3 sales growth of 86% to $29.6 billion and 96% non-GAAP EPS growth to $3.32, and guided for AI revenue to double to $100 billion in 2027. The article argues that recent stock volatility reflects investor concerns over AI-spending durability despite forecasts for global AI infrastructure investment to exceed $1 trillion next year, up from $750 billion this year.

Analysis

The investable distinction is not simply AI-beta: AVGO monetizes hyperscaler concentration through bespoke silicon and networking content, while AMD must prove that incremental accelerator demand converts into durable share gains rather than lower-priced capacity filling. A rising CPU-to-accelerator mix would favor AMD versus NVDA at the compute layer, but it also expands the bottleneck into memory, interconnect and power—supporting AVGO’s switching franchise and, secondarily, VRT. Conversely, custom ASIC adoption is a structural negative for NVDA’s pricing umbrella even if total AI capex remains elevated; the relevant question is whether ASICs displace GPU workloads or unlock incremental inference demand.

Near-term volatility should remain driven by order visibility, not reported growth. The key 1-3 month catalyst is whether hyperscaler capex guidance and supplier backlog commentary validate multi-year deployment schedules; company TAM and revenue targets are not independently sufficient evidence of end-demand. For 6-18 months, AVGO’s risk is customer concentration and program timing: a single delayed tape-out, model architecture change, or customer insourcing decision can create a sharp revenue-air-pocket despite long contractual relationships. AMD’s risk is more direct—gross-margin expansion and accelerator/EPYC mix must outpace expectations to sustain a premium multiple.

Consensus appears too binary on an "AI spending slowdown." A moderation in aggregate capex can coexist with stronger custom-silicon and infrastructure content share, making AVGO relatively insulated versus merchant compute vendors. The contrarian bearish case is that the market is already capitalizing years of revenue before deployment utilization is proven; weak cloud monetization would compress multiples across the stack before chip revenue visibly declines.

Key Decisions for Investors

• Favor AVGO over AMD in a 3-6 month relative-value position: long AVGO / short AMD, beta-adjusted. Target 15-20% relative outperformance with a 7-10% relative stop; close if AVGO discloses ASIC program deferrals, or if AMD delivers materially better-than-expected data-center gross margin and forward accelerator commentary.
• Maintain a modest long AVGO core only after validating next-quarter backlog, customer prepayment/deposit trends, and networking attach rates. Scale exposure rather than chase a sentiment-driven rebound; the upside case requires guidance durability, while a missed custom-silicon timing milestone can produce disproportionate downside.
• Use AMD as an earnings-quality watch item rather than an outright long until management quantifies CPU versus GPU revenue, gross margin, and customer concentration. Initiate only if server share gains are accompanied by sustained margin expansion; revenue growth funded by lower pricing would falsify the bullish rerating thesis.
• Hedge AI compute exposure with a small long VRT or infrastructure basket against a short NVDA overlay only if evidence emerges that ASIC inference deployments are replacing merchant GPU clusters. The trade benefits from a shift in AI spend composition, but should be exited if hyperscalers signal that ASICs are incremental capacity rather than substitution.