Wall Street Analyst: Broadcom's Flat AI Guidance Justified the Selloff, but Nvidia Remains Best-in-Class Chip Play

Broadcom held its AI guidance steady at $100 billion for fiscal 2026 and fiscal 2027, but the stock still fell 12.59% after investors wanted an upward revision. KeyBanc’s John Vinh argued the selloff reflects elevated expectations, supply-chain constraints, and some share loss at Google, while raising his Broadcom target to $575 from $500. He remains constructive on Broadcom as a buy-the-disappointment idea but says NVIDIA is still the best-in-class AI chip name, with NVIDIA shares up 1.94% on the day.

Analysis

The market is penalizing AVGO less for the current print than for the implied change in the slope of expectations: when a name has been re-rated as an AI compounding vehicle, flat forward commentary reads like deceleration even if the absolute revenue base is still accelerating. That creates a tactical mismatch: the stock can de-rate faster than the fundamentals because the buyer base is using a “next leg higher” framework, not a “good quarter” framework.

The more important second-order issue is that the AI buildout is increasingly constrained by infrastructure, not silicon design. That matters because it shifts incremental value away from the chip vendor toward the ecosystem that can solve power delivery, cooling, networking, and campus bring-up; if those bottlenecks persist for 2-4 quarters, forecast risk migrates from timing to mix, which is exactly where custom-silicon narratives get fragile. It also means the market may be underpricing names with direct exposure to grid gear, liquid cooling, and high-voltage electrical equipment while overestimating the near-term monetization rate of AI capacity already sold but not yet deployed.

On GOOGL, any concern about custom ASIC share leakage is more of a margin/share-mix issue than a demand issue. Even if one hyperscaler trims Broadcom content, the broader signal is that top customers are diversifying silicon suppliers to reduce concentration risk, which cap-limits single-vendor upside but lengthens the overall custom-chip runway across the industry. That dynamic should support NVDA’s relative premium because merchant silicon remains the default solution when deployment speed, software, and platform integration matter more than bespoke optimization.

The contrarian read is that the move in AVGO may be overdone relative to the actual medium-term earnings risk. If the issue is guidance conservatism driven by power and customer timing, the stock can rebound sharply once backlog converts into shipments over the next 1-2 quarters; however, if investors keep extrapolating a straight line from announced capacity to revenue, disappointments will recur. The setup favors buying quality on weakness in NVDA and selectively fading broad AI exuberance where valuation already discounts flawless execution.

Key Decisions for Investors

• Buy AVGO on further post-event weakness only if it stabilizes for 3-5 sessions; target a 10-15% rebound over 1-3 months, but keep a hard stop if the next read-through from hyperscaler capex shows delayed deployments.
• Maintain/establish long NVDA vs short AVGO pair trade over 1-2 quarters: NVDA should retain a premium as the cleaner platform winner, while AVGO remains more exposed to expectation resets; aim for relative outperformance of 8-12%.
• Add a basket long in AI infrastructure enablers outside semis, financed by trimming AVGO: names tied to power, grid, and cooling should benefit if the real bottleneck is campus activation rather than chip demand.
• On GOOGL, treat any weakness tied to custom-silicon share concerns as an opportunity to add for 6-12 months; the risk/reward improves if the market over-discounts minor mix shifts versus the larger cloud/backlog monetization story.