Broadcom Drops 10% in 3 Months: Buy, Sell or Hold the Stock?

Broadcom shares have fallen 9.7% over three months as investors weigh AI-driven gross-margin compression, supply bottlenecks, customer concentration and weak non-AI operations against exceptional AI revenue growth. Q3 FY2026 revenue rose 86% year over year to $29.6B, while AI semiconductor revenue jumped 221% to $16.7B and is forecast to reach $21.7B in Q4. Management projects AI revenue of roughly $58B in FY2026, $115B in FY2027 and $230B in FY2028, but gross margin is expected to decline to about 73% in Q4 from 78% a year earlier; Zacks rates AVGO Hold given its 10.34x forward sales valuation.

Analysis

The key valuation issue is not AI revenue growth but incremental economics: if custom silicon continues to displace higher-margin legacy semiconductor content, AVGO can deliver large EPS growth while its earnings-quality multiple compresses. The market will increasingly value AI revenue by gross-profit dollars rather than headline sales, making each quarterly margin guide more important than the revenue beat. In the next 1-3 months, a stable margin floor and evidence that operating expenses remain structurally disciplined are required for the stock to close its relative-performance gap.

Customer concentration creates asymmetric downside because a single hyperscaler deployment delay can shift revenue across quarters while fixed supply commitments remain in place. Conversely, the broader custom-ASIC buildout is a competitive headwind for NVDA at the workload level, but AVGO's networking franchise is partly insulated because Ethernet switching content can attach to both proprietary accelerators and GPU clusters. Optical-component bottlenecks make Coherent (COHR), Lumentum (LITE), and Fabrinet (FN) potentially cleaner second-order beneficiaries if AI cluster build schedules remain supply-constrained.

The contrarian view is that margin dilution may be less damaging than feared if custom accelerators deepen customer lock-in and create recurring networking, connectivity and software pull-through. That thesis is only investable if AVGO demonstrates that gross-margin pressure has bottomed while AI growth broadens beyond a small group of buyers; otherwise, the current premium leaves limited room for a routine execution miss. Over 6-18 months, power availability and data-center construction—not chip demand—are the more probable gating variables, favoring a selective infrastructure basket over a concentrated AVGO bet.

Key Decisions for Investors

• Maintain AVGO at neutral/underweight into the next earnings print; accumulate only if management establishes a credible gross-margin floor and preserves forward operating-margin guidance. Falsify a constructive view if gross margin falls below the guided low-70%s or if a major customer defers deployment.
• Run a 1-3 month equal-dollar pair: long NVDA / short AVGO, sized market-neutral. NVDA has cleaner accelerator economics, while AVGO remains exposed to mix-driven multiple compression; target 8-12% relative return, with a stop if AVGO's gross-margin guide stabilizes and AI revenue materially exceeds guide.
• For AI infrastructure exposure, build a small basket long COHR, LITE, and FN rather than chasing AVGO beta. The setup benefits from optical scarcity irrespective of whether compute demand is served by NVDA GPUs or AVGO-led custom silicon; reassess if hyperscaler capex guidance weakens or optical lead times normalize.
• Watch GOOG, META, AMZN, MSFT, and ORCL capex commentary for deployment timing rather than aggregate AI spend. A shift from announced capex to delayed data-center energization is the earliest warning that AVGO revenue conversion—not demand—will miss expectations.