Down 20%, Should You Buy Broadcom Stock on the Dip? The Answer Might Surprise You.

Broadcom reported Q2 revenue of $22.2 billion, up 48% year over year, with AI semiconductor revenue jumping 143% to $10.8 billion. However, management's Q3 AI revenue guidance of $16 billion fell short of Wall Street's $16.36 billion expectation, and the stock fell 20% from its peak on concerns about slowing AI hardware demand. Valuation remains elevated at 24.9x sales and 64.1x trailing earnings, limiting the case for buying the dip.

Analysis

The selloff is less about one quarter and more about the market repricing the elasticity of AI capex. When a supplier with broad exposure to hyperscalers cannot comfortably raise forward numbers, it implies buyers are getting more disciplined on incremental GPU-adjacent spend and are shifting toward ROI-gated deployments, longer qualification cycles, and more price competition among custom silicon vendors. That is a second-order negative for the whole AI hardware stack: not just accelerators, but the networking, optics, and power ecosystem that has been trading on a reflexive capex supercycle.

The near-term loser is AVGO itself because its multiple was built on the assumption that AI revenue would keep reaccelerating into a higher plateau, not merely step up once. A 200% growth print sounds huge, but the market cares about the slope of the next two quarters and whether 2027 becomes a harder comp rather than a clean air pocket. If enterprise and model vendors continue passing through higher inference costs, the pain will show first in software usage metrics, then in hardware orders with a 1-2 quarter lag.

Relative winners are the large platform owners that can self-fund custom silicon roadmaps and amortize them over enormous internal workloads, especially GOOGL. META also benefits from the same logic, though its spend is more discretionary and therefore more vulnerable if monetization stalls; MSFT is more exposed to the downstream effect because it monetizes AI through usage and could be forced to absorb margin pressure if customers balk at pricing. A subtler beneficiary is NVDA if the market starts preferring the most flexible, highest-performance general-purpose stack over custom solutions when demand becomes less exuberant.

The contrarian read is that the stock may be down enough for a trading bounce, but not enough for an investment case. The market is likely overestimating how quickly custom AI silicon can diversify away from NVDA at scale, while underestimating how quickly CFOs can slow inference consumption when bills rise. The catalyst path is binary: either AI software usage reaccelerates over the next 1-2 quarters, validating the hardware buildout, or hardware multiples compress further as spend decelerates into year-end.

Key Decisions for Investors

• Short-term: stay underweight AVGO for 4-8 weeks; the risk/reward remains skewed to another 10-15% downside if forward AI revenue revisions stop drifting higher.
• Pair trade: long GOOGL / short AVGO for the next 1-2 quarters. Thesis: custom silicon plus in-house demand control makes Alphabet less exposed to third-party pricing pressure than a supplier whose multiple still embeds peak-growth assumptions.
• Use NVDA as a relative hedge against AI capex skepticism rather than a pure long. If the market rotates from custom chips back to best-in-class general-purpose accelerators, NVDA should hold up better than the semi complex over the next 1-3 months.
• Avoid chasing MSFT on AI narrative alone until there is evidence that customer usage is reaccelerating; the cleaner expression is to wait for margin commentary to stabilize, because pricing pressure can hit earnings before revenue.
• For aggressive traders: buy a small AVGO put spread or collar into the next 30-45 days. The setup favors defined-risk downside exposure because valuation leaves little room for another guidance miss, but the oversold bounce risk is non-trivial.