Why Broadcom Stock Crashed Today

Broadcom beat Q2 expectations with non-GAAP EPS of $2.44 versus $2.39 consensus and revenue of nearly $22.2B, while GAAP EPS rose 85% year over year to $1.91. Free cash flow increased 60% and sales grew 48%, with AI semiconductor revenue up 143% and now nearly 50% of total revenue. The company also guided Q3 sales to $29.4B, above the $28.5B Street estimate, even as the stock fell 15.2% on the news.

Analysis

The market is treating this like a classic “good-is-not-good-enough” print, but the deeper message is that Broadcom is becoming a levered proxy on hyperscaler capex intensity rather than on generic semiconductor demand. When one customer category is driving nearly half the revenue base and that segment is still compounding far faster than the rest of the company, the stock starts trading on how much of the AI infrastructure buildout can be monetized without margin leakage. That creates a paradox: near-term upside in fundamentals can coexist with multiple compression if investors fear concentration risk or a normalization of AI spending after the current build phase.

The second-order winner is Alphabet, not Broadcom. If TPU economics continue to justify custom silicon at scale, Google can potentially lower unit inference costs versus merchant GPU alternatives, widening its AI product margin advantage over time. The relative loser is any GPU-centric supplier whose valuation depends on endless share gains in AI training and inference; the street may be starting to distinguish between “AI exposure” and “AI bottleneck ownership,” which is a risk for names whose upside is now priced as if they will capture every layer of the stack.

The selloff also hints at a timing issue: investors may be discounting 6-12 months forward, not this quarter. If AI capex growth slows even modestly in the back half, Broadcom’s high-quality earnings profile could still de-rate because the stock is already being used as a crowded long in the AI trade. The contrarian view is that this pullback is more about positioning than fundamentals; if management can keep AI revenue growth near triple-digits into the next two quarters, the tape should re-rate quickly because the market is effectively underappreciating how much operating leverage sits in the mix shift toward AI silicon.

Key Decisions for Investors

• Buy AVGO on weakness over the next 1-2 trading sessions, but size modestly: use a staged entry because the first leg is likely technical. Risk/reward improves if the stock holds the pre-earnings gap area; downside is another 5-8% if momentum funds continue to de-risk AI winners.
• Pair long GOOGL / short NVDA for a 1-3 month horizon: if custom silicon economics are gaining share, Alphabet captures the cost advantage while NVDA remains more exposed to any deceleration in hyperscaler spend. This is a cleaner expression of AI infrastructure durability than a directionally bullish single-name bet.
• For event-driven traders, sell downside puts in AVGO only after volatility stays elevated for 2-3 sessions; the market is pricing a growth scare, not a balance-sheet issue. Target short-dated premium with defined risk rather than outright stock, because the fundamental floor remains strong.
• Avoid chasing NVDA into this tape unless the next hyperscaler commentary re-accelerates capex guidance. Relative performance could lag AVGO/GOOGL for several weeks if the market continues rotating from ‘AI compute scarcity’ to ‘AI monetization efficiency.’
• If AVGO rebounds 8-10% from current levels without new data, take partial profits on any tactical long: this remains a crowded positioning unwind setup, so upside can be fast but fragile.