Broadcom Is Up More Than 25X in 10 Years. Can AI Drive the Next Chapter?

Broadcom’s fiscal Q3 revenue rose 85.5% to $29.591 billion, ahead of the $29.44 billion estimate, while non-GAAP EPS of $3.32 beat the $3.24 consensus; AI semiconductor revenue increased 221% to $16.70 billion. Management guided to Q4 revenue of $34.8 billion and expects AI semiconductor revenue of about $115 billion in fiscal 2027 and $230 billion in fiscal 2028, but the stock is 29% below its 52-week high amid supply bottlenecks and customer concentration concerns. The article’s model assigns a $423.79 price target versus a cited reference price of $351.19, while noting that constrained supply or weaker hyperscaler spending could impair the outlook.

Analysis

The key question is not whether AI demand exists, but whether Broadcom can convert customer plans into shipped XPUs and networking revenue while constrained inputs cap near-term deliveries. If supply—not demand—is binding, missed shipments may shift revenue across quarters rather than destroy it; however, prolonged bottlenecks could also give customers time to reconsider designs or diversify suppliers. Track delivery timing and customer commitments, not headline AI revenue targets alone.

Broadcom’s XPU opportunity pressures NVIDIA most directly where hyperscalers substitute custom silicon for workloads suited to their own chips. That is not necessarily a one-for-one loss: expanding AI workloads may support both custom accelerators and GPUs. Google’s TPU relationship reinforces the strategic opportunity but also illustrates customer concentration. A spending pause or redesign by one large buyer would therefore matter disproportionately. Marvell faces competitive pressure in custom silicon, though the article’s peer valuation comparison is not enough to establish relative mispricing; verify comparable growth, business mix, and estimates before leaning on it.

Near term, supply availability and quarterly shipment conversion are the catalysts. Over 6–18 months, customer diversification, XPU adoption, and networking attachment determine whether elevated expectations become durable earnings. The reported gross-margin decline should be monitored alongside mix and operating margin; a mix-driven dip differs from pricing deterioration. The article’s model target and confidence rating are promotional inputs, not independent evidence, and its price references are inconsistent.

Contrarian risk: the pullback may look like an attractive entry while the market is actually discounting execution and concentration risk. Falsify the constructive thesis if guidance is cut for demand rather than timing, major customers defer deployments, or margins weaken beyond explainable mix effects.

Key Decisions for Investors

• AVGO: Prefer staged exposure over chasing a headline target. Add only as reported shipments and forward guidance confirm that constrained wafers, substrates, HBM, and power are easing; reassess if guidance is reduced or customer timing slips.
• Relative-value watch: Consider long AVGO / short MRVL only after verifying comparable estimates and business mix. The thesis is Broadcom’s ability to capture custom-silicon and networking demand; cap risk because a broad AI-capex drawdown or a Marvell-specific rerating can overwhelm the pair.
• NVDA: Avoid treating Broadcom’s custom-chip growth as an automatic short signal. Monitor hyperscaler capex and workload allocation; a broad spending slowdown would hurt both, while continued GPU demand alongside TPU/XPU adoption would weaken the substitution thesis.
• Over the next 1–3 months, track supply-constrained shipment conversion, customer concentration, and margin commentary. No high-conviction options trade is warranted from this article alone; position sizing depends on verified estimates, valuation, and current market pricing.