Broadcom: 5 Reasons To Buy This Dip

Broadcom's custom AI accelerators and networking silicon are driving rapid growth, supported by co-design partnerships with six frontier-model developers. Higher-memory custom-chip products and rising memory costs are pressuring gross margins, while operating leverage continues to expand EBIT margins. Google's supplier diversification and intensifying custom-chip competition represent a risk to Broadcom's future TPU design share.

Analysis

AVGO’s AI upside is increasingly a systems-integration story rather than a pure accelerator-volume story: proprietary interconnect, switching and co-design create meaningful customer switching costs, but also concentrate revenue in a small number of hyperscaler capex budgets. The near-term market should reward continued AI revenue beats, yet gross-margin dilution means incremental revenue will not translate into EPS at the same rate investors associate with AVGO’s software-heavy historical mix. The key 1-3 month debate is whether EBIT leverage can outrun HBM and advanced-packaging cost inflation; a gross-margin miss alongside intact revenue guidance would likely compress the AI premium multiple.

GOOG supplier diversification is strategically rational but should not be read as a near-term displacement event. Qualification cycles for advanced ASICs, HBM integration and high-radix networking are long, making incumbent content resilient through the next several product ramps; the more material risk is AVGO receiving a smaller share of the next TPU generation’s value stack 6-18 months out. A second-order beneficiary of broader custom-silicon adoption is MRVL, where hyperscaler design-win optionality is more material to valuation, while memory suppliers MU and SK Hynix benefit from rising accelerator memory intensity even if AVGO’s unit economics weaken.

Consensus may be underestimating the asymmetry around margins: investors are likely to tolerate modest gross-margin erosion if AI revenue growth remains elevated, but a deceleration in AI growth would expose that the mix shift has reduced AVGO’s margin floor. Watch HBM pricing, substrate/advanced-packaging availability, and evidence that hyperscalers are moving from bespoke co-design toward multi-vendor sourcing. The thesis is falsified near term by sustained gross-margin expansion with stable AI growth despite higher memory costs, or by a material new hyperscaler design win that extends AVGO’s revenue visibility beyond the current product cycle.

Key Decisions for Investors

• Maintain a tactical long AVGO into the next earnings window only if AI revenue expectations have not moved materially above company-visible demand; target a 5-10% upside on another revenue/EBIT beat over 1-3 months, with risk control on any gross-margin-guide cut not offset by higher EBIT guidance.
• Express the custom-silicon buildout via a relative-value long AVGO / short SMH position over 3-6 months: AVGO has differentiated ASIC/networking exposure, while the hedge reduces broad AI semiconductor beta. Exit if AVGO’s AI growth decelerates materially versus the broader semiconductor complex or Google announces a significant alternate ASIC supplier.
• Add MU as a watch-list beneficiary rather than an immediate AVGO hedge; initiate only after confirmation that HBM contract pricing remains firm into the next two quarters. The catalyst is memory-content growth across custom accelerators, while the principal risk is a conventional DRAM/NAND inventory correction overwhelming HBM strength.
• Monitor GOOG’s TPU supplier disclosures, packaging-capacity commitments and custom-chip hiring for evidence of multi-vendor qualification. A confirmed second-source award is a 6-18 month negative for AVGO’s future content share, not necessarily a reason to short AVGO before the affected product-generation revenue is visible.