5 Reasons to Buy Broadcom Stock Before Its Sept. 2 Earnings

Broadcom is set up as a key AI-chip supplier ahead of its Sept. 2 earnings, driven by custom TPU growth linked to Alphabet’s AI capex ramp to $195B–$205B (and higher next year). The article cites $21B of Anthropic TPU orders and a $200B Samsung Electronics agreement for HBM, manufacturing, and advanced packaging through 2030—positioning Broadcom with a supply-chain moat. It also highlights an optical networking/“co-packaged optics” upside and notes the stock trades at 18x forward P/E for fiscal 2027, suggesting valuation support if Q3 results and guidance are strong.

Analysis

Broadcom is increasingly a leverage point on hyperscaler capex, but the real edge is supply control, not just chip design. Securing scarce HBM, advanced packaging, and leading-node capacity turns each custom-ASIC win into a quasi-locked revenue stream; that is structurally better than competing on merchant accelerators where pricing can be reset every cycle. Second-order, that same scarcity should keep TSM and SSNLF as indirect winners while making AVGO harder to displace in future custom-silicon bids.

The overlooked upside is optical. If AI buildouts keep shifting from compute-limited to interconnect-limited, AVGO gets paid twice: once on the accelerator content and again on the networking layer that moves data between racks. That creates a longer duration earnings compounding story than the market usually assigns to a semiconductor name, especially if attach rates rise faster than unit shipments over the next 6-18 months.

The main risk is that the market is extrapolating design wins into linear revenue too quickly. Custom-chip ramps are lumpy, customer-concentrated, and often lag recognition by quarters; if one or two hyperscalers slow deployment or if NVDA’s next-gen roadmap narrows the performance gap, the rerating can stall fast. Consensus may also be underweight the possibility that the best trade is not AVGO itself, but the infrastructure bottlenecks around it.

Key Decisions for Investors

• Add to AVGO on any post-earnings fade, with a 3-6 month hold; thesis breaks if management does not lift medium-term revenue visibility or shows TPU/custom-ASIC revenue is less durable than implied.
• Initiate a 1-3 month relative-value long AVGO / short NVDA if earnings confirms optical attach-rate gains and customer concentration is not worsening; target 5-10% spread capture, cover if NVDA commentary re-accelerates demand.
• Overweight TSM and SSNLF as capacity toll-collectors on AI silicon buildouts; these are the cleaner second-order beneficiaries if AVGO’s custom-chip pipeline converts.
• Stay underweight MU versus the AI-capacity winners until there is evidence its HBM share can expand meaningfully; it has the right theme, but less direct leverage to the capacity bottleneck AVGO is monetizing.
• Use GOOGL capex commentary as the leading indicator: if Alphabet pauses upward revisions in the next 1-2 quarters, reduce AVGO exposure because the custom-chip thesis is most dependent on continued hyperscaler investment.