Broadcom: The Market Is Wrong

Broadcom reported record Q2 results, with revenue up 48% and net income up 88%, while its semiconductor segment grew 143% year over year and now accounts for 49% of sales. AI orders topped $30 billion, about 3x shipped revenue, extending visibility through 2028, but the stock fell about 15% after earnings as Q3 AI revenue guidance of $16 billion came in below aggressive expectations. The report is fundamentally strong, but the guidance disappointment and post-earnings selloff make the tone mixed.

Analysis

The post-earnings selloff looks less like a fundamental break and more like the market punishing a high bar reset: AVGO is now being valued as an AI infrastructure quasi-utility, so even strong execution gets discounted if the forward step-up is not as explosive as the crowd modeled. The important second-order effect is that multi-year customer commitments reduce near-term revenue volatility but also compress optionality for surprises; that tends to cap multiple expansion because investors stop paying for scarcity and start underwriting a longer-duration annuity.

The real competitive implication is that AVGO’s deal structure reinforces a two-tier AI supply chain. Hyperscalers and frontier-model labs are effectively locking in custom silicon capacity, which disadvantages merchants of generic accelerator exposure and shifts bargaining power toward the few vendors with design wins and packaging scale. That should be incrementally negative for substitute compute pathways over the next 12-24 months, while benefiting upstream advanced packaging, HBM, and power-management vendors that can absorb the rising content-per-server bill.

The consensus is probably missing that the headline miss may be a timing issue rather than a demand issue. A $30B order book against a much smaller current revenue run-rate implies the debate should move from “is AI demand real?” to “how quickly can AVGO convert backlog into recognized revenue without margin leakage?” If conversion slips, the stock can stay range-bound for quarters; if it accelerates, the current drawdown may prove overdone because the market is implicitly treating 2028 visibility as if it were already fully discounted.

Near term, the main risk is positioning: AVGO had become a crowded long within AI infrastructure, so de-rating can continue for days to weeks if AI peers also guide conservatively. Over months, the key catalyst is whether additional hyperscaler/custom silicon wins are announced, because each incremental win raises the probability that current guidance remains understating the terminal revenue base rather than overstating it.

Key Decisions for Investors

• Buy AVGO on a staggered basis over 1-3 weeks, using the post-earnings gap as entry; target a 6-12 month hold if AI order conversion remains intact, with downside capped by a multi-year backlog and upside tied to multiple re-rating if backlog converts faster than expected.
• Pair trade: long AVGO / short a basket of generic AI accelerator exposure over 2-4 months; thesis is that custom-silicon backlog and customer lock-in should outperform commodity compute names if the market stays selective on AI capex quality.
• Express a tactical mean-reversion view via AVGO call spreads 1-2 quarters out, financed by selling upside against the recent implied-vol spike; favorable if the stock stabilizes once the guidance reset is digested, with defined risk after the post-earnings move.
• For GOOGL and META, treat the selloff as confirmation that top-tier hyperscalers are still willing to commit capex to proprietary AI infrastructure; stay long both on a 6-12 month horizon, as custom silicon should improve unit economics versus off-the-shelf alternatives.
• Set a watch item on AVGO margins and delivery cadence next two quarters; if revenue conversion lags backlog growth, reduce exposure, since the market will punish any sign that the 2028 visibility story is becoming too back-end loaded.