Broadcom Is Down 20% From Its All-Time High. Here's Why It's a Screaming Bargain.

Broadcom’s upside case hinges on custom AI chips ramping in 2027, with CEO Hock Tan targeting more than $100B in AI-semiconductor revenue in 2027 versus $10.8B in Q2. The article argues the stock is not fully pricing that growth, valuing AVGO at 20x 2027 earnings (or 30x for a more normalized scenario) and projecting ~50% potential gains over 12–18 months if Wall Street estimates are met.

Analysis

AVGO is increasingly a 2027 cash-flow story, not a 2026 one. That matters because the stock is now sensitive to timing slippage: if customer conversion into volume orders moves even 1-2 quarters right, the market will likely compress the out-year multiple before it will re-accelerate earnings estimates. In the near term, the stock can stay range-bound even if the fundamental story is intact because the next leg up depends on order visibility, not just design-win rhetoric.

The bigger second-order winner is the hyperscaler customer set, especially GOOGL and META, if custom silicon meaningfully lowers inference cost and improves unit economics versus general-purpose accelerators. That would shift bargaining power toward the cloud operators and away from the merchant GPU ecosystem, which is a subtle margin headwind for NVDA even if its training franchise remains dominant. Put differently: AVGO’s upside is partly a sign that the AI supply chain is moving from "scarcity pricing" toward "cost optimization," which tends to favor the largest platforms with internal engineering scale.

The contrarian risk is that consensus may be extrapolating a clean ramp from a very messy execution cycle. If the second wave of customers does not show up on schedule, the market could re-rate AVGO as a premium mature semi rather than a growth compounder, and that would cap upside well before the bull case fully plays out. The key falsifier is any 2027 guide that stops getting raised, or evidence that the customer ramp remains concentrated in one name; in that case, the multiple likely does the down work faster than earnings do the up work.

Key Decisions for Investors

• Long AVGO on weakness for a 6-12 month horizon; thesis is that the stock is being valued too conservatively on 2027 earnings, but size modestly because the ramp timing is the real risk. Falsify if out-year AI guidance stops improving.
• Pair trade: long AVGO / short NVDA on a beta-adjusted basis for 3-9 months. This expresses relative share-shift toward custom silicon without making a blanket bearish call on AI capex. Risk: broad AI spend re-acceleration favors both names and can swamp the spread.
• Watch GOOGL and META as indirect beneficiaries rather than immediate longs; if custom silicon lowers inference cost, they should see margin support before the market fully prices it. Best entry is on any post-earnings weakness where spend guidance rises slower than operating leverage.
• Use a catalyst alert, not a full position, for NVDA: if upcoming hyperscaler capex updates show no custom-chip substitution, the AVGO/NVDA relative-value thesis weakens quickly. Reassess if NVDA gross margin or share gains re-accelerate.
• If buying AVGO outright, prefer staged entries over chasing strength; the cleanest risk/reward is during drawdowns before 1H27 order visibility improves. The trade should be sized for a rerating, not for a straight-line rally.