Broadcom Has Trailed the S&P 500 This Year. That Shouldn't Last Much Longer.

The key mechanism is not “AI demand is strong” but that custom silicon is turning into a cost-optimization tool for hyperscalers. That favors AVGO because ASIC programs tend to be sticky once qualified, which can extend revenue duration and improve visibility versus merchant GPU cycles. The second-order winner is GOOGL: if inference economics keep improving, it can keep spending aggressively while still lowering cost per query, which supports cloud share and ad/AI product rollout.

The market is likely underweighting how much of AVGO’s near-term multiple depends on sequential AI revenue momentum, not just year-over-year growth. If the company clears the implied AI run-rate into September, the stock can re-rate even without a major guidance raise because the current valuation still leaves room versus faster-growing semi peers. The main loser in the narrative is MRVL, not because its economics are broken, but because any incremental skepticism about “second-source” ASIC share can compress its multiple if investors decide AVGO remains the cleaner execution vehicle.

Contrarian risk: the consensus may be too confident that Nvidia’s beat automatically implies a clean read-through. GPU strength proves AI capex is alive, but it does not prove AVGO’s customer mix or program ramp cadence. If September commentary shows one or two hyperscalers pausing custom silicon orders, AVGO could de-rate quickly; the reversal would show up first in Q4 guide rather than the reported quarter. The thesis is falsified if AI revenue comes in meaningfully below the implied $16B path or if management tones down 2026 growth language.