Better Buy After the Semiconductor Sell-Off: Marvell or Broadcom?

Broadcom reported Q2 revenue of $22.2 billion, up 48% year over year, with AI semiconductor revenue surging 143% to $10.8 billion and management guiding to $16.0 billion in next-quarter AI chip sales. Marvell also posted strong results, with revenue up 28% to $2.4 billion and fiscal 2027 growth now expected around 40%, but the stock still trades at about 90x earnings versus Broadcom's roughly 64x. The article argues Broadcom is the better buy after the sell-off because it is growing faster and offers a more reasonable valuation.

Analysis

The key takeaway is that the AI infrastructure trade is starting to bifurcate between narrative premium and cash-generating scale. Marvell’s endorsement-driven rerating looks fragile because its upside depends on a narrower set of hyperscale program wins, while Broadcom’s mix is broadening from “AI optionality” into a demonstrable earnings engine with far better self-funding capacity. That matters because in the next 1-2 quarters, the market will likely reward the name that can absorb disappointment without multiple compression, not the one with the most exciting headline beta.

Broadcom’s second-order advantage is that its cash flow gives it more room to keep winning sockets through packaging, networking, and software adjacency without needing perfect execution. If AI capex keeps broadening beyond the first-wave GPU clusters, Broadcom should capture more of the plumbing spend than Marvell, whose earnings sensitivity is higher to a handful of design cycles and customer concentration. The flip side is that any slowdown in hyperscaler order timing will hit MRVL first and harder, because the stock is priced like a durable monopoly rather than a cyclical supplier.

The contrarian miss is that the current debate may underweight duration risk for both names. Even if AI demand remains healthy, the next leg of returns will likely come from margin durability and diversification, not from revenue growth alone; on that score, AVGO has the cleaner setup. MRVL could still outperform on short bursts of sentiment, but it now needs flawless follow-through, while AVGO merely needs to keep converting AI demand into cash at scale.

Key Decisions for Investors

• Go long AVGO vs MRVL as a 1-3 month pair trade; the relative valuation gap and lower execution fragility favor Broadcom if AI capex stays firm. Use a stop if MRVL re-accelerates bookings and AVGO guidance inflects upward again.
• Initiate a tactical long AVGO position on pullbacks over the next 2-6 weeks; risk/reward improves if the market continues to de-rate narrative-driven winners and reward free-cash-flow compounders.
• Avoid chasing MRVL after the endorsement-driven spike; if long, hedge with short-dated calls or a partial short against semis to protect against a program-delay multiple reset.
• For options, favor AVGO call spreads 3-6 months out rather than outright calls; the thesis is continued rerating through earnings revisions, not a straight-line melt-up.
• Watch hyperscaler capex commentary and any sign of AI networking mix shift over the next earnings season; a broadening spend cycle is the catalyst for AVGO outperformance, while any customer delay is the main bear trigger for MRVL.