Credo vs. Broadcom: Which AI Connectivity Stock Is the Better Pick?

Credo reported fiscal Q1 revenue of $479 million, up 115% year over year, with non-GAAP EPS up 130.8% to $1.20 and a 48.2% operating margin; it forecasts more than 85% revenue growth in fiscal 2027. Broadcom's fiscal Q3 revenue rose 85.5% to $29.59 billion as AI semiconductor revenue jumped 221% to $16.7 billion, and it projects AI semiconductor revenue of about $58 billion in fiscal 2026, $115 billion in 2027 and $230 billion in 2028. The article favors Credo for higher-growth connectivity upside and a slightly lower 9.5x forward-sales multiple versus Broadcom's 9.93x, though Credo's 84% revenue concentration among four customers and Broadcom's expected gross-margin decline to roughly 73% remain key risks.

Analysis

The relevant distinction is not headline growth but durability of the revenue base. CRDO's valuation near AVGO's despite far greater customer concentration embeds a successful optical transition and continued hyperscaler share gains; any pause by a single large customer can create an outsized estimate reset. In contrast, AVGO's custom-silicon programs create switching costs across design, networking, software and supply-chain qualification, supporting a longer-duration backlog even if near-term margins dilute on accelerator mix.

For the next 1-3 months, AVGO is vulnerable to a mechanical gross-margin and operating-expense debate as lower-margin AI hardware displaces software revenue; this could create a better entry rather than impair the multi-year earnings power. CRDO needs evidence that optical revenues are scaling at gross margins comparable to its legacy connectivity franchise, not merely initial design wins. The key competitive risk is that higher-speed optical demand broadens the addressable market for Marvell (MRVL), Coherent (COHR) and Lumentum (LITE), limiting CRDO's ability to retain premium pricing.

Consensus appears too focused on CRDO's percentage growth and too dismissive of AVGO's mix shift. AVGO's incremental AI revenue may carry lower gross margin, but its absolute operating-profit and free-cash-flow conversion can still compound materially if custom accelerators displace merchant GPUs at scale. Conversely, CRDO's apparent sales-multiple discount is not a discount after adjusting for concentration, smaller liquidity, and the execution dependency of multiple product ramps; the appropriate comparison is to high-beta optical peers rather than AVGO's diversified platform multiple.

Over 6-18 months, the more consequential second-order effect is custom-accelerator proliferation: it increases demand for Ethernet switching, DSPs, retimers and optical modules, but also shifts buyer power to a handful of cloud customers. This favors scaled suppliers with broad product stacks such as AVGO, while CRDO requires continued technological differentiation to avoid becoming a qualified second-source component vendor.

Key Decisions for Investors

• Initiate a 6-12 month pair: long AVGO / short CRDO, sized dollar-neutral. The thesis is valuation convergence after adjusting for CRDO's concentrated revenue base; target 15-25% relative outperformance, with a stop if CRDO discloses broadening demand across additional hyperscalers and optical gross margins sustain near corporate levels.
• Use any post-earnings AVGO selloff tied solely to gross-margin compression to add long exposure over 1-3 months. Falsify if AI revenue guidance is reduced, custom-accelerator deployment schedules slip materially, or semiconductor operating margin falls faster than mix effects can explain.
• Avoid chasing CRDO until quarterly disclosure demonstrates that optical revenue is converting into durable, diversified production volume rather than sampling or early ramp activity. A watch trigger is customer concentration falling meaningfully from current levels while optical growth supports sustained upward FY estimates.
• For AI-connectivity beta without CRDO-specific concentration risk, monitor MRVL and COHR as alternative beneficiaries of 800G/1.6T deployment. Prefer only after confirmation of improved data-center order trends; these names carry greater pricing and inventory-cycle risk than AVGO.