Marvell's competitive landscape differs by product line. In optical interconnect DSPs (PAM), the primary competitors are Coherent Corp. and Intel's Altera division, neither of which has matched Marvell's volume ramp pace at 800G and 1.6T. The more meaningful structural competitor is Broadcom (AVGO), which competes directly in both custom XPU design and optical DSPs, with deeply entrenched positions at Google (TPU) and Meta. Broadcom reported FY2025 revenue of $63.9 billion and trades at approximately 27.5x EV/Sales versus Marvell at approximately 17.8x TTM.^10,11

In custom silicon, Marvell's differentiation is specialization in networking-attached compute — XPUs optimized for inference workloads — rather than Broadcom's broader ASIC platform. The reported Google discussions would represent expansion of Marvell's hyperscaler relationships without necessarily displacing Broadcom's primary TPU position. In data center switching, Marvell competes with Broadcom's Tomahawk/Trident line, Cisco's Silicon One, and Intel's Tofino, all with greater installed bases. In scale-up interconnects addressed by Celestial AI, competitors include Ayar Labs and other CPO start-ups, but none with a confirmed large-scale hyperscaler design win at product launch as of Q1 FY2027.⁵

"Marvell is the first company to productize 200-gigabit per lane technology, enabling the 1.6T transition now underway. While this generation is expected to continue to grow through the end of the decade, Marvell has already demonstrated 400-gig per lane technology. We expect that this will position us to enable the industry's subsequent transition to 3.2T."
— Matthew Murphy, Chairman & CEO, Q4 FY2026 Earnings Call, March 5, 2026⁵

Company	Latest FY Revenue	GAAP Gross Margin	EV/Sales (TTM)	Primary AI Focus
Marvell (MRVL)	$8.2B (FY2026)	51.0%	~17.8x	Optical interconnect, custom XPU, switching, CPO
Broadcom (AVGO)	$63.9B (FY2025)	67.8%	~27.5x	Custom ASIC (XPU), AI networking, software
Microchip (MCHP)	$4.4B (FY2025)	56.1%	~7.0x	Embedded / industrial, limited AI exposure

Marvell Technology: Engineering the Backbone of the AI Data Center

The Power Delivery Gap