Broadcom and Marvell Technology are at the forefront of the data-center boom powered by artificial intelligence, with both companies reporting record revenues and significant growth driven by demand for custom silicon. In Q2 fiscal 2026, Broadcom posted $22.2 billion in total revenue, including $10.8 billion from AI semiconductor sales, supported by commitments from major clients like Google and OpenAI. Meanwhile, Marvell reported $2.418 billion in Q1 fiscal 2027 revenue, attributing its growth to exceptional AI bookings. As hyperscalers increasingly opt for specialized custom accelerators instead of general-purpose GPUs, both companies are well-positioned to capitalize on this shift, making them appealing for traders looking to leverage the ongoing AI buildout.

Broadcom: Broadcom is a global semiconductor and infrastructure-software company whose portfolio includes networking, broadband, wireless, storage, and enterprise tools from its VMware acquisition. In the current AI buildout, Broadcom designs custom XPUs and high-speed networking silicon sold directly to hyperscalers and AI frontier-model companies. It serves as the scale player in the custom-silicon trade, with its AI segment now the primary growth driver.
Hock Tan: Hock Tan is the CEO of Broadcom and has led the company through its expansion into AI semiconductors and infrastructure software. He is directly relevant to the news as the executive commenting on Broadcom’s custom AI accelerator and networking demand from major hyperscalers during recent earnings discussions.
Matt Murphy: Matt Murphy is the CEO of Marvell Technology and oversees its shift toward data-center and AI-focused silicon. He is directly relevant to the news through his statements on exceptional AI bookings and raised revenue outlook tied to custom XPU programs.
Marvell Technology: Marvell Technology designs data-infrastructure silicon including custom compute accelerators, electro-optics, Ethernet switching, and interconnect products for modern data centers. In the AI context, it focuses on a smaller set of hyperscaler XPU programs and data-center networking, positioning it as the higher-beta challenger to larger custom-silicon providers. Its growth is concentrated in the data-center segment, which has become the dominant revenue contributor.

Custom Silicon Trend: Hyperscalers are turning to specialized custom accelerators rather than relying exclusively on general-purpose GPUs for their AI workloads.
AI Networking Importance: High-speed networking and interconnect silicon are essential for linking thousands of accelerators within hyperscale AI clusters.
Equity Derivatives Access: Major semiconductor names tied to the AI buildout are now available for leveraged trading via perpetual contracts on specialized platforms.