NVIDIA Bets $2 Billion: New Alliance for AI Custom Chips Emerges

On March 31, Nvidia and Marvell Technology, Inc. (NASDAQ: MRVL) announced a strategic partnership to link Marvell’s custom XPUs and scale-up networking solutions to Nvidia’s NVLink Fusion-based AI factories and AI-RAN ecosystem, enabling the construction of heterogeneous AI infrastructure compatible with Nvidia systems.

As part of this collaboration, Nvidia will invest $2 billion—approximately KRW 2.6–2.7 trillion—in Marvell equity to bolster AI data center infrastructure and advance optical communications and silicon photonics technologies.

Separately, Marvell’s General Counsel Mark Kasper sold around 17,000 shares on April 1 and 2 in the open market, cashing in roughly $1.9 million (about KRW 25 billion), while CFO Willem Meintjes sold approximately $4 million worth of stock on April 15 under a 10b5-1 plan. With substantial RSUs and performance-based shares vesting, he also returned shares to cover tax obligations and received new RSUs, maintaining his holdings’ value at about $31 million (around KRW 40 billion).

Investors have noted that in its recent earnings report, Marvell disclosed fiscal 2026 revenue of about $8.2 billion, with more than 70 percent coming from data center sales—driven by growing demand for AI data center connectivity and custom chips.

Marvell also introduced 2-nanometer-class optical transport DSPs and ultra-high-speed optical modules, outlining a strategy to expand solutions that combine high-speed interconnects with custom silicon within the Nvidia NVLink Fusion ecosystem.

Headquartered in California, Marvell is a fabless semiconductor company that designs and supplies infrastructure chips and network semiconductors for data centers, cloud services, communications networks, and storage. Notably, it has broadened its presence in high-bandwidth interconnects—such as custom AI accelerators for major hyperscalers, Ethernet switches, and optical transceivers—and, bolstered by partnerships with Nvidia and major customers like Amazon Web Services and Microsoft, is emerging as a key player in the AI infrastructure supply chain.