NVIDIA Invests $2 Billion in Marvell: Strategic Partnership for Next-Generation AI Infrastructures

NVIDIA has announced a $2 billion investment in Marvell Technology, accompanied by a strategic partnership aimed at expanding the hardware and infrastructure ecosystem dedicated to artificial intelligence.

The core of this collaboration is represented by NVLink Fusion, the program that opens its proprietary NVLink interconnection to selected third parties. In this scenario, Marvell will contribute to the development of customized XPUs and networking solutions compatible with NVLink, while NVIDIA will provide key components such as Vera CPUs, NVLink interconnections, ConnectX NICs, BlueField DPUs, and Spectrum-X switches, in addition to its AI computing infrastructure.

The goal is to enable a heterogeneous infrastructure capable of integrating accelerators, networking, and storage within a coherent ecosystem. This approach aims to offer greater flexibility to customers looking to develop customized solutions while maintaining compatibility with the NVIDIA technology stack.

Another pillar of the partnership involves the joint development of technologies in the field of silicon photonics, essential for improving the speed and energy efficiency of interconnections in data centers, where data transmission increasingly represents a bottleneck.

The two companies will also collaborate in the telecommunications front, with the aim of transforming 5G and future 6G networks into true AI infrastructures. In this context lies the AI-RAN initiative, which aims to integrate AI processing capabilities directly into radio networks, making them an active part of the computational ecosystem.

From a financial perspective, the announcement had an immediate impact on the markets: Marvell's stock saw a strong increase of over 10% in trading, while NVIDIA showed a more modest increase. This operation is part of a series of similar investments by the company led by Jensen Huang, which in recent months has allocated significant capital to technology companies linked to AI, including Synopsys, CoreWeave, Coherent, Lumentum, and Nebius.

NVIDIA is therefore progressively extending its reach beyond GPUs, aiming to control the entire AI infrastructure supply chain: from silicon to interconnections, through to data centers and telecommunications networks.