NVIDIA Invests USD 2 Bn in Marvell to Boost Next-Generation AI Infrastructure

Tech leader NVIDIA has invested USD 2 billion in Marvell Technology as part of a strategic partnership aimed at strengthening next-generation artificial intelligence (AI) infrastructure.

The collaboration focuses on combining Marvell’s custom chips and high-speed networking capabilities with NVIDIA’s AI computing platform, including its NVLink Fusion system.

NVLink is a high-speed connection technology that enables faster communication between processors such as CPUs and GPUs. NVLink Fusion expands this by allowing companies to integrate their own custom chips into NVIDIA’s ecosystem while maintaining seamless connectivity.

The partnership is expected to give companies building AI data centres greater flexibility. Marvell will contribute custom silicon, optical connectivity and networking technology, while NVIDIA will provide GPUs, CPUs, networking hardware and the software platform that integrates these components into a unified system.

Both companies are also set to collaborate on silicon photonics, a technology that uses light instead of electrical signals to transfer data. This approach is seen as a way to improve speed and energy efficiency in large-scale AI systems.

The move further strengthens NVIDIA’s position in the AI infrastructure market. Even when companies deploy custom chips developed by firms like Marvell, those systems will remain connected through NVIDIA’s platform. This ensures that NVIDIA continues to play a central role in AI computing environments.

For Marvell, the partnership provides access to NVIDIA’s extensive AI ecosystem and customer base. The company already develops custom AI chips for major cloud providers such as Amazon and Microsoft. Integration with NVIDIA’s platform is expected to make these chips easier to deploy and more attractive for large AI data centres.

Industry experts note that the key challenge in AI is shifting from building powerful chips to efficiently connecting thousands of them within data centres. Networking, optical connectivity and energy efficiency are becoming critical factors. The NVIDIA-Marvell partnership specifically targets these challenges, particularly bandwidth and power consumption.

The deal also reflects a broader shift towards modular AI infrastructure, where companies specialise in different components that work together within a shared system rather than building everything independently.

The investment in Marvell is part of NVIDIA’s wider strategy to build a comprehensive AI ecosystem. NVLink Fusion plays a central role by enabling third-party chips to integrate into NVIDIA’s platform while still relying on its technology.

In recent months, NVIDIA has invested about USD 2 billion each in companies such as Coherent, CoreWeave, Lumentum and Nebius, spanning areas like cloud computing and optical networking. The company has also announced partnerships across telecom, industrial software and cloud sectors, underlining its push to expand across the entire AI supply chain.