Nvidia Invests in Competitor: $2 Billion to Strengthen AI Market Control

Nvidia has made a strategic move by investing $2 billion in Marvell — one of its key competitors in niche AI hardware segments. This is not just an investment; it’s a way to integrate a rival’s technology into its own ecosystem and consolidate market leadership.

Collaborative Development and Technology Exchange

According to Nvidia’s press release, the deal involves close integration of Marvell into Nvidia’s “AI factory” through NVLink Fusion and the AI-RAN ecosystem. Both companies can address their weaknesses:

• Marvell will supply XPU processors and scalable networking solutions fully compatible with NVLink Fusion;
• Nvidia will provide access to Vera CPUs, ConnectX network cards, and Bluefield DPU units.

For large data center operators, this offers more flexibility: components from both companies can be combined without compatibility issues.

Networks of the Future: 5G, 6G, and Optical Data Transfer

The companies plan to jointly develop AI-RAN infrastructure for 5G networks and future 6G standards. AI will enable dynamic network resource management, which is crucial for autonomous vehicles, smart cities, and latency-sensitive applications.

Special focus is on silicon photonics — a technology that replaces copper connections with optical ones, speeding up data transfer between chips. This is a necessary step to support high-performance AI models and large-scale data processing.

Why Nvidia Supports a Competitor

At first glance, investing in Marvell, which along with Broadcom supplies specialized ASIC chips to giants like Google and Amazon, may seem counterintuitive.

However, Nvidia’s strategy is clear: if you can’t prevent clients from using third-party chips, make them compatible with your ecosystem. This allows the company to maintain control over industry standards, even if its own GPUs are not used in a specific project.