NVIDIA INVESTS $2 BILLION IN MARVELL, DEEPENS AI INFRASTRUCTURE PUSH

Nvidia is investing $2 billion directly into Marvell Technology, expanding a partnership aimed at strengthening artificial intelligence infrastructure as global AI spending accelerates.

- 

The agreement tightens integration between Nvidia's AI computing platforms and Marvell's custom silicon, networking, and photonics technologies — key components in scaling next-generation data centers.

The investment underscores how chipmakers are aligning more closely across the AI hardware stack, from GPUs to networking fabrics, as enterprises and cloud providers race to support rapidly growing token generation and model deployment demands.

Industry estimates place the broader AI infrastructure opportunity at roughly $630 billion over the coming years, intensifying competition among semiconductor and networking firms.

INTEGRATING NETWORKING, PHOTONICS, AND AI COMPUTE

Nvidia CEO Jensen Huang said the investment will enable customers to tap into Nvidia's expanding AI infrastructure ecosystem while building specialized compute environments tailored to large-scale AI workloads.

Modern AI systems rely not only on powerful GPUs but also on ultra-fast data movement between chips, racks, and data centers. As models grow larger and inference demand rises, interconnect performance has become a critical bottleneck.

Marvell plays a central role in that layer. The company develops custom silicon, high-speed networking chips, and optical interconnect solutions that help move massive amounts of data efficiently across servers.

Adding momentum to the partnership, Marvell recently completed a $3.25 billion acquisition of Celestial AI, bringing in photonic fabric technology designed to enhance chip-to-chip communication. Photonics — which uses light rather than traditional electrical signals to transmit data — promises lower latency and higher bandwidth, particularly valuable for GPU-heavy AI clusters.