NVIDIA deepens AI infrastructure push with Marvell deal
NVIDIA has invested $2 billion in Marvell Technology, expanding their NVLink Fusion partnership and further integrating Marvell’s custom silicon capabilities into NVIDIA’s fast-growing AI infrastructure ecosystem.
The investment is aimed at strengthening NVIDIA’s interconnect strategy—critical plumbing for large-scale AI systems—by pulling one of the world’s leading custom chip designers closer to its platform. By reinforcing NVLink Fusion, NVIDIA is positioning its technology stack as a preferred foundation for hyperscalers and enterprise buyers building next-generation AI clusters, where bandwidth, latency, and system-level integration can dictate performance and total cost of ownership.
Raising the stakes ahead of UALink
The move also signals intensifying competition in chip-to-chip connectivity standards. Industry attention has been building around UALink, a rival interconnect effort associated with Broadcom. By investing directly in Marvell, NVIDIA appears to be accelerating ecosystem alignment around NVLink Fusion before alternative standards gain broader adoption.
Why Marvell matters
Marvell is a key player in custom silicon, networking, and data center connectivity—areas that increasingly define AI infrastructure design. A tighter partnership could help NVIDIA deliver more cohesive platform offerings that combine compute, networking, and interconnect technologies, while giving Marvell a stronger commercial pathway through NVIDIA’s customer base.
Neither company disclosed additional deal terms in the announcement. The investment underscores how strategic capital is being deployed not just to secure supply, but to shape standards and lock in platform influence as AI infrastructure spending continues to surge.
Related
