Nvidia said it struck a strategic partnership with Marvell to plug the chip maker into its NVLink Fusion ecosystem and invested $2 billion in the company, aiming to give customers more options to build custom accelerators and scale-up networks for AI. Under the arrangement, Marvell will deliver custom XPUs and NVLink Fusion–compatible networking while Nvidia supplies CPUs, DPUs, NICs, NVLink interconnects, Spectrum‑X switches and rack‑scale compute. The companies also plan to co-develop silicon photonics and optical interconnects and push Nvidia’s Aerial AI‑RAN to turn 5G/6G telecom networks into AI infrastructure. The move underscores Nvidia’s effort to lock in an end‑to‑end platform as demand for AI inference surges, even as both firms cautioned that execution, market conditions and regulatory factors could affect outcomes.
Related article:
