Overview
- Nvidia disclosed Tuesday a $2 billion equity stake in Marvell alongside a pact to link Marvell’s custom processors and networking to Nvidia’s NVLink Fusion rack‑scale AI platform.
- The collaboration includes optical interconnects and silicon photonics, which use light to move data faster and with less power across crowded AI servers.
- Under the plan, Marvell provides custom XPUs and NVLink Fusion‑compatible scale‑up networking, while Nvidia contributes its Vera CPU, ConnectX network cards, BlueField data‑processing units, NVLink interconnects, Spectrum‑X switches, and rack‑scale compute.
- Marvell shares jumped about 7% to 11% intraday after the announcement, with coverage framing the deal as validation of Marvell’s role in AI networking and custom silicon.
- The move extends a string of Nvidia’s $2 billion stakes in AI supply‑chain firms as Big Tech forecasts hundreds of billions in AI infrastructure spending this year, reinforcing Nvidia’s position even as customers design their own chips.