Nvidia’s Jensen Huang calls Marvell “the next trillion-dollar company,” as NVLink Fusion deal deepens

Nvidia CEO Jensen Huang said at Computex 2026 in Taipei that Marvell Technology is “the next trillion-dollar company,” and investors quickly reacted. Marvell’s stock jumped by more than 25% in early trading on June 2, with Reuters reporting that the market capitalization stood far below the $1 trillion level Huang referenced.

The comment arrives as Nvidia and Marvell have been tightening collaboration around AI infrastructure. On March 31, Nvidia and Marvell announced a strategic partnership to connect Marvell to Nvidia’s “AI factory and AI-RAN ecosystem” through Nvidia NVLink Fusion. NVLink Fusion is described as a rack-scale platform intended to let customers build semi-custom AI infrastructure using Nvidia’s NVLink ecosystem, with Marvell contributing custom “XPUs” (accelerated processing units designed for specific workloads) and compatible scale-up networking.

Under the partnership, Nvidia said it would supply supporting technologies including Vera CPU, ConnectX network interface cards, BlueField data processing units, NVLink interconnect, and Spectrum-X switches, alongside rack-scale AI compute. Marvell said the companies would also collaborate on silicon photonics, a set of optical components used to move data over light for high-bandwidth AI connectivity.

The deal also extends beyond data-center connectivity. The companies said they would work to transform telecommunications networks into AI infrastructure using Nvidia’s Aerial AI-RAN for 5G/6G, and to advance networking for AI that includes optical interconnect solutions and silicon photonics technology.

A key financial announcement is Nvidia’s commitment to the relationship. Reuters reported that Nvidia invested $2 billion in Marvell earlier in the year, and the firms’ March 31 announcement similarly said Nvidia had invested $2 billion in Marvell as part of the expanded collaboration.

Marvell’s own results show the company is already seeing momentum from AI-linked infrastructure demand. In its fiscal 2026 earnings release, Marvell reported record revenue of $8.195 billion, up 42% year over year, and forecast continued acceleration in fiscal 2027. In its first-quarter fiscal 2027 results, Marvell reported net revenue of $2.4178 billion and showed that data center revenue made up 76% of total revenue.

Separately, Reuters reported that Marvell forecast its custom chips business could surpass $10 billion in revenue in fiscal 2029, as cloud companies expand AI data centers and invest in custom chips to reduce reliance on Nvidia’s processors. The “next trillion-dollar company” framing, however, was not accompanied by any disclosed valuation target or timeline for when Marvell might reach that level.

For AMD investors, Huang’s remark was pointed at Marvell rather than AMD, but the underlying theme matters for the whole compute and networking layer of AI systems: as customers scale AI factories, the market increasingly values companies that can provide custom silicon, fast interconnects, and integration pathways into dominant GPU ecosystems. What to watch next is whether more customers publicly adopt Marvell-based custom XPUs and NVLink Fusion-compatible networking, and whether Marvell’s custom silicon guidance continues to hold as Nvidia and Marvell expand the ecosystem.