Nvidia CEO Jensen Huang told attendees at the Computex trade show in Taipei on Tuesday that Marvell Technology, a data-infrastructure semiconductor firm based in Santa Clara, California, could become the next trillion-dollar chip company. Huang said the forecast is driven by demand for AI hardware that is “going through the roof” as autonomous AI agents — software that can run workflows and solve problems independently — begin to scale.
“Useful AI has arrived,” Huang said. He described a strategic shift in the architecture of massive AI data centers, where computing is becoming increasingly disaggregated and distributed, making advanced connectivity central to system performance.
“That’s the reason why Marvell is so essential,” Huang told Murphy during the joint keynote. “That’s why you’re going to be the next trillion-dollar company.”
Murphy, whose company specializes in data-infrastructure semiconductors and high-speed networking technology for data centers, said focusing solely on processor or memory “fails to capture the full picture of hardware efficiency.” He said the next major wave of AI innovation will be driven by interconnection.
Huang also addressed the industry’s shift from copper wiring to optical communications, or silicon photonics, for data transmission inside AI systems. He advocated a pragmatic approach: use optical links where technically necessary and copper everywhere else.
“You use optics wherever you must, and you use copper wherever you can,” he said.
Nvidia announced a strategic partnership with Marvell in March, saying it had invested $2 billion in the company. Marvell’s stock closed 7% higher Monday and surged more than 15% in after-hours trading on the Blue Ocean alternative trading system following Huang’s remarks.
The appearance at Computex is the latest signal of deepening ties between the two chip makers as Nvidia, which has a market capitalization above $3 trillion, pushes to build out the networking fabric needed for the next generation of AI data centers.