Nvidia CEO Jensen Huang tells audiences Marvell could be the next trillion-dollar company, pointing to the AI data-center “connectivity” bottleneck

Nvidia CEO Jensen Huang said Marvell Technology could become the next trillion-dollar company, arguing that Marvell is essential to the connectivity needed for modern AI data centers. His remarks, made during an onstage appearance with Marvell CEO Matthew Murphy at Computex Week in Taipei, came as semiconductor investors increasingly focus on the infrastructure that keeps AI workloads flowing across thousands of chips in large clusters.

Huang framed the challenge as one of distributing and aggregating compute in a system where workloads are broken into many parts. In that setup, he said, “connectivity” is what makes the overall system work, because the distributed components must share data quickly and reliably. He called Marvell “so essential,” linking its networking and connectivity chips to the practical requirements of moving bandwidth inside data centers.

The market reaction was immediate. CNBC reported that Marvell shares jumped roughly 32% on June 2 after Huang’s comments, describing it as Marvell’s biggest one-day gain ever, narrowly surpassing its previous best performance in May 2023. The surge highlighted how much weight traders assign to high-profile endorsements from Nvidia, the dominant supplier of AI accelerators and an influential driver of the broader AI chip supply chain.

Marvell specializes in designing high-performance chips used in global data infrastructure, including cloud computing, AI, enterprise networking, and 5G carrier networks, according to CNBC. The company’s positioning fits Huang’s argument that AI scaling is not only about accelerating math, but also about efficiently networking compute, memory, and bandwidth across large distributed environments.

The remarks also come amid evidence that Nvidia is investing in adjacent technologies that could reduce bottlenecks for AI communication. CNBC reported that Nvidia has invested $2 billion in Marvell and other companies designing photonic technology, which uses light to transmit data and is often discussed as a potential path toward higher bandwidth and lower energy costs than conventional electrical links. Photonics remains early and complex, but the investment indicates Nvidia’s interest in the communications layer, not just the processing layer.

Even with Huang’s confidence, the exchange did not include a detailed roadmap that would clarify exactly how Marvell’s products will translate into long-term revenue and margins. Public comments like these can reflect broad strategic alignment, but they do not substitute for disclosed contract terms, capacity commitments, or manufacturing timelines that would be required to precisely gauge financial impact.

For investors, the key unanswered question is what parts of Marvell’s portfolio most directly address the connectivity requirements Huang emphasized, and how quickly those solutions are being adopted by AI infrastructure customers. Marvell may benefit from increased attention, but without additional company guidance or contract disclosures tied to the “trillion-dollar” framing, the durability of any market enthusiasm will likely depend on subsequent order trends and customer design wins.

Jensen Huang’s endorsement is a reminder that the AI build-out is evolving into an end-to-end systems competition. What to watch next is whether Marvell provides further detail on demand drivers for its networking and connectivity products in AI data centers, and whether Nvidia’s communications-focused investments, including photonics, translate into measurable scale at the edge of the infrastructure stack.