Shares of Marvell Technology (NASDAQ: MRVL), a company focused on customized AI chips and optical interconnect technologies for large-scale AI data centers, and one of the largest partners behind Amazon AWS’s Trainium AI ASIC program, surged more than 30% during Tuesday’s US trading session.

The core catalyst was undoubtedly Nvidia CEO Jensen Huang’s appearance alongside Marvell CEO Matt Murphy at Computex 2026 in Taipei.

During the discussion, Huang stated that Marvell has the potential to become the next publicly traded company with a market capitalization exceeding US$1 trillion, while emphasizing its critical role in large-scale AI data center interconnect infrastructure.Such a direct endorsement from the most influential leader in the AI computing industry significantly strengthened market confidence in Marvell’s long-term growth trajectory.

Marvell’s CEO emphasized that as demand for AI large-language-model computing resources and large-scale AI infrastructure clusters continues growing exponentially, improvements in computing performance are increasingly outpacing advances in data movement and high-speed interconnect capabilities both within and between data centers.

As a result, the next major bottleneck for AI infrastructure expansion will be connectivity rather than computing power at individual nodes.

Marvell’s optical interconnect, ultra-high-speed DSP, and silicon photonics technology stack sits directly at the center of this emerging infrastructure bottleneck.

Jensen Huang’s broader outlook on the future of AI computing has also become an important foundation behind growing investor optimism toward Marvell.

He argued that AI commercialization has entered the era of practical, agent-driven AI applications.

This means token generation is no longer merely supporting research and experimentation — it is increasingly generating real economic value.

As AI workloads become more distributed and decomposed across large-scale systems, demand for massive interconnect capacity has moved beyond simple node-level upgrades.

System-level scalability has become the central challenge.

As distributed AI architectures gradually become standardized, the interconnect layer is expected to become one of the fastest-growing and highest-value segments within the AI ecosystem.

This forms the core foundation of Marvell’s long-term value proposition.

Jensen Huang also emphasized that Nvidia’s US$2 billion strategic investment in Marvell and the partnership between the two companies is not based on one-way dependency, but rather ecosystem complementarity.

Under the NVLink Fusion framework, Marvell can provide hyperscale customers with customized optical interconnect and ASIC solutions, while Nvidia contributes its GPU ecosystem and system-level platforms.

Together, the partnership aims to ensure that future data center infrastructure achieves the right balance between flexibility and high performance.

This combination of hardware, optical interconnect infrastructure, and ecosystem integration positions Marvell for a potentially long-duration growth cycle across future AI factories, AI-RAN deployments, and edge AI infrastructure.

These factors collectively support the thesis that Marvell could evolve into one of the core trillion-dollar companies within the AI infrastructure ecosystem.

Market Analysis:

Marvell’s technology portfolio and expertise in high-speed optical DSPs, optical interconnect modules, and custom ASIC/XPU development occupy some of the most scarce and defensible segments within the AI infrastructure stack.

As demand for both AI training and inference computing continues accelerating, these components may prove even harder to replace than standalone computing chips themselves.

This positions Marvell as more than simply another AI chip stock.

Instead, the company increasingly sits at the architectural core of optical interconnect infrastructure, data center DCI (Data Center Interconnect) networks, and the broader data movement ecosystem.

As a result, Marvell may possess a longer-term and potentially more stable pathway for revenue growth and profit expansion than many traditional semiconductor companies.