- NVIDIA CEO Jensen Huang reportedly referred to Marvell as a potential trillion-dollar company.
- Marvell has evolved into a major AI infrastructure provider spanning custom silicon, optical networking, and data center technologies.
- The company is benefiting from explosive AI spending by hyperscalers including Amazon, Microsoft, and Google.
Technology executives praise partners all the time. Predicting they could become trillion-dollar companies is another matter entirely. That is why investors immediately took notice when NVIDIA CEO Jensen Huang reportedly pointed toward Marvell CEO Matt Murphy and declared, “The next trillion-dollar company, ladies and gentlemen.”
The comment alone would have generated headlines. Combined with reports that NVIDIA has invested roughly $2 billion into Marvell through its NVLink Fusion partnership, it has sparked fresh debate about whether one of the AI industry’s most important companies is still being overlooked by the broader market.
Marvell Is No Longer Just a Networking Company
Many investors still associate Marvell with networking chips and data communications hardware. While those businesses remain important, the company’s role within the AI ecosystem has expanded dramatically over the last few years.
Today, Marvell sits at the center of several critical AI infrastructure trends. The company develops custom AI accelerators, optical networking solutions, photonic interconnect technologies, Ethernet switching silicon, and storage systems that support modern data centers. As AI workloads continue scaling, demand for these technologies has grown rapidly.
In many ways, Marvell is supplying the infrastructure that allows the broader AI industry to function. While companies like OpenAI, Anthropic, and Google build advanced models, Marvell provides many of the tools needed to move data, connect systems, and support increasingly complex computing environments.
The Hyperscaler Opportunity Is Massive
One of Marvell’s most valuable businesses today is custom silicon. Rather than producing general-purpose chips, the company helps major technology firms design specialized processors tailored to their own AI workloads.
Marvell is involved in projects linked to Amazon’s Trainium platform, Microsoft’s Maia chips, and Google’s Axion initiative. Those relationships place the company directly alongside some of the largest AI infrastructure investments happening anywhere in the world.
Management expects custom silicon revenue to exceed $10 billion annually by fiscal 2029. That projection reflects the growing preference among hyperscalers to develop proprietary AI hardware rather than relying exclusively on off-the-shelf solutions.
AI Networking Demand Keeps Growing
Training advanced AI models requires enormous amounts of computing power, but raw processing is only part of the equation. Data must move efficiently between thousands of GPUs and servers, creating huge demand for networking infrastructure.
This is where Marvell’s optical networking business becomes especially important. Every new AI cluster requires increasingly sophisticated connectivity solutions capable of handling vast amounts of information with minimal latency.
As companies continue spending hundreds of billions of dollars expanding AI infrastructure, networking demand is expected to rise alongside it. Marvell remains one of the industry’s leading suppliers in this area, giving it exposure to another major growth engine beyond custom silicon.
Strategic Acquisitions Are Expanding Its Reach
The company has also strengthened its position through targeted acquisitions. Its purchase of Celestial AI added advanced photonic interconnect technology, an area many experts believe could become increasingly important as AI systems scale further.
Marvell also acquired XConn, expanding its capabilities in next-generation server architecture and connectivity. These moves help broaden the company’s presence across multiple layers of the AI infrastructure stack rather than concentrating growth in a single segment.
The strategy reflects a long-term effort to become indispensable regardless of which AI models, platforms, or applications ultimately dominate the market.
A Different Way to Invest in AI
Much of the attention surrounding artificial intelligence focuses on companies building chatbots, foundation models, and AI software products. Marvell offers a different investment thesis.
Rather than betting on which AI model wins, Marvell benefits from the infrastructure spending required to support the entire ecosystem. Whether enterprises adopt Claude, ChatGPT, Gemini, or future models that do not yet exist, they will still need chips, networking hardware, and data center technologies.
That makes Marvell one of the clearest “picks and shovels” plays within the AI boom.
Calling any company a future trillion-dollar business remains a bold prediction. Yet Marvell’s growing role across custom silicon, networking, photonics, and AI infrastructure makes the argument more compelling than many investors initially realize. If hyperscaler spending continues at its current pace, NVIDIA’s comment may eventually look less like speculation and more like an early glimpse of where the industry was headed.
