It’s Not Nvidia, Broadcom, or Micron.)

Business
Advertisements


While Wall Street obsesses over graphics processing units (GPUs), Marvell Technology (NASDAQ: MRVL) quietly powers the networking backbone that makes large-scale AI possible. Its under-the-radar role within AI data centers, combined with a $2 billion strategic boost from Nvidia and a privileged position in the next wave of hyperscaler capital expenditure (capex) budgets, position Marvel for outsize gains that could eclipse the company’s better-known peers.

Understanding Marvell’s overlooked role in AI infrastructure

Marvell designs high-speed Ethernet switches that move data at ultra-high speeds and low latency rates across clusters of server racks. Its product line also includes network interface cards and data processing units (DPUs) that off-load encryption and load-balancing tasks from central processing units (CPUs).

Will AI create the world’s first trillionaire? Our team just released a report on the one little-known company, called an “Indispensable Monopoly” providing the critical technology Nvidia and Intel both need. Continue »

One of the biggest reasons Marvell’s technology is overlooked is that it’s not directly used for training generative models. Instead, the company’s hardware ensures that every watt and byte inside an AI cluster is used efficiently. This is important because a single faulty switch or congested link can idle an entire rack of GPUs — ultimately costing developers both time and wasted capital.

Person looking at a stock chart on a computer.
Image source: Getty Images.

Nvidia’s investment is a discounted tailwind

A recent catalyst arrived for Marvell following Nvidia’s $2 billion strategic investment and partnership. The collaboration aims to leverage Marvell’s data center networking and custom silicon divisions to accelerate the next generation of Ethernet switches and DPUs specifically optimized for Nvidia’s AI platforms.

This relationship provides Marvell with immediate chip design wins inside the ecosystems that AI hyperscalers are already buying by the tens of billions. Over the next year, investors should be on the lookout for higher growth in the company’s networking ASICs and volume DPU shipments as the partnership with Nvidia ramps up.

In my eyes, the broader market has not yet fully priced the multiyear growth trajectory from Nvidia — or the secular themes accelerating investment in AI infrastructure — into Marvell’s valuation. This lag creates a compelling asymmetric buying opportunity: The upside could be enormous amid the AI infrastructure supercycle, yet the stock price has not fully caught up.


Leave a Reply

Your email address will not be published. Required fields are marked *