It only took one comment from Nvidia CEO Jensen Huang to send Marvell Technology (MRVL.US) soaring. On June 2, the optical interconnect leader saw its pre-market trading spike by more than 22%, hovering around $269.62 as we went to press.
During today’s Computex 2026 keynote, Marvell CEO Matt Murphy shared the stage with Huang to dive into why “connectivity” is shaping up to be the next massive wave for tech investors.
“Over the last few years, AI has thrown some serious curveballs at our infrastructure,” Murphy explained. “We’ve watched the industry tackle major bottlenecks one by one. First up was raw compute. To keep pace with AI’s explosive growth, we needed a massive jump in processing power—and Nvidia absolutely crushed it, even becoming the world’s first $5 trillion company.”
According to Murphy, the next big hurdle is memory. “Larger models demand insane amounts of RAM and bandwidth, and memory makers are already scaling up production to keep pace. We’ve just seen three new trillion-dollar tech giants emerge from that exact space. But the bottleneck is shifting again. Right now, it’s connectivity that’s pushing the boundaries of what our infrastructure can handle.”
Case in point: just last May, storage powerhouses Samsung, Micron, and SK Hynix all crossed the trillion-dollar market cap threshold in quick succession.
Murphy didn’t hold back, positioning Marvell as the “undisputed leader in connectivity.” He added, “When we started this journey, data center work made up less than 10% of our revenue. We went all in. Last quarter, that segment alone drove 75% of our total income, and it’s still climbing fast.”
Huang fully backed that sentiment, flat-out calling Marvell “the next trillion-dollar company.” He pointed out that both firms are tightening their partnership to build out the critical networking and interconnect infrastructure that next-gen AI data centers desperately need.
In his view, data center architecture is undergoing a massive shift as AI moves past the training phase into full-scale commercial deployment. We’re moving away from centralized compute setups toward a distributed, heterogeneous mesh that spans entire facilities. The glue holding all those GPUs, CPUs, storage arrays, and memory modules together? High-speed interconnect networks.
“Compute is now scattered across the whole data center floor, and interconnect tech is the only thing pulling it all together,” Huang noted. He emphasized that modern AI agents constantly juggle reasoning, data retrieval, memory management, and multi-model collaboration. That workflow demands unprecedented bandwidth and rock-solid networking—which perfectly explains why Marvell’s numbers have been skyrocketing lately.
For those unfamiliar with the space, Marvell is a heavy hitter in data center networking, switching silicon, optical interconnects, custom ASICs, and storage solutions. In plain English, they make the chips and hardware that let thousands of servers inside a data center talk to each other at lightning speed and work in sync. The company currently sits at a $192 billion market valuation.
Back on March 31, Nvidia officially announced a strategic tie-up with Marvell. The goal is to plug Marvell directly into Nvidia’s AI Factory and AI-RAN ecosystems using NVLink Fusion. They’re also joining forces on silicon photonics, and Nvidia has already dropped a $2 billion investment into Marvell to seal the deal.
Reflecting on the deal at the time, Murphy said, “Doubling down with Nvidia really underscores how critical high-speed interconnects, optical networking, and accelerated infrastructure have become for scaling AI. By linking our best-in-class analog tech, optical DSPs, silicon photonics, and custom silicon directly to Nvidia’s expanding AI ecosystem via NVLink Fusion, we’re giving our clients a blueprint for highly scalable, ultra-efficient AI infrastructure.”
Huang added that we’ve officially hit an AI inflection point, forcing data centers worldwide to upgrade their setups. While Nvidia’s chips remain the go-to standard, this partnership opens the door for hyperscalers who prefer designing their own custom silicon. Now, they can seamlessly plug those proprietary chips straight into Nvidia’s hardware and software ecosystem.
He also pointed out that several major cloud providers already run massive fleets of Nvidia hardware alongside their own in-house ASICs. Thanks to this collaboration, they can now wire their custom silicon directly into a unified system alongside Nvidia’s GPUs and LPUs.
Marvell’s latest Q1 earnings report paints a pretty impressive picture. The company pulled in a record-breaking $2.418 billion in net revenue, marking a solid 28% jump year-over-year. On the margins, GAAP gross profit hit 52.1%, while non-GAAP came in at 58.9%. Earnings per share landed at $0.04 (GAAP) and $0.80 (non-GAAP).On top of that, operating cash flow surged to a record $638.8 million for the quarter.
During the earnings call, management also raised the roof on their future outlook, now projecting that optical interconnect revenue will jump over 70% by the end of fiscal 2027—a massive bump from their earlier 50% estimate.
Nvidia’s own recent numbers back up this explosive growth trend. For Q1 of fiscal 2027 (ending April 26, 2026), revenue hit roughly $81.6 billion, up about 20% sequentially and 85% year-over-year. Net profit climbed to $58.3 billion (up 36% sequentially, 211% annually). The real engine behind those figures? Data center sales, which raked in around $75.2 billion for the quarter—a 21% sequential and 92% annual jump.
Looking ahead to Q2, Nvidia is guiding toward $91 billion in revenue, representing roughly a 2% sequential increase. Keep in mind, that figure doesn’t include data center compute sales from China. Margins are expected to stay rock-solid at 74.9% to 75.0% across both GAAP and non-GAAP metrics, with the full-year effective tax rate projected between 16.0% and 18.0%.
That forecast comfortably tops the $86.84 billion consensus from Refinitiv analysts, clearly signaling Nvidia’s confidence in sustained AI compute demand. But honestly, Wall Street is keeping its eyes on what’s really driving that growth momentum under the hood.
