Nvidia CEO Jensen Huang pushed the company to start developing AI-specific chips in 2010, more than a decade before AI became a hot topic today, staying years ahead of the market. A similar move in 2020, doubling down on data center networking through strategic acquisitions, created one of the company’s most profitable and fastest-growing divisions, but with little fanfare.
In just a few years, Nvidia’s networking business, designed to connect data centers, has grown to become the company’s second-largest source of revenue after computing. Nvidia’s latest earnings show that it reported $11 billion in revenue last quarter, up 267% year-over-year, and brought in more than $31 billion in revenue for the year.
Driven by growth in AI processing, this division includes technologies such as NVLink, which powers communication between GPUs on data center racks. Nvidia InfiniBand Switch, an in-network computing platform. Spectrum-X, an Ethernet platform for AI networking. Among others, there are co-packaged optical switches.
Nvidia’s networking business includes all the technology needed to build “AI factories,” data centers designed for training AI models.
Kevin Cook, senior equity strategist at Zacks Investment Research, told TechCrunch that NVIDIA’s networking business is one of the company’s most impressive new areas. “[Nvidia’s networking business] It reported $11 billion in the quarter. “That number is bigger than Cisco’s networking business and about the same size as our full-year forecast,” Cook said, adding that it could accomplish in one quarter what Cisco’s business accomplishes in a year.
Nevertheless, this business segment has received less attention than the company’s much larger chip business. Additionally, it is not as widely advertised as the company’s game business, which is its core business, and its business size is nearly three times smaller.
Nvidia’s networking business has its roots in Mellanox, a networking company founded in Israel in 1999 and acquired by Nvidia in 2020 for $7 billion.
tech crunch event
San Francisco, California
|
October 13-15, 2026
Kevin Deierling is senior vice president of networking at Nvidia. He joined the company through its acquisition of Mellanox. Deierling told TechCrunch that the reason people don’t know about Nvidia’s networking business may be because he did a bad job marketing it, but he doesn’t like that answer.
“People think of networks as just, ‘I have a printer, I need to connect to it,'” Dearing says. “Mr. Jensen said this on the first day he bought us, that the data center is the new unit of computing. Networking is not just about moving small amounts of data between computing nodes; it’s actually the foundation.”
Deierling said he didn’t really understand why Huang bought the company when he bought it, but he understands now. By developing the network business in parallel with the GPU business, the company will be able to sell chips equipped with the optimal technology.
“When Mr. Jensen acquired Mellanox in 2020, he realized that it was the missing piece to make GPUs a complete package,” said Cook, the Zacks analyst.
Deierling added that he believes another aspect of Nvidia’s networking success is that the company only sells its technology as a full-stack solution rather than individual components, and that it doesn’t actually sell the technology itself, but through its partners.
“I can’t think of any other company that does this. [the] “We have full stack capabilities,” Dearing said. We build a complete compute stack, a fully integrated stack, and bring it to market through all of our partners. ”
Nvidia just announced a slew of all-new updates to its networking system during Huang’s keynote on March 16 at the company’s annual Nvidia GTC technology conference. The company announced the Nvidia Rubin platform, which includes six new chips that power an “AI supercomputer.” Nvidia also announced products such as the new Nvidia Inference Context Memory Storage platform and the more efficient Nvidia Spectrum-X Ethernet Photonics Switch.
Dearing said networking is “no longer just a peripheral for connecting printers and other slow I/O devices.” “This is the basis of computers. In the olden days, there was something called a backlining inside the computer. Now, the network is the backlining of the AI factory, and it’s very important.”
Source link
