Nvidia's $60 billion networking business

Nvidia’s networking business is no longer a side note to its AI chip franchise. The company said on May 20 that data-center networking revenue reached a record $14.8 billion in its first quarter of fiscal 2027, up 199% from a year earlier and 35% from the prior quarter. That disclosure gives a hard operating number behind a May 23 analysis by 24/7 Wall St. that valued the business at about $60 billion. The figure matters because Nvidia’s AI buildout depends on more than GPUs. Nvidia sells the interconnect gear — including InfiniBand and Ethernet products inherited and expanded after its Mellanox acquisition — that links large clusters of chips inside AI data centers. Jensen Huang said on the earnings release that the “buildout of AI factories” is accelerating, while the company separately broke out networking as a distinct line within its previous data-center reporting. (investor.nvidia.com) ### Where does the $60 billion figure come from? 24/7 Wall St. published the estimate on May 23 and argued Nvidia’s networking and interconnect products could support a business worth about $60 billion. The outlet compared the segment’s potential scale with Broadcom’s AI networking and broader infrastructure portfolio. Nvidia itself did not assign a standalone valuation to networking in its May 20 earnings release. (investor.nvidia.com) What it did provide was revenue: $14.8 billion for one quarter. On an annualized basis, that run rate would imply roughly $59.2 billion, which appears to be the arithmetic behind the outside estimate. That is an inference from Nvidia’s reported quarterly figure, not a company forecast. (247wallst.com) ### What exactly is Nvidia selling in networking? Nvidia’s data-center networking business centers on the systems that move data between AI chips and servers at very high speed. Those products include InfiniBand and Ethernet-based switches, adapters and interconnect components that let large AI clusters operate as a single system. Nvidia has increasingly pitched its platform as a full-stack data-center offering rather than a chip-only sale. (investor.nvidia.com) The company’s May 20 release showed how large that piece has become relative to the rest of its data-center segment. Nvidia reported $60.4 billion in data-center compute revenue and $14.8 billion in data-center networking revenue in the quarter ended April 26, 2026. ### Why is Broadcom part of the comparison? Broadcom is one of the clearest public benchmarks because it already reports AI revenue that includes networking. (investor.nvidia.com) Broadcom said on March 4 that first-quarter AI revenue was $8.4 billion, up 106% from a year earlier, driven by demand for custom AI accelerators and AI networking. Chief Executive Hock Tan said on Dec. 11 that the company expected first-quarter fiscal 2026 AI semiconductor revenue of $8.2 billion, driven by custom AI accelerators and Ethernet AI switches. Those disclosures do not provide a pure networking-only number for Broadcom, but they show why analysts and commentators use Broadcom as a reference point. Both companies are selling key plumbing for AI data centers, though Broadcom’s reported AI revenue combines multiple product categories. ### Why does this matter beyond Nvidia’s GPU sales? Nvidia said on May 20 that it is shifting to a new reporting framework built around Data Center and Edge Computing, with Data Center split into Hyperscale and ACIE, for AI clouds, industrial and enterprise customers. (broadcom.com) That reporting change reflects how the company is framing growth across entire AI systems rather than only processors. The next public checkpoint is June 5, when Broadcom is scheduled to report fiscal second-quarter results, including updated AI revenue guidance. Nvidia’s next formal update will come with its fiscal 2027 second-quarter earnings materials on its investor relations site. (broadcom.com) (investor.nvidia.com)