Nvidia's networking business nears $60 billion

Nvidia’s networking business is drawing new attention after the company disclosed record data-center networking revenue in its latest quarterly results, giving investors a clearer view of how much of the AI buildout extends beyond GPUs. Nvidia said on May 20 that first-quarter fiscal 2027 data-center networking revenue rose to $14.8 billion, up 199% from a year earlier and 35% from the prior quarter. At that pace, the unit is running at about $59.2 billion on an annualized basis. The figure underpins a May 23 analysis by 24/7 Wall St. that argued Nvidia’s networking operation is nearing a $60 billion business. ### Where does the near-$60 billion figure come from? Nvidia’s own earnings release provides the math. The company said data-center networking revenue was $14.8 billion in the quarter ended April 26, 2026. Multiplying that quarterly figure by four yields roughly $59.2 billion, which is the basis for describing the business as nearing a $60 billion run rate. (investor.nvidia.com) The May 23 24/7 Wall St. analysis framed that number as a sign that Nvidia’s networking arm has become a major business in its own right, rather than a supporting line item to GPU sales. Yahoo Finance’s pickup of the report also said the quarterly figure put Nvidia’s networking business at a $60 billion annualized run rate. ### What exactly is Nvidia counting as networking revenue? (investor.nvidia.com) Nvidia said it is changing its reporting structure, but it still disclosed the old sub-market breakdown for comparison. Under that prior framework, “data center networking” sat alongside “data center compute,” and the company reported $14.8 billion for networking in the latest quarter. (247wallst.com) Nvidia’s networking portfolio includes InfiniBand and Ethernet products used to connect large clusters of AI chips across data centers. On its Spectrum-X product page, Nvidia says the Ethernet platform is designed to accelerate AI networking and improve performance for large-scale AI workloads. ### Why has networking become such a large piece of the AI buildout? Jensen Huang said on May 20 that the “buildout of AI factories” is accelerating, and Nvidia tied its reporting change to growth in hyperscale and purpose-built AI data centers. (investor.nvidia.com) Those facilities require not just processors but also high-speed interconnects to move data among thousands of chips with low latency. (nvidia.com) Broadcom has made a similar point from the Ethernet side of the market. Hock Tan said in December that Broadcom expected fiscal first-quarter 2026 AI semiconductor revenue of $8.2 billion, driven by custom AI accelerators and Ethernet AI switches. That guidance shows how central networking hardware has become to AI infrastructure spending across suppliers. (investor.nvidia.com) ### How does Nvidia stack up against Broadcom? Broadcom does not report a directly comparable standalone networking segment in the same format, which makes one-to-one comparisons difficult. Its public guidance grouped AI semiconductor revenue at $8.2 billion for fiscal first quarter 2026, with custom accelerators and Ethernet AI switches both contributing. (investors.broadcom.com) S&P Global Market Intelligence estimated in February that Broadcom’s AI networking revenue, including high-speed Ethernet switching products, would be about $2.3 billion in that quarter. That estimate is not a company figure, but it illustrates why analysts and market commentators are increasingly treating Nvidia as a leading force in AI networking as well as AI compute. (investors.broadcom.com) ### What should investors watch next? Nvidia’s next formal update will come with its fiscal 2027 second-quarter results after the company guided for second-quarter revenue of about $91 billion. Any continuation in the disclosed networking line, or further detail under its new reporting framework, will show whether the near-$60 billion run rate is sustained. Broadcom’s next earnings updates will provide the clearest public benchmark on the Ethernet side. (spglobal.com) In its most recent annual results, Broadcom said first-quarter fiscal 2026 AI semiconductor revenue was expected to reach $8.2 billion, driven by custom AI accelerators and Ethernet AI switches. (investors.broadcom.com) (investor.nvidia.com)