Key Highlights

  • Networking Revenue in the fiscal third quarter, which ended October 25, 2026, went past USD 21 billion, according to Nvidia's reported figures.
  • Co-packaged optics switches have reached Volume production, and one deployment links more than 10,000 GB300 GPUs.
  • Management anticipates total CPU revenue will more than double in fiscal 2028, because Demand for Vera CPU exceeds available supply.
  • Edge Computing sales climbed 42% to USD 8.2 billion, while physical AI is expected to surpass USD 25 billion this fiscal year.

Networking Emerges as a Major Revenue Line

Nvidia (NASDAQ:NVDA) is building substantial revenue in the networking hardware that connects its graphics processors. Sales in this area exceeded USD 21 billion during the fiscal third quarter, a level that alone would place it among the largest chip operations globally. Each large AI cluster requires fast connections within every rack and among thousands of racks, and the company supplies switches as well as the interconnect technology behind them. Nvidia stated that its co-packaged optics switches, which mount optical components right next to the switch silicon, have entered volume production. One of the earliest production installations links more than 10,000 GB300 GPUs through these switches. The company asserts that, compared with traditional pluggable optics, the design delivers roughly five times the power efficiency, about fifteen times the reliability and around 10% less latency. These gains matter because networking can absorb a large share of a facility's power budget, and one failed link can halt lengthy Training jobs. Bundling compute, networking and software lifts revenue per Data Center and makes the platform harder to swap for parts from other vendors.

CPUs Add a Second Engine of Growth

Outside graphics, the central processor represents a second growth driver. The Vera CPU ships inside Nvidia's Rubin rack systems and is also available as a separate product. Leadership reported that demand for Vera far exceeds what can be produced, and the names mentioned on the call included CoreWeave, Meta, Nscale and OpenAI, joining cloud clients disclosed in the prior quarter. Total CPU revenue should more than double in fiscal 2028. Agentic AI helps account for this demand: agents that invoke tools, query databases and execute code generate heavy general computing work around every model request, and that work runs on CPUs instead of accelerators. Should agent usage keep rising, the number of CPUs relative to GPUs in AI facilities could increase, opening fresh territory for Nvidia against long-established server chip suppliers. Nvidia is further widening its software footprint through a pending purchase of Hugging Face, a platform serving more than 18 million developers and 200,000 companies. The deal would bring Nvidia nearer to the developers who choose models and hardware, tightening the connection between its open models and its chips.

Edge, Graphics and Physical AI Segments

Nvidia now presents results by market platform. Edge Computing, which covers workstation, robotics and automotive offerings, brought in USD 8.2 billion during the quarter, a 42% rise from a year earlier and 14% above the previous quarter. Workstation sales built on Blackwell were a principal contributor. The Graphics segment, spanning gaming and professional visualization, reported USD 9.8 billion, up 62% year on year. Physical AI, which includes robotics, autonomous vehicles and factory automation, has almost tripled from a year earlier and is expected to pass USD 25 billion in fiscal 2027. Those businesses remain modest beside USD 107.6 billion of data center revenue, yet they lessen dependence on a small number of hyperscale purchasers. Investors will watch whether these lines keep growing if AI cluster spending slows. A more varied revenue base could soften a downturn and creates additional routes to growth if AI spreads into factories, vehicles and PCs. Management's confidence about doubling CPU revenue indicates it already sees that broadening underway.

Where Things Stand

Networking above USD 21 billion, an expected doubling of CPU revenue and a growing physical AI Business suggest Nvidia is widening its growth beyond accelerators. These lines are still smaller than data center GPUs, but they lessen reliance on a single product and a small set of buyers.