Nvidia is expanding its dominance from GPUs into networking and CPUs to offer integrated AI system solutions.
Nvidia's networking business has reached a significant milestone, with annual networking revenue for its 2026 fiscal year reaching $31 billion—a 142 percent increase from the previous year and more than 10 times the level when the company acquired Mellanox Technologies in 2020. This achievement surpassed Cisco's networking revenue of $28 billion for its 2025 fiscal year, with Nvidia generating nearly $3 billion more than Cisco in their most recent quarters. CEO Jensen Huang highlighted this achievement on the company's February 25 fourth-quarter earnings call, stating "Nvidia is the world's largest networking business." The growth was driven by a continued ramp of NVLink compute fabric for its Grace Blackwell GB200 and GB300 rack-scale platforms as well as expansion of its Spectrum-X Ethernet and Quantum InfiniBand networking platforms.
Huang explained the scale of the networking opportunity, noting that "every rack comes with nine nodes of switches, and each one of them has two chips in it, and in the future, they'll have more." He emphasized Nvidia's rapid expansion in the Ethernet market, which it entered about two years ago: "We're probably the largest Ethernet networking company in the world today—and surely will be soon. And so Spectrum-X Ethernet has been a home run for us." This networking expansion reflects Nvidia's broader strategy of vertical integration, which the company calls "extreme co-design"—developing networking, computing, and chip technologies in tandem to deliver data center-scale computers optimized for AI workloads. CFO Collete Kress articulated this philosophy, stating that Nvidia's "ability to extreme co-design across compute and networking, across chips, systems, algorithms and software" enables the company to deliver continued performance gains.
Beyond networking, Nvidia is expanding into CPUs as a stand-alone offering, signaling increased competitive pressure on Intel and AMD. The company has announced deals with neocloud provider CoreWeave and hyperscaler Meta to supply its upcoming Vera CPU for their data centers. Huang explained that AI applications are increasingly learning to use tools that run in CPU-only compute environments, and that "Vera was designed to be an excellent CPU for post-training. And some of the use cases in the entire pipeline of artificial intelligence includes using a lot of CPUs." The company is also extending its accelerator capabilities through a reported $20 billion non-exclusive licensing deal with AI chip designer Groq, which includes hiring members of Groq's team, including co-founder and CEO Jonathan Ross. Huang stated that "as we did with Mellanox, we will extend Nvidia's architecture with Groq innovations to enable new levels of AI infrastructure, performance and value."
Despite intensifying competition—including AMD's recent announcement of supplying 6 gigawatts of AI infrastructure powered by its Instinct GPUs to Meta and the continuing success of homegrown AI chip efforts by Amazon Web Services and Google—Huang and CFO Kress conveyed confidence in Nvidia's continued dominance. This confidence is supported by substantial resources: an annual research and development budget approaching $20 billion and a significant investment war chest being deployed to invest in companies of all sizes. The company views its positioning in the broader AI infrastructure market, where it expects hyperscalers to spend nearly $700 billion this year, as enabling it to extend its dominance well beyond AI chips into product categories where it did not participate 10 years ago.