Nvidia, a dominant player in the artificial intelligence sector, is experiencing substantial growth not only from its leading graphics processing units but also from a lesser-known, yet increasingly vital, area: data center networking. This strategic business unit is rapidly expanding, indicating a significant shift in the company's growth trajectory and its potential for future valuation.
During its fiscal 2027 first quarter, ending April 26, Nvidia recorded an impressive $14.8 billion in data center networking revenue, marking a remarkable 199% surge year-over-year. This figure dramatically surpasses the $8.6 billion generated by this segment throughout the entirety of fiscal 2024, highlighting an accelerating trend. Projections for its fiscal years indicate a continuous upward trajectory, with networking revenue climbing from $8.6 billion in fiscal 2024 to $13 billion in fiscal 2025, and an estimated $31.4 billion in fiscal 2026. This exponential growth underscores the burgeoning opportunities within the networking domain as artificial intelligence clusters expand and necessitate more robust infrastructure to interlink thousands of accelerator chips.
The burgeoning importance of networking in contemporary AI systems cannot be overstated. High-performance networking is as crucial as raw computing power, particularly for training and executing advanced AI models. These operations demand the rapid transfer of enormous volumes of data among numerous accelerators and data storage solutions. Insufficient network capabilities can create bottlenecks, leading to underutilized GPUs and hindered AI workloads. Nvidia's NVLink technology facilitates connections among GPUs within powerful AI rack systems, while Spectrum-X Ethernet is responsible for linking servers and racks across expansive data centers. As AI systems scale, Nvidia stands to gain considerably by providing both networking and interconnect technologies alongside its core computing hardware.
Further solidifying its position, custom AI chips, while representing a competitive challenge to Nvidia's GPUs, are also reinforcing the networking opportunity. The NVLink Fusion rack-scale platform offers a strategic advantage by enabling certain custom processors to integrate seamlessly with Nvidia's networking and infrastructure technologies. This allows Nvidia to capitalize on the increasing investments in AI infrastructure, even when it is not the primary supplier of the AI accelerators. Collaborations, such as the expanded partnership with Marvell Technology, further exemplify this strategy. In this alliance, Marvell contributes custom accelerators and the necessary high-speed networking hardware, while Nvidia supplies critical technologies like NVLink, Spectrum-X switches, ConnectX network adapters, and BlueField data processing units. Although Nvidia faces competition, with IDC reporting Arista Networks' share of the data center Ethernet switching market only marginally below Nvidia's 21.5%, the company is strategically broadening its offerings to encompass more of the AI data center's technological requirements, thereby reducing its reliance solely on GPU sales.
Industry analysts anticipate Nvidia's earnings per share to increase significantly in fiscal 2028, with current estimates around $12.90. Should the company maintain its current forward valuation multiples, its market capitalization could potentially reach $7.5 trillion, a substantial increase from its approximate $5.3 trillion valuation today. While potential valuation compression could temper this upside, the networking business is no longer a secondary consideration for Nvidia. As AI factories continue to grow in scale and become more communication-intensive, Nvidia's expanded capacity to deliver comprehensive infrastructure solutions, beyond just its chips, is poised to become an increasingly pivotal element in its next phase of market expansion and valuation growth.
