NVIDIA Becomes the Top Datacenter Ethernet Switch Vendor as AI Infrastructure Spending Hits Record Levels
NVIDIA is no longer dominating only GPUs and AI accelerators. The company has now taken the leading position in the datacenter Ethernet switching market by revenue for the first time, as global demand for AI infrastructure continues to surge.
In Q1 2026, the datacenter Ethernet switch market reached a record $15.4 billion in revenue, marking a 39.8% increase year over year. The rapid growth is being driven by hyperscalers, cloud providers, and enterprise datacenters racing to build the networking backbone required for large-scale AI systems.
NVIDIA’s rise in this segment is especially notable because the company has traditionally been viewed as the leader in GPUs rather than networking hardware. However, the expansion of AI factories and massive GPU clusters has changed what customers need. Modern AI datacenters require extremely fast, reliable, and tightly integrated networking, and NVIDIA is positioning itself as a complete infrastructure provider rather than just a chip supplier.
NVIDIA captured 21.5% of the datacenter Ethernet switch market in Q1 2026, generating $2.1 billion in quarterly revenue. That represents a massive 192.7% year-over-year increase, showing how quickly the company’s networking business is scaling.
A major reason behind this growth is NVIDIA’s Spectrum-X platform. Designed specifically for AI workloads, Spectrum-X combines high-performance Ethernet switching with NVIDIA BlueField DPUs and LinkX cables. The platform is built to support large GPU clusters, where fast communication between thousands of processors is essential for training and running advanced AI models.
This approach gives NVIDIA a powerful advantage. Instead of selling isolated components, the company can offer an end-to-end AI infrastructure stack that includes GPUs, CPUs, DPUs, switches, cables, and software. For hyperscalers and enterprises investing billions into AI datacenters, that level of integration is becoming increasingly attractive.
The broader market numbers also show just how quickly AI infrastructure is reshaping datacenter spending. Hyperscaler and enterprise datacenters accounted for around $10 billion of Ethernet switch revenue in Q1 2026, growing 61% compared to the same period in 2025.
Regional demand was strong across the globe. The Americas led the market with 49.7% year-over-year growth, followed by EMEA with 32.2% growth. The Asia-Pacific region also expanded significantly, posting 25.9% growth.
High-speed networking is now one of the most important parts of AI datacenter design. Demand for 400G and 800G switches remains strong as companies deploy larger and more powerful AI clusters. In Q1 2026, 800G switches represented 35.8% of datacenter Ethernet revenue, while 200G and 400G switches together accounted for 34.1%. Combined, these high-speed switching categories made up roughly 70% of global datacenter Ethernet revenue.
The rise of 800G networking is particularly important because AI workloads depend heavily on low-latency, high-bandwidth communication. As AI models become larger and training clusters expand, networking can become a bottleneck if it cannot keep pace with GPU performance. This is why platforms like Spectrum-X are gaining traction among companies building next-generation AI factories.
NVIDIA’s leadership in datacenter Ethernet switching highlights a major shift in the technology industry. The company is no longer just the dominant force in AI accelerators. It is expanding across the full AI infrastructure stack, from compute and networking to system-level integration.
As global datacenters continue to scale for AI, NVIDIA’s ability to provide optimized hardware across multiple layers gives it a stronger position in one of the fastest-growing markets in tech. With record Ethernet switch revenue, strong demand for 400G and 800G deployments, and rapid adoption of Spectrum-X, NVIDIA is proving that its AI ambitions extend far beyond GPUs.
