How Nvidia Quietly Seized the Data Center Ethernet Crown

Few corporate climbs feel quiet until they’re obvious. Nvidia’s latest move into data center Ethernet switching reads that way: sudden in headline, gradual in execution.

According to IDC, Nvidia claimed the top revenue spot in the data center Ethernet switch market for Q1 2026 as the overall market swelled to $15.4 billion. The driver was an intense surge in demand for networking hardware tuned to AI workloads and sprawling hyperscaler builds.

Nvidia captured 21.5% of the data center Ethernet market after Spectrum‑X posted $2.1 billion in quarterly revenue, a staggering 192.7% year‑over‑year increase.

That growth hasn’t happened in isolation. Spectrum‑X bundles high‑performance switching with BlueField DPUs and NVIDIA LinkX cabling, a package designed for the large GPU clusters that underpin modern AI training and inference. The result: customers aren’t just buying switches; they’re buying an integrated network layer purpose‑built for AI factories.

Macro numbers back the narrative. Enterprise and hyperscaler datacenter spending alone hit $10 billion in the quarter, up about 61% from the same period in 2025. Regionally, the Americas led with near‑50% annual growth (49.7%), followed by EMEA at 32.2% and APAC at 25.9%—evidence that the AI-driven buildout spans continents, not just cloud hubs.

Port speeds tell a similar story. 800G switches accounted for roughly 35.8% of datacenter Ethernet revenue in Q1, while 200G and 400G gear together made up about 34.1%—meaning these high‑speed platforms represented roughly 70% of total Ethernet revenue. In short: bandwidth and low‑latency fabrics are now the currency of AI infrastructure.

Nvidia’s rise in switching complements its established dominance in GPUs and growing presence with Vera CPUs. Together, this tightens the company’s hold on the full AI stack—from accelerators through the interconnects that stitch clusters into usable systems. That integration is attractive to large cloud providers and enterprises building private AI campuses.

Whether incumbents can respond with comparable hardware‑software integration, or whether other vendors will instead specialize and coexist in niche roles, is the immediate strategic question. For now, Nvidia has shown that control over compute and the networking that binds it is a potent combination—one that may redraw vendor maps across the datacenter.