The datacenter Ethernet switching market reached $15.4 billion in the first quarter of 2026, a 39.8% surge that coincided with NVIDIA claiming the number-one vendor position, according to IDC's latest tracker data.

The figure marks one of the steepest expansions on record for a segment that had already posted sustained double-digit growth through the AI buildout cycle. A near-40% jump in a single quarter points to hyperscale and enterprise buyers accelerating network refreshes alongside GPU deployments, rather than spreading capital expenditure across the year.

What does the ranking change mean?

NVIDIA's ascent to the top of the datacenter Ethernet switching market displaces the incumbent vendor hierarchy that Cisco, Arista Networks and Huawei have anchored for roughly a decade. IDC's tracker places NVIDIA first by revenue for the quarter.

The shift is structural, not marginal. NVIDIA's networking business — built on Spectrum-X Ethernet platforms and the InfiniBand portfolio inherited with Mellanox in 2020 — now sells into the same AI clusters that drive its accelerator revenue. Customers buying racks of GPUs increasingly take the switching fabric in the same transaction, collapsing the traditional separation between compute and network procurement.

For competitors, the consequence is direct: Ethernet switching revenue growth is pooling around AI-class fabrics with high-radix, high-speed ports, while classic enterprise access and aggregation layers grow far more slowly. Vendors without an AI-networking story capture a shrinking share of a rapidly expanding market.

How large is the market now?

IDC puts the first-quarter 2026 datacenter Ethernet switching market at $15.4 billion, up 39.8% from the same period a year earlier. That growth rate implies the market added well over $4 billion in revenue year over year in a single quarter.

Sustained at that pace, the segment would clear $60 billion in annual revenue in 2026 — a scale that rivals the entire switching and routing market of just a few years ago.

What is driving the surge?

Three demand vectors explain the acceleration:

• AI cluster construction. Front-end and back-end networks for large language model training and inference require Ethernet fabrics at port speeds and densities far above legacy datacenter norms.
• Spectrum-X attach. NVIDIA has positioned its Ethernet platform as the default networking option alongside its accelerators, converting GPU orders into switching revenue.
• Speed upgrades. 800G port adoption is moving from early hyperscale deployments toward broader availability, lifting average selling prices across the segment.

What should buyers and investors watch next?

The key question is durability. A 39.8% quarterly growth rate assumes AI infrastructure spending keeps compounding, and capital-expenditure guidance from the largest hyperscalers will test that assumption over the rest of 2026.

Competitive responses are already in motion. Cisco, Arista and Juniper-owner HPE have all staked positions in AI back-end Ethernet, and white-box vendors led by Celestica and Accton continue to expand share at hyperscalers. Whether NVIDIA converts a single quarter's lead into a durable number-one position depends on whether its attach-rate advantage survives as rivals ship comparable high-speed platforms.

The next two quarterly tracker releases from IDC will show whether the first quarter of 2026 was an inflection point or a peak. For now, the data is unambiguous: datacenter Ethernet switching is growing at AI-accelerator velocity, and NVIDIA holds the top of the market it helped reorder.