AI Back-End Network Switch Sales Exceed Front-End Networks for the First Time

AI Back-End Network Switch Sales Surpass Front-End Networks in 2Q 2026



In a significant turn of events in the networking industry, recent findings from the Dell'Oro Group highlight a pivotal moment for AI back-end networks. For the first time, sales of switches designed for AI back-end applications have outsold those intended for front-end networks as of the second quarter of 2026. This achievement underscores the rapidly evolving landscape of the network switch market and hints at the essential transformations underway in the sector.

According to Sameh Boujelbene, Vice President at Dell'Oro Group, the growth trajectory of AI back-end networks has been phenomenal. "In just over three years since their inception, AI back-end networks have not only caught up with but have overtaken the sales of established front-end networks, a feat that indicates a significant market shift."

The report emphasizes that the switch spending in the back-end network category is not just growing faster; the overall market size itself has expanded to surpass that of front-end networks. This shift poses intriguing implications for Ethernet switch vendors, as companies navigate through a transforming marketplace.

Moreover, supply chain factors are increasingly shaping the competitive landscape. Dell'Oro Group predicts that the market will remain constrained by supply issues for at least the next one to two years. This forecast places added pressure on vendors to secure the necessary components to meet the rising demand for AI back-end networking solutions.

Ethernet's Leading Role



Ethernet switches have maintained their dominance within the AI back-end category, specifically tailored for scale-out and scale-across use cases. Further developments are anticipated, as scale-up applications are expected to emerge later this year, potentially fueling even further growth in this sector. Industry analysts note that Ethernet's stronghold in this domain positions it favorably against competing technologies.

Celestica has retained its top position in terms of sales of Ethernet AI back-end switches, with NVIDIA following closely. Meanwhile, Arista has found its place in third, though there remains potential for a closer competition, contingent on the handling of deferred AI revenue. Cisco emerged forth in terms of market share expansion, indicating a potential resurgence in their strategy toward AI networking solutions.

The Technological Evolution



Switches operating at 800 Gbps have dominated the market, accounting for a significant majority of the shipments and revenue within AI back-end networks this quarter. With advancements paving the way, 1.6 Tbps switches are currently in the sampling phase, expecting a ramp-up in deployment during the latter half of 2026.

Ultimately, Dell'Oro Group's quarterly report on Data Center Switches for AI Back-end Networks provides a comprehensive overview of how these technologies are performing. It encompasses not only vendor revenue but also other metrics such as the number of ports shipped and average selling prices, which branch across a wide range of speeds, facilitating informed decision-making for stakeholders.

This report also delves into the geographical revenue breakdown and various customer segments, including major U.S. cloud providers and leading Chinese cloud companies.

Looking Ahead



With the underlying trends suggesting a robust market for AI-centric technologies, the networking landscape will likely continue to evolve significantly in the coming years. Businesses looking to stay competitive must keep a close eye on market dynamics and technological advancements to capitalize on the emerging opportunities in AI back-end networks.

As AI and networking technologies continue to converge, the implications for both technology providers and consumers alike are vast. Companies navigating this new terrain will need a strategic approach to maintain their positions as leaders in this ever-evolving market.

Topics Telecommunications)

【About Using Articles】

You can freely use the title and article content by linking to the page where the article is posted.
※ Images cannot be used.

【About Links】

Links are free to use.