Arista Expands AI Network Push From Switches to Rack-Scale Ethernet

Arista's lead announcement is Etherlink SU-144, an Ethernet-based scale-up architecture that connects up to 144 accelerators, or XPUs, within a single-hop domain and up to 1,024 XPUs across racks. It is built on the Ethernet for Scale-Up Networking, or ESUN, initiative.

According to SiliconANGLE, the move reflects a shift in AI infrastructure: the network can no longer be treated as an independent pipe moving data between servers, but has become a tightly integrated, rack-scale backplane. Arista, whose identity has been built on the network as a distinct layer, is arguing that compute is being built around the network rather than the network disappearing into compute.

The scale-up offering comes in three physical form factors. In an orthogonal chassis, servers sit in front and full-height switch cards in the back connect directly, with no backplane or cabling, drawing on Arista's modular chassis heritage. It supports 100 to 400 kilowatts of liquid cooling and is designed for vendors focused on training workloads that need the largest possible scale-up domain. A cabled backplane places switches in the middle of the rack and connects them to servers through copper cable cartridges; it is less dense but more serviceable and lets operators split bandwidth between intra-rack and inter-rack connections. A cross-rack design uses a dedicated switch rack to connect separate accelerator racks, making it the most flexible and fastest to market because it avoids long co-engineering cycles, but it adds interconnect cost and power. Vijay Vusirikala, Arista's distinguished lead for AI systems and networks, said copper runs out of juice very quickly, so anything beyond two racks needs optics.

For scale-out, Arista is offering liquid-cooled rack reference designs in eight-, 12- and 16-switch configurations built around its 7060EX7 switches. Each delivers 102.4 terabits per second, plus fiber patch panels, cooling manifolds, drip trays with active leak detection, rack management controllers and power shelves with battery backup. Integration partners build, test and deliver the configured racks. The fabric supports Ultra Ethernet Consortium specifications, Multipath Reliable Connection resiliency, multi-plane routing and dynamic and cluster load balancing.

Customers can run Arista EOS or an open network operating system on the common Netdi foundation. Arista also said next-generation optics, including its XPO pluggables and open co-packaged and near-packaged optics, will increase rack capacity from 1.6 petabits per second to 6.5 petabits per second and cut data center footprint by nearly half.

The OCP Global Summit runs Oct. 12-15 in San Jose, where Arista will have a fully assembled liquid-cooled rack at its booth. According to SiliconANGLE, the booth will depart from Arista's traditional engineering story, focusing less on switching and more on manifolds, leak handling, fiber management and power. According to Arista's OCP event page, co-founder Andy Bechtolsheim will present at 1 p.m. Oct. 13 on 'The Gigawatt AI Data Center Era.'

Among the takeaways, Arista is targeting the one part of the AI network it has not owned: scale-up, where Nvidia's NVLink has dominated. Arista said this will not change for Nvidia-based systems anytime soon. Vusirikala told SiliconANGLE there is no near-term opportunity because that is proprietary. The opportunity lies in non-Nvidia accelerators, and a partner list including AMD, Microsoft's Maia, Qualcomm and d-Matrix shows that market is real, with Arista betting that heterogeneous compute will become the norm. The company is also making the case for a single network stack across all three fabrics, which have historically been separate.