Arista Networks announced the 7060XE7 Series, a new set of 1.6T Ethernet networking platforms aimed at rack-scale AI infrastructure for training and inference. The company says the systems support both scale-up and scale-out AI fabrics, with configurations listed at up to 102.4 Tbps throughput.
Arista Networks announced the 7060XE7 Series, a new portfolio of 1.6T Ethernet networking platforms designed for rack-scale AI infrastructure.
In a June 9 press release, Arista said the 7060XE7 Series is intended to support both scale-up and scale-out AI fabrics used for next-generation training and inference workloads. Business Wire also carried Arista’s announcement, describing the release as focused on 1.6T networking for AI infrastructure and noting customer and partner comments from Meta, Microsoft, Oracle, AMD and Broadcom.
Arista’s product page describes the 7060XE7 Series as “high-performance 1600G solutions for AI architectures.” The company lists systems with 64x1600G or 128x800G configurations and up to 102.4 Tbps throughput, positioning the line for high-bandwidth connections inside AI data centers.
AI infrastructure increasingly depends on fast networking to connect accelerators, servers and storage across large training and inference deployments. Arista’s announcement frames the 7060XE7 Series as part of that shift, with the company emphasizing rack-scale infrastructure and Ethernet-based fabrics.
The distinction between scale-up and scale-out is central to Arista’s positioning. Scale-up networking typically focuses on connecting accelerators within tightly coupled systems, while scale-out fabrics connect larger groups of systems across a data center. Arista says the 7060XE7 portfolio targets both patterns, which are relevant as AI workloads demand higher bandwidth and lower latency between compute resources.
Arista’s 7060XE7 product page specifies 1600G and 800G port options, including 64x1600G and 128x800G configurations. The same product listing cites throughput of up to 102.4 Tbps.
According to Business Wire’s version of the announcement, Arista’s release included availability windows for the new platforms. The company’s announcement also included statements from major technology companies and silicon partners, including Meta, Microsoft, Oracle, AMD and Broadcom, indicating the product line is being introduced with support from several firms involved in AI infrastructure and networking ecosystems.
Arista did not describe the 7060XE7 Series as a complete AI system. Instead, the company presented it as a networking layer for AI fabrics, intended to move data between compute components in high-performance training and inference environments.
The announcement reflects continued investment in Ethernet networking for AI data centers. While some AI systems use specialized interconnects, Arista’s release focuses on Ethernet-based infrastructure and higher-speed links, including 1.6T connections.
For data center operators, the relevant details are the port configurations, throughput figures and intended deployment patterns. Arista’s published specifications point to systems built for dense AI networking requirements, while the company’s release frames the 7060XE7 Series as part of broader rack-scale AI infrastructure planning.
Arista’s announcement and product documentation provide the core technical positioning: 1.6T Ethernet platforms, 64x1600G or 128x800G configurations, and up to 102.4 Tbps throughput for AI architectures.
Arista Networks announced the 7060XE7 Series, a new portfolio of 1.6T Ethernet networking platforms designed for rack scale AI infrastructure.
New 1.6T systems for AI fabrics In a June 9 press release, Arista said the 7060XE7 Series is intended to support both scale up and scale out AI fabrics used for next generation training and inference workloads.
Business Wire also carried Arista’s announcement, describing the release as focused on 1.6T networking for AI infrastructure and noting customer and partner comments from Meta, Microsoft, Oracle, AMD and Broadcom.
Continue reading
