Edgecore Networks introduced two 102.4 Tbps open networking switches aimed at hyperscale AI and cloud data center fabrics, marking the industry’s first commercially announced platforms at this capacity. The new AIS1600-64O and AIS800-128O systems run on Broadcom’s 3nm Tomahawk 6 silicon and target high-bandwidth, low-latency AI/ML clusters that are rapidly transitioning to 800G and 1.6T Ethernet connectivity.
The AIS1600-64O (1.6T) is a 3RU switch with 64 x 1.6T OSFP1600 ports and Flexport technology supporting breakout speeds from 100G to 1.6T. The AIS800-128O is a 4RU system with 128 x 800G OSFP800 ports and up to 512 logical ports on a single chip, designed for high-radix 800G fabrics and two-tier 400G NIC deployments using 800G breakout. Edgecore said the platforms reduce hop counts and latency in large-scale GPU clusters while supporting emerging 800G and 1.6T NIC generations.
Both systems integrate Tomahawk 6 features including hardware-based link failover and dynamic load balancing (DLB/GLB) to reduce Job Completion Time (JCT) in AI workloads. Built on a 3nm process, the switches target improved power efficiency per gigabit and support 30W per port operation across all 64 ports in the AIS1600-64O. Edgecore also emphasized open networking support, including compatibility with commercial and open-source network operating systems, PTPv2 and SyncE timing, and deep telemetry for AI fabric observability. The company will showcase the platforms at MWC Barcelona 2026 and OFC 2026.
• AIS1600-64O: 64 x 1.6T OSFP1600 ports in 3RU form factor
• AIS800-128O: 128 x 800G OSFP800 ports in 4RU form factor
• 102.4 Tbps switching capacity powered by Tomahawk 6 (3nm)
• Up to 512 logical ports on a single chip
• Breakout flexibility from 100G to 1.6T
• Hardware-based failover and DLB/GLB load balancing
• Support for PTPv2, SyncE, and open-source or commercial NOS
“The transition to 1.6T Ethernet is a critical milestone in meeting the relentless bandwidth demands of next-generation AI clusters. By utilizing our latest Tomahawk 6 silicon, Edgecore is delivering a high-performance, power-efficient platform to the open networking ecosystem that empowers customers to scale their infrastructure with confidence,” said Hasan Siraj, Vice President of Product Management, Core Switching Group at Broadcom.
🌐 Analysis: The move to 102.4 Tbps switching aligns with hyperscaler roadmaps that target 800G today and 1.6T uplinks for next-generation AI clusters. Broadcom’s Tomahawk 6 positions merchant silicon at the center of this transition, while open networking vendors such as Edgecore seek to capture design wins in disaggregated AI fabrics competing with vertically integrated systems from NVIDIA and others.
🌐 We’re tracking the latest developments in networking silicon. Follow our ongoing coverage at: https://convergedigest.com/category/semiconductors/
🌐 We’re launching the “Data Center Networking for AI” series on NextGenInfra.io and inviting companies building real solutions—silicon, optics, fabrics, switches, software, orchestration—to share their views on video and in our expert report. To get involved, send a note to [email protected] or [email protected].
