AIS1600-64O

The World’s Fastest Switch Powered by Broadcom Tomahawk® 6.

AI / ML | AI Switch

  |  AIS1600-64O

AI Switch Unleashing 102.4T of Ultimate Computing Power

AIS1600-64O

The Edgecore AIS1600-64O is a 64-port 1.6T AI switch powered by Broadcom Tomahawk 6. Delivering 102.4 Tbps capacity and sub-microsecond latency, it eliminates bandwidth bottlenecks for next-gen hyperscale AI/ML clusters.

Unprecedented Scale for Generative AI Workloads

The AIS1600-64O is engineered to break bandwidth bottlenecks in hyperscale AI and ML clusters. Incorporating the groundbreaking Broadcom Tomahawk® 6 (BCM78914) silicon, it delivers an industry-leading 102.4 Tbps switching capacity with 512 x 200G PAM4 SerDes technology.

Designed for sub-microsecond latency and massive scalability, the AIS1600-64O enables the largest flat 2-tier networking topologies, significantly reducing Job Completion Time (JCT) for complex AI training models.

64 Ports of 1.6T Connectivity

Support for 64 x OSFP1600 ports allows for seamless connection to next-gen 1.6T optics or 800G NICs via breakout options. This high-density 3RU platform eliminates network blocking and maximizes throughput for data-intensive applications

Power Efficiency & Telemetry

Leveraging a monolithic 3nm process die for superior power efficiency. The platform includes programmable in-band telemetry for deep network visibility and includes a powerful Intel® Xeon® D-1713NTE (Ice Lake-D) CPU for robust management

Cognitive Routing & Load Balancing

Built to handle traffic spikes in AI clusters, the switch features Cognitive/Adaptive routing, Dynamic Load Balancing (DLB), and Global Load Balancing (GLB). Hardware-based link failover ensures maximum resilience and fastest recovery

High Radix & Unified Buffer

Supporting up to 512 logical ports (Highest Radix) and equipped with a 267MB unified fully-shared packet buffer, the AIS1600-64O effectively manages congestion and micro-bursts without packet loss

Broadcom Tomahawk® 6 (BCM78914)

Features

Extreme Connectivity

Port Density : 64 x 1.6T OSFP1600 ports.

Flexibility : Supports breakout options to 800G, 400G, 200G, and 100G via Flexport™ technology.

Power : Up to 30W power budget per port to support advanced transceivers.

Advanced Silicon & Architecture

Chipset : Broadcom Tomahawk 6C BCM78914 (Monolithic 3nm process).

Scale : Highest Radix with up to 512 logical ports per chip.

Performance : Low latency design with a 267MB unified, fully-shared packet buffer.

Optimized for AI/ML Workloads

Routing : Cognitive/Adaptive routing with Dynamic (DLB) and Global Load Balancing (GLB).

Congestion Control : Features Drop Congestion Notification (DCN), Fast CNP, and PFC-aware ECN.

Resiliency : Hardware-based link failover minimizes Job Completion Time (JCT).

Next-Gen Protocols

SRv6 : Comprehensive support including Micro SID (uSID), Generalized SID (GSID), and canonical modes (Subject to NOS).

Virtualization : Full support for VxLAN.

Scheduling : Credit-based Flow Scheduler (CBFS) for precise traffic management (Subject to NOS).

Efficient Hardware Design

Form Factor : Compact 3 RU chassis with front-to-back (AFO) airflow.

Redundancy : Hot-swappable 7+1 fan modules and load-sharing PSUs (2700W AC or 3200W DC).

Safety : Integrated e-fuses for transceiver and component protection.

Management & Synchronization

Open Networking : Pre-loaded with ONIE for automated installation of compatible NOS.

Precision Timing : Built-in SyncE and PTPv2 support (1PPS, 10MHz, ToD).

Control : BMC module with Serial-over-LAN capabilities.

Use Cases
AIS1600-64O Use Cases

Future-Proof Architecture : The Key to Seamless 800G to 1.6T

Flexibly breakout single ports to multiple servers, revealing an effortless upgrade path from current setups to next-gen infrastructure.

Related Solutions

Related Resource

2026 Product Catalogue