Keysight Debuts Purpose-Built 1.6T Ethernet AI Workload Emulation Platform

Keysight Debuts Purpose-Built 1.6T Ethernet AI Workload Emulation Platform to Validate Next-Generation AI Fabrics

Platform reduces risk, cost, and development time with real-world AI workload emulation and data-center-ready scalability across emerging 224G lanes

Keysight announced AresONE 1600GE, a scalable 1.6T Ethernet AI workload emulation platform designed to validate next generation AI fabrics operating over emerging 224G SerDes. The solution enables network equipment and silicon chip manufacturers, hyperscalers, and AI data center operators to emulate large scale AI compute nodes and validate AI fabrics before and after deployment.

As AI workloads drive rapid growth in scale-up and scale-out networking, the industry is transitioning to higher data rates and increased AI switch radix (number of ports on a switch) via fan-out of 800GE and 1600GE ports, improving AI fabric efficiency and reducing network tiers. However, validating new designs at these speeds presents significant challenges, including link integrity over 224G lanes, congestion behavior under microbursts, and realistically emulating AI collective communications.

To address these challenges, the purpose-built AresONE 1600GE platform combines high-density 1.6T architecture with Keysight AI Data Center Builder (KAI DC Builder) software, delivering realistic AI workload emulation required to validate next-generation AI fabrics. With KAI DC Builder, the platform emulates real world AI workloads, including multiple full stack RoCEv2 connections and a broad range of collective communication patterns. Teams can assess key fabric KPIs under conditions that closely mirror production environments, while integrated data collection and analytics improve visibility, helping engineers fine tune fabric performance.

Unifying AI workload emulation, physical layer validation, and traffic and protocol testing into a single platform, the AresONE 1600GE delivers:

AI data center and network validation at scale: Enables AI fabric deployment assessment using real AI workloads and GPU cluster emulation at high speeds. Tests GPUs, AI NICs, and accelerators pre- and post-deployment, measures network impact on job completion time, and evaluates performance isolation, load balancing, and congestion control to optimize AI training performance.
High-density OSFP 1600 hardware architecture: The rack-mount platform delivers four OSFP 1600 ports supporting flexible fan out configurations of 1 × 1600GE, 2 × 800GE, 4 × 400GE, or 8 × 200GE over 224G SerDes electrical lanes, enabling link bring-up and large-scale traffic validation in a single scalable system.
Full-stack validation from physical layer through AI-driven data and control plane: Supports optical and electrical link validation, forward error correction (FEC) and physical coding behavior, and full Layer 2/3 protocol and traffic analysis within one platform — correlating signal integrity, protocol behavior, and application performance.
Scalable data center-ready system design: High-performance processing and memory enable automated, repeatable testing, seamless lab and data center integration, and reliable day-to-day operation across development and validation cycles.
Integrates with Keysight IxNetwork: Emulates high-scale Layer 2/3 protocols and traffic at lane speed, providing engineers with greater visibility to identify issues earlier and accelerate product readiness.

Alan Weckel, Founder and Technology Analyst, 650 Group, said: “The market is rapidly shifting from foundational training to inference and agentic models. The AI interconnect market (scale-up, scale-out, scale-across, and front-end) networks will approach $200B by 2030 led by record growth in 800G and 1.6T Ethernet. The move to 1.6T will be the largest and fastest cycle ever in the industry. Keysight’s additional AI validation software features along with AresONE 1600GE hardware will help customers scale in this new market opportunity.”

Ram Periakaruppan, Vice President and General Manager, Keysight's Network Test and Security Solutions, said: “As AI workloads continue to scale, data centers are pushing speeds and complexity the industry has never seen before. With AresONE 1600GE, Keysight provides hyperscalers and other AI innovators a practical, proven way to validate next-generation network devices and infrastructure under real-world conditions, enabling faster deployment with greater confidence and significantly less risk.”

Source: Keysight media announcement

Follow @PipelineWire