Keysight Unveils AI Data Center Builder for Network Architecture, Host Design

US-based Keysight Technologies has introduced the Keysight AI Data Center Builder, a software suite designed to validate and optimise AI infrastructure by emulating real-world workloads. 

This tool evaluates how new algorithms, components, and protocols impact AI training performance, the company announced. 

The KAI Data Center Builder integrates large language models (LLMs) and other AI model training workloads into the design and validation of AI infrastructure components, including networks, hosts, and accelerators.

Keysight customers can access a library of LLM workloads like GPT and Llama, along with popular model partitioning schemas.

Ram Periakaruppan, VP and general manager of network test and security solutions at Keysight, emphasised the importance of early validation in AI infrastructure design. 

“As AI infrastructure grows in scale and complexity, the need for full-stack validation and optimisation becomes crucial. To avoid costly delays and rework, it’s essential to shift validation to earlier phases of the design and manufacturing cycle.” 

The KAI Data Center Builder enhances AI training by aligning model partitioning strategies with AI cluster topology and configuration. Key considerations include the scale-up design of GPU interconnects, scale-out network design, and configuration of network load balancing. 

The solution reproduces network communication patterns of real-world AI training jobs to accelerate experimentation and provide insights into performance degradation. 

The KAI Data Center Builder is part of the KAI architecture, which offers end-to-end solutions for scaling AI data centers. Juniper Networks, a US-headquartered company in AI networking and cloud, has been one success story by improving its AI network efficiency with Keysight. 

“With Keysight, we are able to replicate the real data centre environment—specifically for AI data centre requirements, such as RoCEv2,” said Mahesh Subramaniam, director of product management at Juniper Networks.

Recently, Keysight partnered with Samsung and NVIDIA to develop AI models to improve radio access network (RAN) performance for 5G advanced and 6G technologies. This occurred early next month at the Mobile World Congress (MWC) 2025 in Barcelona.