Keysight Technologies Inc. (NYSE: KEYS) launched Keysight AI Inference Builder, an emulation and analytics platform designed to validate AI inference infrastructure at scale. The company announced the platform will be demonstrated at NVIDIA GTC within NVIDIA DSX Air digital twin environments.

The platform addresses testing challenges as the AI industry shifts from training large language models to deploying them for inference. Traditional testing methods using synthetic traffic generation or GPU benchmarks cannot accurately reproduce latency-sensitive workload behavior across compute, networking, memory, storage and security layers, according to the company.

KAI Inference Builder recreates realistic inference workload patterns and models industry-specific usage patterns to validate AI infrastructure, applications and data center deployments. The platform targets AI cloud providers, hardware vendors and application developers seeking to measure and optimize real-world inference performance.

The platform includes industry-specific benchmarking capabilities that emulate usage patterns for AI models in finance, healthcare and other sectors. It evaluates inference workflows from user request to model response and can perform client-only emulation to identify performance bottlenecks across the AI infrastructure stack.

"Inference is the key to unlocking AI's ROI, but that can be challenging to achieve when system resources aren't optimized for capacity and performance," said Ram Periakaruppan, vice president and general manager of Network Test & Security Solutions at Keysight.

NVIDIA's Amit Katz, vice president of networking, said the integration with NVIDIA DSX Air provides an environment to eliminate performance volatility and enables partners to emulate real inference workloads before deployment.

Keysight will demonstrate the platform at NVIDIA GTC in San Jose from March 16-19 at both the NVIDIA booth and Keysight's booth 3217.