VIAVI Solutions debuts Ultra Ethernet Transport validation for AI data center networks

2 Sources

Share

VIAVI Solutions has launched the industry's first Ultra Ethernet Transport validation solution for AI networks, offering GPU-free testing capabilities for hyperscalers and cloud providers. The new VIAVI TestCenter offering validates AI back-end networks using the Ultra Ethernet Consortium's stack, addressing the challenge of scaling AI clusters to millions of endpoints without relying on physical GPUs.

VIAVI Solutions Introduces Industry-First Ultra Ethernet Transport Validation

VIAVI Solutions has unveiled what it describes as the industry's first validation solution for AI fabrics using Ultra Ethernet Transport, marking a shift in how organizations test and deploy large-scale AI infrastructure

1

2

. The GPU-free offering for VIAVI TestCenter targets hyperscalers, cloud providers, neocloud providers, and network equipment manufacturers across the broader Ultra Ethernet ecosystem. This launch addresses a growing challenge in AI data center networks: validating performance and resiliency without the cost and complexity of deploying dedicated GPU infrastructure for testing purposes.

The solution validates AI back-end networks using the Ultra Ethernet Stack developed by the Ultra Ethernet Consortium, which is purpose-built for large-scale AI and High-Performance Computing workloads

2

. UEC 1.0 introduces the UET protocol that delivers advanced congestion control and massive scalability to accelerate multi-vendor adoption of Ethernet as a primary transport for AI networks.

Emulating Complex AI Traffic Patterns at Scale

The VIAVI TestCenter platform emulates the transport layer of UET, replicating realistic, stateful AI traffic patterns at scale

1

. These capabilities include reliable ordered and unordered delivery, packet trimming, congestion control, and dynamic multipathing. The solution also supports full-fidelity emulation of AI workloads such as collective communications and large language model flows, which are critical for training and deploying modern AI systems.

Aniket Khosla, Vice President of Product Management for Optical Transport and High-Speed Ethernet at VIAVI Solutions, explained the practical necessity driving this innovation: "AI clusters will soon scale to millions of endpoints, which means relying on physical GPUs alone to validate network behavior is no longer practical"

1

. This statement underscores the scalability challenges facing organizations building next-generation AI infrastructure.

Comprehensive Load-Balancing Validation

The platform enables comprehensive validation of load-balancing mechanisms across AI fabrics, including ECMP, packet spraying, and flowlet switching

2

. These capabilities matter because effective load-balancing directly impacts the performance of distributed AI training and inference workloads. Mahesh Subramaniam, Senior Director of Product Management for AI Data Centers at HPE, noted that the VIAVI TestCenter platform provides traffic emulation and visibility for validating congestion control and transport performance

1

. HPE's collaboration with VIAVI Solutions includes UET transport validation using the Juniper QFX5240 platform and Junos Evolved.

Market Position and Future Implications

VIAVI Solutions, which provides test and measurement and optical technologies for data centers and communication networks, has seen remarkable momentum with a market capitalization of $12.97 billion and shares surging 450% over the past year to $52.94

1

. Revenue grew 30.6% in the last twelve months, with seven analysts recently revising earnings upward and the company expected to return to profitability this year. The launch of this validation solution for AI fabrics positions VIAVI Solutions to capture growing demand as organizations accelerate deployment of high-speed AI networks. As AI clusters continue expanding toward millions of endpoints, the ability to validate network behavior without physical GPUs becomes increasingly valuable for reducing infrastructure costs and accelerating time-to-deployment.

Today's Top Stories

© 2026 TheOutpost.AI All rights reserved