VIAVI Solutions launches world’s first Ultra Ethernet validation solution for AI data centers

VIAVI Solutions launches world’s first Ultra Ethernet validation solution for AI data centers

The TestCenter-based tool lets hyperscalers and equipment makers validate next-gen networking without burning through expensive GPUs

VIAVI Solutions has released what it calls the first validation solution built specifically for Ultra Ethernet Transport, a protocol designed to fix the networking bottlenecks that traditional Ethernet creates inside AI data centers. The tool, which launched on June 23, plugs directly into VIAVI’s existing TestCenter platform and lets companies emulate realistic AI workload traffic without needing actual GPUs to do it.

What Ultra Ethernet actually solves

The Ultra Ethernet Consortium, whose founding members include AMD, Arista, Broadcom, Cisco, HPE, Intel, Meta, and Microsoft, was established around 2023 to tackle exactly this problem. The group released its formal Specification 1.0 on June 11, 2025, codifying the Ultra Ethernet Transport protocol with improvements to congestion control and scalability for AI and high-performance computing environments.

The tool validates several key capabilities that the UET protocol introduces. These include reliable ordered and unordered packet delivery, packet trimming, congestion control mechanisms, and dynamic multipathing.

Advertisement

Who this is built for

VIAVI is targeting three distinct customer segments: hyperscalers building their own AI infrastructure, cloud service providers offering AI compute to customers, and the networking equipment manufacturers whose switches and NICs need to speak fluent UET.

The TestCenter platform supports emulation of collective communications patterns and large language model traffic flows. It also handles load-balancing validation for strategies like Equal-Cost Multi-Path routing and flowlet switching.

“AI clusters will soon scale to millions of endpoints,” said Aniket Khosla, VP of Product Management at VIAVI.

Khosla emphasized that the validation tool is designed to help users deploy high-performance AI fabrics with greater speed and cost-efficiency. Given that AI data centers are now operating on multi-terabit connections, the margin for networking errors is essentially zero.

The broader context

VIAVI has previously demonstrated UET interoperability capabilities that earned recognition at Interop industry events. Those demonstrations helped establish credibility with the networking community and the UEC membership before this commercial product launch.

What this means for investors

VIAVI Solutions (NASDAQ: VIAV) is positioning itself as the testing and validation layer for what could become the default networking protocol in AI data centers. Every switch vendor, NIC manufacturer, and hyperscaler that adopts UET will need to verify their implementation works correctly, and right now VIAVI is the only game in town for that specific function.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.

VIAVI Solutions launches world’s first Ultra Ethernet validation solution for AI data centers

VIAVI Solutions launches world’s first Ultra Ethernet validation solution for AI data centers

The TestCenter-based tool lets hyperscalers and equipment makers validate next-gen networking without burning through expensive GPUs

VIAVI Solutions has released what it calls the first validation solution built specifically for Ultra Ethernet Transport, a protocol designed to fix the networking bottlenecks that traditional Ethernet creates inside AI data centers. The tool, which launched on June 23, plugs directly into VIAVI’s existing TestCenter platform and lets companies emulate realistic AI workload traffic without needing actual GPUs to do it.

What Ultra Ethernet actually solves

The Ultra Ethernet Consortium, whose founding members include AMD, Arista, Broadcom, Cisco, HPE, Intel, Meta, and Microsoft, was established around 2023 to tackle exactly this problem. The group released its formal Specification 1.0 on June 11, 2025, codifying the Ultra Ethernet Transport protocol with improvements to congestion control and scalability for AI and high-performance computing environments.

The tool validates several key capabilities that the UET protocol introduces. These include reliable ordered and unordered packet delivery, packet trimming, congestion control mechanisms, and dynamic multipathing.

Advertisement

Who this is built for

VIAVI is targeting three distinct customer segments: hyperscalers building their own AI infrastructure, cloud service providers offering AI compute to customers, and the networking equipment manufacturers whose switches and NICs need to speak fluent UET.

The TestCenter platform supports emulation of collective communications patterns and large language model traffic flows. It also handles load-balancing validation for strategies like Equal-Cost Multi-Path routing and flowlet switching.

“AI clusters will soon scale to millions of endpoints,” said Aniket Khosla, VP of Product Management at VIAVI.

Khosla emphasized that the validation tool is designed to help users deploy high-performance AI fabrics with greater speed and cost-efficiency. Given that AI data centers are now operating on multi-terabit connections, the margin for networking errors is essentially zero.

The broader context

VIAVI has previously demonstrated UET interoperability capabilities that earned recognition at Interop industry events. Those demonstrations helped establish credibility with the networking community and the UEC membership before this commercial product launch.

What this means for investors

VIAVI Solutions (NASDAQ: VIAV) is positioning itself as the testing and validation layer for what could become the default networking protocol in AI data centers. Every switch vendor, NIC manufacturer, and hyperscaler that adopts UET will need to verify their implementation works correctly, and right now VIAVI is the only game in town for that specific function.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.