Artificial intelligence news.
Straight from the original announcements.

Lambda deploys NVIDIA Photonics switches across 10,000+ GPUs

November 16, 2026, 17:03By the AI News Bites Editorial TeamSource: Lambda
Lambda deploys NVIDIA Photonics switches across 10,000+ GPUs
Image: Lambda

Lambda says it has completed the world’s first large-scale deployment of co-packaged optics networking, connecting 10,000+ NVIDIA Blackwell Ultra GPUs at its Quebec campus. The company reports lower latency, more stable signals and cooler optics than conventional pluggable networking.

Key points
  • Lambda connected 10,000+ NVIDIA Blackwell Ultra GPUs in Quebec.
  • The deployment uses NVIDIA Photonics switches with 144 800Gb/s ports.
  • CPO latency was 3.8 percent lower at rack level and 10.1 percent lower at pod level.
  • CPO optics had a median temperature of 36°C, versus 51°C for pluggable optics.
  • Lambda says the optics provide up to 15x higher reliability.

CPO connects thousands of GPUs

The deployment uses NVIDIA Photonics networking switches across Lambda’s Quebec campus. Lambda first announced its early adoption of NVIDIA CPO networking in November 2025, and has now benchmarked the technology against conventional optics in rack, scalable unit and pod configurations.

Co-packaged optics places optical engines alongside the switch ASIC. In contrast, conventional pluggable optics convert electrical signals to light at the switch’s front panel. According to Lambda, the shorter electrical path, reduced from inches to millimeters, cuts signal loss, removes the need for retimer chips and improves network resiliency.

“Networking has to keep pace with the demands of increasingly large AI clusters. Bringing co-packaged optics into a deployment of this scale is an important step toward building the efficient and resilient infrastructure needed to support the next generation of AI,” said Dave Ward, President of Cloud Services at Lambda.

Lower latency and cooler operation

Lambda says CPO delivered identical throughput and more stable signal integrity compared with conventional pluggable optics in its benchmarks. Latency was lower at every tested scale, with the difference increasing from 3.8 percent at rack level to 10.1 percent at pod level.

The CPO optics also recorded a lower median temperature, 36°C compared with 51°C for conventional pluggable optics. Lambda says cooler and more stable optics drive up to 15x higher reliability, helping keep large training jobs running.

The NVIDIA Photonics switches used in the deployment feature 144 800Gb/s ports, liquid cooling and 18 removable external light sources. NVIDIA describes the networking system as reducing power consumption and improving resiliency. “In an AI factory, the network is a business accelerator. Every interruption stalls the job, and every watt counts. NVIDIA Photonics cuts network power, improves resiliency and delivers up to 15x higher reliability, so Lambda’s customers get more productive GPU hours from their investment,” said Gilad Shainer, Senior Vice President Networking at NVIDIA.

A deployment tied to a lower-energy campus

Lambda says the Quebec campus has 142 MW of secured, nearly 100% renewable energy capacity. The site enables free cooling for up to 80% of the year and is guaranteed to a PUE of 1.2, with no process water consumption associated with server cooling.

The deployment will be featured in NVIDIA’s SC26 special address, which will provide a closer look at Lambda’s implementation of CPO networking at scale. For Lambda, the announcement links optical networking improvements with the growing infrastructure demands of large AI clusters. The company says its customers range from AI researchers to enterprises and hyperscalers, making network latency, reliability and power use relevant to the operation of its AI cloud infrastructure.

Related news

NVIDIA and Microsoft Extend AI Infrastructure From Azure to Windows PCsTop story
Chips and infrastructureNovember 19, 2026, 21:00

NVIDIA and Microsoft Extend AI Infrastructure From Azure to Windows PCs

NVIDIA and Microsoft are expanding their accelerated computing partnership across Azure, industrial systems and Windows PCs, adding access to Vera Rubin infrastructure…

Lambda reports faster AI training with NVIDIA Blackwell Ultra GPUs
Chips and infrastructureNovember 16, 2026, 17:31

Lambda reports faster AI training with NVIDIA Blackwell Ultra GPUs

Lambda reports a 2.86× improvement in mixture of experts language model training and a 1.50× improvement in image model training in the latest MLPerf round. The results…

Lambda Deploys NVIDIA Photonics Across a 10,000-GPU AI Factory
Chips and infrastructureNovember 16, 2026, 17:00

Lambda Deploys NVIDIA Photonics Across a 10,000-GPU AI Factory

Lambda is using NVIDIA Photonics co-packaged optics switches to connect NVIDIA GB300 NVL72 racks containing more than 10,000 GPUs. NVIDIA says the deployment delivers…