HaiAI123

Curated Global AI Tools Directory

Industry update
Flat

Lambda connects 10,000+ GB300 GPUs with NVIDIA co-packaged optics switches

11/17/2026 — 11/17, 00:19·1 sources (all)·1 reports (all)

Key takeaways

  • NVIDIA's blog says Lambda used NVIDIA Photonics co-packaged optics switches to connect GB300 NVL72 racks at a scale of over 10,000 GPUs, cut…

Background & analysis

On November 17, 2026, NVIDIA's blog reported that cloud provider Lambda is deploying co-packaged optics (CPO) switches from NVIDIA Photonics to connect NVIDIA GB300 NVL72 racks at a scale of more than 10,000 GPUs. According to the post, training an AI model requires thousands of GPUs to work together, which calls for a network that is fast, efficient, and resilient, and Lambda is addressing the networking layer of that requirement with the CPO switches. Lambda is using the switches across its AI factory environment, and NVIDIA states that the design delivers lower power consumption, higher resiliency, and a longer mean time between interruptions (MTBI). Beyond the headline deployment figure, NVIDIA also notes that in its testing the rack-level network latency dropped by roughly 4 percent, and Pod-level latency dropped by up to roughly 10 percent, while throughput remained on par with traditional pluggable optical modules. The report presents the deployment as Lambda's approach to scaling the networking portion of large-scale AI training infrastructure.

AI-generated from 1 reports · updated 12 hours ago

updream
24-hour heatHeat index 70 · peak 97 · 11h ago
24 hours agonow

Reports on this story headlines open the original

Today
  1. NVIDIA's blog says Lambda is using NVIDIA Photonics co-packaged optics switches to connect GB300 NVL72 racks with more than 10,000 GPUs. In the company's tests, network latency dropped by about 4% at rack scale and up to about 10% at pod scale while maintaining the same throughput as conventional pluggable optics.

    NVIDIA BlogFirst-partyAI score 46

Other stories people are talking about

How is heat calculated?About the method

Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.

This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.

Surge
Discussion rising fast
New
First report within 6 hours
Rising
Still gathering discussion

Back to AI industry updates →