Saturday, 21 November 2026

Processor Press

The day's headlines, with the volume turned up.

AI Wire is original plain-English briefs — not satire. About →

Processor Press

AI Wire

AI Wire · not satire. Original plain-English briefs inspired by public announcements — not investigative reporting.

Lambda Boosts AI Training Resilience with NVIDIA Photonics Co‑Packaged Optics

Training modern AI models requires thousands of GPUs to work in concert, and the network that stitches them together must be fast, efficient and fault‑tolerant. Lambda has turned to NVIDIA’s co‑packaged optics (CPO) switches to meet that need. The CPO switches are built directly into the GPU racks, allowing Lambda to interconnect its NVIDIA GB300 NVL72 racks at line‑rate speeds without the extra cabling and latency that traditional switches introduce.

With the new CPO‑based fabric, Lambda can link more than 10,000 GPUs across its data‑center. The result is a more resilient network that can recover quickly from node or link failures, and one that delivers higher throughput for the massive data shuttles that AI training demands. In practice, this means faster model convergence and lower operational costs for Lambda’s customers.

The partnership also underscores NVIDIA’s growing focus on photonics as a key enabler for next‑generation AI infrastructure. By embedding optical switching directly into the rack, Lambda and NVIDIA are pushing the limits of what can be achieved with conventional copper‑based networking. While the exact performance gains are still being measured, the move signals a broader shift toward optical interconnects in high‑density GPU clusters.

if this one landed.