Unbox one of NVIDIA's first co-packaged optics switches with us. See why we bet on CPO early.
When we design large GPU clusters, the network is no longer a background system. It's part of the compute envelope. At the 800G and NVIDIA GB300 NVL72 scale, the back-end fabric accounts for 86% of networking power in a three-layer cluster. This matters even more as we shift towards agentic...
When we design large GPU clusters, the network is no longer a background system. It's part of the compute envelope. At the 800G and NVIDIA GB300 NVL72 scale, the back-end fabric accounts for 86% of networking power in a three-layer cluster. This matters even more as we shift towards agentic workloads. A single request can fan out across multiple model calls, tool calls, retrieval steps, and reasoning passes. That creates more east-west traffic inside the cluster and puts the network closer to the critical path of token generation. If GPUs are waiting on data or recovering from failures, token throughput drops. Lambda is taking an early look at co-packaged optics, starting with the NVIDIA Quantum-X InfiniBand Photonics Q3450-LD switch. In this post, you'll learn what the hardware is, what changes operationally, insights from the unboxing, and why this architecture fits large NVIDIA GB300 NVL72 clusters.Source: Lambda Labs — Published — Category: Models