Nvidia begins shipping TSMC‑produced CPO switches to alleviate AI data‑center bottlenecks

Nvidia announced the start of shipments for its new CPO switches. The switches are manufactured by TSMC, a leading semiconductor foundry. They are designed to address performance bottlenecks in AI data centres. The hardware aims to improve throughput and latency for AI workloads. Nvidia says the switches complement its GPU and networking portfolio. Early customers will integrate the product into existing AI clusters. Analysts view the move as a response to rising demand for AI infrastructure. Nvidia expects the switches to help maintain its market leadership.