Nvidia and silicio photonics: horizon 2026

According to Digitimes, Nvidia is accelerating plans for the adoption of silicio photonics, with the goal of starting mass production by 2026. This strategic move could revolutionize the way data is transferred within and between computing systems, particularly for applications that require very high bandwidth, such as large language model (LLM) inference and high-performance computing (HPC).

Silicio photonics promises to overcome the limitations of traditional electrical interconnects, offering higher speeds and lower energy consumption. The large-scale adoption of this technology by Nvidia could trigger a significant transformation in the optical supply chain, with implications for component manufacturers and service providers.

For those evaluating on-premise deployment of AI workloads, there are trade-offs between performance, costs, and infrastructure requirements. AI-RADAR offers analytical frameworks on /llm-onpremise to evaluate these options.