• Nokia wants to be the leading provider of AI datacentre networking technology
• It has set up the AI Networking Innovation Lab in California and attracted a number of key partners to develop and test new network designs

In what looks like a pretty smart move as AI datacentre operators try to figure out the  most efficient designs for their facilities, Nokia has set up what it calls the AI Networking Innovation Lab in Sunnyvale, California, to “accelerate the development of next-generation networking technologies for artificial intelligence (AI) infrastructure.”

According to the vendor, the lab will serve as a “testing ground for Nokia Validated Designs [NVDs] and a co-innovation hub with global AI and cloud partners.” 

The vendor, which has identified datacentre-related networking as its primary area of potential sales growth in the coming years as it seeks to take advantage of the ‘AI supercycle’, has already used the lab’s facilities to develop some of these NVDs and has been set up to help validate the designs “under real AI networking conditions, covering training and real-time inference workloads,” the vendor noted in this blog. It noted that some NVDs are already available for download and inspection and that some are “tailored with partners for specific use cases” – for example, it has developed one for sovereign AI with Lenovo  and another for training and inferencing with AI server vendor Supermicro and storage specialist Weka. 

Nokia, which took Nvidia on board as an investor last year, is pushing the partnership angle hard for this lab, noting that it has already signed up AMD, Everpure, Keysight, Nscale and Viavi Solutions, as well as Lenovo, Supermicro and Weka, as lab collaborators. 

Nokia noted: “A strong example of ecosystem collaboration is the development of highly optimised and validated designs for front-end and back-end networks in AI factories built on AMD and Nvidia GPUs and network interface cards (NICs).”

But why has Nokia felt the need to set up this facility and develop these new designs? It explains: “AI workloads are fundamentally changing how datacentre networks must operate. The performance, scale and precision required to support large-scale AI training and distributed, real-time inference place unprecedented demands on networking infrastructure. To address these challenges, Nokia is adopting a new approach to how technologies are integrated, tested and deployed from the ground up for the AI era.” 

- Ray Le Maistre, Editorial Director, TelecomTV