According to the new AI Networks for AI Workloads report by Dell’Oro Group – a trusted source for market information about the telecommunications, security, networks and data centre industries – spending on switches deployed in AI back-end networks is forecast to expand the data centre switch market by 50%.
Current data centre switch market spending is on front-end networks used primarily to connect general-purpose servers. AI workloads will require a new back-end infrastructure buildout. The competition intensifies between InfiniBand and Ethernet as manufacturers vie for market dominance in AI back-end networks. While InfiniBand is expected to maintain its lead, Ethernet is forecast to make substantial gains, such as 20 revenue-share points by 2027.
“Generative AI applications usher in a new era in the age of AI, standing out for the sheer number of parameters that they have to deal with,” said Sameh Boujelbene, Vice President, Dell’Oro Group. “Several large AI applications currently handle trillions of parameters, with this count increasing tenfold annually. This rapid growth necessitates the deployment of thousands or even hundreds of thousands of accelerated nodes.
“Connecting these accelerated nodes in large clusters requires a data centre-scale fabric, known as the AI back-end network, which differs from the traditional front-end network used mostly to connect general-purpose servers.
“This predicament poses the pivotal question: what is the most suitable fabric that can scale to hundreds of thousands and potentially millions of accelerated nodes while ensuring the lowest Job Completion Time (JCT)? One could argue that Ethernet is one speed generation ahead of InfiniBand. Network speed, however, is not the only factor,” added Boujelbene.
“Congestion control and adaptive routing mechanisms are also important. We analysed AI back-end network build-outs by the major Cloud Service Providers (such as Google, Amazon, Microsoft, Meta, Alibaba, Tencent, ByteDance, Baidu and others) as well as various considerations driving their choices of the back-end fabric to develop our forecast.”
Additional highlights from the AI Networks for AI Workloads Report:
- AI networks will accelerate the transition to higher speeds. For example, 800Gbps is expected to comprise the majority of the ports in AI back-end networks by 2025, within just two years of the latest 800Gbps product introduction.
- While most of the market demand will come from Tier 1 Cloud Service Providers, Tier 2/3 and large enterprises are forecast to be significant, approaching US$10 billion over the next five years. The latter group will favour Ethernet.