Interconnect functionality within data centres is the critical infrastructure that allows systems and semiconductors to talk to each other. For two decades, interconnect was taken for granted and included as a piece of every server, storage system, and networking device.
However, the expanded use of artificial intelligence (AI), machine learning (ML), server acceleration and memory expansion combined with the quest for efficiencies in hyperscale, edge, and private cloud infrastructure are causing disruption. These market factors are forcing the industry to reinvent and logically disaggregate the interconnect based on technologies such as Ethernet and Compute Express Link (CXL) to create better economies of scale and efficiencies.
The 650 Group report, Interconnect Semiconductor Market 2022-2027, forecasts the worldwide market for the portion of semiconductors used for interconnect in data centres will reach US$25 billion in revenue by 2027.
“The interconnect technologies that currently sit in disparate systems and silicon will need to come together to address the scalability and sustainability of cloud and data centre computing,” said Alan Weckel, a technology analyst at 650 Group.
“Similar to the offload that occurred in NICs and the scale of Ethernet in the cloud, which allowed those markets to double in size, we expect that the interconnect market will have a higher growth rate than the overall systems market and nearly double in size by 2027 as new products and solutions enter the market.” Alan Weckel