As artificial intelligence workloads grow larger and more complex, the infrastructure powering modern data centres is undergoing a rapid transformation. The shift is being driven by one key challenge: how to move enormous volumes of data faster, cheaper and more efficiently.
Traditionally, copper cables and printed circuit boards (PCBs) have formed the backbone of short-distance connections inside data centres. PCBs are widely used for internal server connections, while copper cables link components within and between servers. Their popularity stems from relatively low costs and lower power consumption for short-range communication.
However, the limitations of copper-based technologies are becoming increasingly visible as AI computing scales up. Higher transmission speeds and longer connection distances can weaken signal quality, making traditional copper systems less efficient for next-generation AI infrastructure.
This is where optical fibre technologies are gaining importance. Fibre-based connections are emerging as the preferred option for long-distance and high-speed interconnections because they can transmit data with lower latency and reduced signal loss.
The evolution of AI data centres is also reshaping how connectivity technologies are deployed. Industry experts point to three major trends. First, copper cable technology is advancing from Direct Attach Copper (DAC) to Active Copper Cable (ACC) and Active Electrical Cable (AEC), enabling longer-distance connectivity than earlier systems.
Second, PCBs are no longer limited to intra-tray applications. They are increasingly being used for connections within racks, helping operators improve integration and simplify hardware layouts.
Third, optical technologies are expanding beyond traditional “scale-out” networking and moving into “scale-up” architectures. Technologies such as Active Optical Cables (AOC) and Co-Packaged Optics (CPO) switches are now supporting shorter-distance, high-bandwidth connections as well.
The transition reflects a broader race within the technology industry to build AI data centres that can deliver faster bandwidth, support larger computing clusters, reduce deployment complexity and lower long-term operating costs.
