The mass production of Co-Packaged Optics (CPO) has not encountered any delays. The perceived discrepancies in the market may largely be attributed to the misinterpretation of timelines, particularly the confusion between the stages of 'small-batch validation and market introduction' and 'widespread adoption across the entire industry.' Presently, the primary bottleneck for CPO, as well as high-speed optical communication products, lies in the upstream supply of indium phosphide laser chips. The setup of production lines for these chips, along with the rigorous customer validation processes, are both time-intensive endeavors, thereby constraining the expansion of production capacity.

Looking ahead, CPO continues to represent the optimal solution for ultra-large model training clusters in the long run. However, in the near term, a variety of alternatives, including Near-Packaged Optics (NPO), Linear Pluggable Optics (LPO), traditional pluggable optical modules, or multi-channel parallel technologies, will collectively cater to the burgeoning demand for high-bandwidth computing. The deployment of high-speed optical interconnect technology is anticipated to unfold in a phased manner, with upstream capacity enhancements spearheading the initiative and progressively advancing through distinct stages.