Institution: Storage faces a structural shortage lasting up to several years; large-scale deployment of CPO is delayed until the end of 2028

Jin10 Data July 10 (reported): SemiAnalysis founder Dylan Patel, a leading research institution focusing on semiconductors and AI infrastructure, recently discussed on a podcast the core dynamics and investment logic across the current AI infrastructure stack. Dylan emphasized that storage faces a structural shortage lasting for years, with still 2 to 3 times upside potential; meanwhile, although agents and reinforcement learning are driving CPU demand, the sell-side market has priced this too high—CPU growth is mainly driven by historical “catch-up,” and its absolute value in AI servers is still far less than that of GPUs. Dylan believes that the timeline for the long-awaited large-scale rollout of co-packaged optics (CPO) has been clearly pushed back to the end of 2028 to 2029, unexpectedly extending the benefit period for copper-cable connectors.
(From the Jin10 Data app)