The piece by Kamal Mann critiques the traditional cloud storage model, particularly its limitations when handling the demands of modern AI workloads. Mann points out that the existing infrastructure is often not designed to manage the massive volumes of data generated by AI applications. He highlights that these workloads require not only substantial storage capacity but also low-latency data access, which current systems struggle to provide. Mann delves into the inefficiencies of the "store everything" approach. He argues that indiscriminate data storage leads to increased costs and complexity, as organizations grapple with managing vast amounts of irrelevant or redundant data. He suggests that businesses need to adopt a more strategic approach to data management, focusing on relevance and accessibility rather than sheer volume. The article also touches on potential solutions, including the evolution of hybrid cloud models and edge computing. These approaches could help alleviate some of the pressures on centralized cloud storage by distributing data processing closer to where it's generated. Mann emphasizes the necessity for organizations to rethink their data strategies to effectively support AI initiatives and improve overall operational efficiency.