The inefficiencies exposed by AI workloads point to a fundamental shift in how storage and compute must interact.

InfoWorld is a source of news, analysis, and commentary on technology trends, IT strategies, and business innovation. With a focus on enterprise technology and digital transformation, InfoWorld offers insights and guidance for IT decision-makers, software developers, and technology professionals. From  articles on cloud computing and cybersecurity to product reviews and industry trends, InfoWorld helps readers navigate the complexities of modern IT environments and make informed decisions to drive business success.

InfoWorld

Traditional cloud architectures separate storage and compute, which worked well for structured analytics but creates significant inefficiencies for AI workloads. AI pipelines repeatedly read, transform, and reshape large volumes of unstructured data, causing redundant I/O, GPU underutilization (93% of organizations report this), and data scientists spending up to 80% of their time on data prep. The proposed solution is 'smart storage' — moving compute closer to where data lives so transformations happen once and are reused across pipelines. This reduces data movement, improves GPU utilization, and lowers infrastructure costs. The author, who founded DataPelago to address this problem, argues the next generation of AI infrastructure will be defined by how intelligently storage and compute are integrated.

Why AI requires rethinking the storage-compute divide