NVIDIA's HBM shortage slows GPU development, SSD gains momentum - KuCoin

This article explores how storage technologies are evolving in the era of AI agents, fundamentally reshaping data processing paradigms. As AI agents transition from single model invocations to continu... Read original →

The shortage of HBM from NVIDIA is forcing a reconfiguration of the GPU supply chain, temporarily constraining its expansion in AI inference and training. However, this bottleneck is accelerating SSD vendors like Samsung to advance storage computing capabilities, particularly through NVMe-enabled features such as encryption and lifecycle governance, creating competitive differentiation. From a geopolitical perspective, the fragility of global semiconductor supply chains is becoming more evident, especially at key nodes in Taiwan, China and Hong Kong, China, prompting companies to adopt localized and diversified sourcing strategies. In response, NVIDIA may accelerate partnerships with IDM manufacturers to boost HBM production at 3nm and below, while competitors could pursue in-house storage architectures to reduce dependency. Over the next 12–24 months, the evolving demands of AI agents will drive a new generation of storage devices with intelligent indexing, secure traceability, and cross-domain collaboration, transforming storage from a data container into a decision-making core.