AI’s memory stack: What investors need to know

Investing.com reports that artificial intelligence is creating a growing memory bottleneck, with demand spreading beyond high-bandwidth memory into conventional DRAM, NAND flash, and storage. This new memory landscape presents investment opportunities, as AI workloads such as training, inference, retrieval-augmented generation, and agentic AI place different demands on the memory stack.

Training is highly compute-intensive and limited by memory bandwidth, while inference requires GPUs and high-bandwidth memory (HBM) for the compute-bound initial stage and memory-bound subsequent decode stage. The emerging technologies, including CXL memory, Nvidia’s Storage Next initiative, and CMX context storage, aim to balance performance, capacity, and cost as AI systems require larger pools of memory.

Memory manufacturers, such as Samsung Electronics, SK hynix, Micron, SanDisk, Seagate, and Western Digital, are developing technologies like high-bandwidth flash to address the memory challenges posed by AI.

Brief written by urgent.news from Investing.com's own syndicated text. Machine-written — may contain errors; check the original before relying on it.