Morgan Stanley anticipates that as artificial intelligence evolves towards autonomous action, the demand for central processing units (CPUs) will surge. This shift is expected to reshape data center development and broaden investment beyond the graphics chips that have powered the current AI boom.
Analyst notes suggest that the computing bottleneck is moving towards CPUs and memory as AI transitions from generation to autonomous tasks, intensifying general-purpose compute requirements. While demand for graphic processing units (GPUs) remains robust, the next wave of AI will emphasize coordination over raw power.
Morgan Stanley estimates that agentic AI could inject an additional $32.5 to $60 billion into the data-center CPU market, already valued at over $100 billion by 2030. CPUs are increasingly serving as the control layer for AI systems executing multi-step tasks. Consequently, memory demand is also poised for sharp increases.
This broadened spending is expected to benefit chipmakers, memory suppliers, and manufacturers. Companies in supply-constrained sectors of the ecosystem could see enhanced pricing power. Potential beneficiaries identified include Nvidia, AMD, Intel, and Arm for CPUs and accelerators, and Micron, Samsung, and SK Hynix for memory, with TSMC and ASML noted for chipmaking and equipment.