Investment banking giant Morgan Stanley released analysis on Sunday suggesting that evolving artificial intelligence technology will expand semiconductor investment opportunities far beyond the graphics processing chips that have fueled the current AI surge.
The financial services firm predicts that as artificial intelligence evolves from simple response generation to independent decision-making capabilities, computing demands will shift significantly toward central processing units and memory systems.
“As AI transitions from generation to autonomous action, the computing bottleneck is shifting towards CPU and memory, driving a step-change in general-purpose compute intensity,” Morgan Stanley analysts wrote in their weekend research note, while emphasizing that graphics processor demand continues at high levels.
The bank’s projections indicate that autonomous AI systems could contribute between $32.5 billion and $60 billion to a data center CPU marketplace already valued at over $100 billion by the end of this decade.
This technological shift involves what experts call “agentic AI” – sophisticated systems capable of independent task planning and execution rather than simply responding to user commands. Morgan Stanley analysts believe future AI development will prioritize coordination capabilities over pure computational strength.
Central processors are becoming essential control mechanisms for AI applications handling complex, multi-step operations. Meanwhile, memory component requirements are expected to surge dramatically, expanding AI-related spending across multiple technology sectors including chip manufacturing, memory production, and equipment suppliers.
The investment firm suggests that companies operating in supply-limited market segments may gain enhanced pricing advantages as demand increases.
Morgan Stanley identified several potential beneficiaries of this market evolution, including Nvidia, AMD, Intel and Arm for processors and accelerators; Micron, Samsung and SK Hynix for memory solutions; and TSMC and ASML for manufacturing and equipment needs.
