Morgan Stanley: Agent AI Will Expand Chip Spending, Moving Beyond GPUs to CPUs
Morgan Stanley says increasingly autonomous AI may boost demand for central processing units (CPUs), reshape data center construction, and expand investment beyond the graphics chips that have dominated the AI boom so far.
Morgan Stanley said in a note on Sunday that “as AI transitions from generative to autonomous action, compute bottlenecks are shifting to CPUs and memory, driving a step-change in general-purpose compute intensity,” adding that demand for graphics processing units (GPUs) remains strong.
Morgan Stanley estimates that agent AI will add $32.5 billion to $60 billion to the data center CPU market by 2030, which already exceeds $100 billion.
Agent AI refers to systems capable of planning tasks and taking action on their own, rather than simply responding to prompts.
Morgan Stanley said the next wave of agent AI will be driven more by orchestration than raw computing power.
CPUs are increasingly acting as the control layer for AI systems managing multi-step tasks.
Memory demand will surge, expanding AI spending beyond GPUs to chipmakers, memory suppliers, and manufacturing.
The brokerage added that companies with constrained supply in the ecosystem may gain more pricing power.
Morgan Stanley identified Nvidia, AMD, Intel and Arm in CPUs and accelerators; Micron, Samsung and SK Hynix in memory; and TSMC and ASML in chip manufacturing and equipment as potential beneficiaries.