Agentic AI Shift Poised to Broaden Chip Spending Beyond GPUs

Key Moments

Agentic AI Seen Redefining Data-Center Compute Needs

Morgan Stanley said on Sunday that increasingly autonomous artificial intelligence is likely to spur greater use of central processing units (CPUs), reshape how data centers are built, and extend AI-related investment beyond the graphic processing units (GPUs) that have led the current wave of AI infrastructure spending.

“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 said in a note, while emphasizing that demand for GPUs remains strong.

Market Impact and CPU Demand Outlook

According to Morgan Stanley, agentic AI – systems that can plan tasks and take actions independently rather than only reacting to user prompts – could add between $32.5 billion and $60 billion to the data-center CPU segment by 2030. The firm said this would come on top of a CPU market that already exceeds $100 billion.

The note described CPUs as playing an expanding role as the control layer within AI architectures, managing and orchestrating complex, multistep workflows as these systems advance.

Shift From Pure Compute to Coordination

Morgan Stanley said that the next phase of agentic AI development will be driven more by coordination functions than by sheer computing power alone. As AI models increasingly manage sequences of tasks and make decisions autonomously, the brokerage expects CPU-centric control and memory to become more critical components of the overall stack.

Rising Memory Needs and Broader AI Spending

The firm projected a sharp increase in memory demand as AI systems evolve, which it said is likely to extend AI-related capital expenditure beyond GPUs to a wider range of semiconductor and equipment providers. Morgan Stanley indicated that this expansion could benefit chipmakers, memory manufacturers, and companies involved in chip fabrication and tools.

The brokerage also noted that firms operating in areas of the supply chain that remain capacity constrained could experience greater pricing power as demand for AI-related hardware accelerates.

Potential Beneficiaries Across the AI Hardware Stack

Morgan Stanley highlighted a group of companies it views as well positioned in this environment. In CPUs and accelerators, it pointed to Nvidia, AMD, Intel, and Arm. In memory, it cited Micron, Samsung, and SK hynix. Across manufacturing and equipment, the firm named TSMC and ASML as key players.