According to Reuters, Morgan Stanley said in a report released the previous day that as AI moves beyond content generation toward autonomous execution, CPUs and memory, along with GPUs, are becoming key components that determine overall performance.
The bank said the shift could change how data center servers are configured and lift demand for CPUs, while adding that GPU demand remains strong.
Morgan Stanley estimated that agentic AI could add $32.5 billion to $60 billion to a data center CPU market expected to exceed $100 billion by 2030.
As AI workloads are broken into multiple steps, the bank said, CPUs are likely to serve as a control layer and memory demand could also rise quickly.
Companies operating in supply-constrained areas could gain greater pricing power, it said.
Morgan Stanley listed Nvidia, AMD, Intel and Arm as potential beneficiaries in CPUs and accelerators; Micron, Samsung Electronics and SK hynix in memory; and TSMC and ASML in manufacturing and equipment.
Reuters earlier reported, citing recent earnings outlooks from ASML and TSMC, that U.S. Big Tech companies are still expanding investment in AI infrastructure.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.