Morgan Stanley Flags CPU Demand Boost From Agentic AI

According to Reuters, Morgan Stanley said in a research note that increasingly autonomous, or agentic, AI systems could add $32.5-60 billion to the data-center CPU market by 2030, within a server CPU market Morgan Stanley expects to exceed $100 billion. Reuters reports the note frames the next wave of agentic AI as more driven by coordination and multistep task management than by raw parallel compute, which shifts part of the compute bottleneck toward CPUs and memory even as demand for GPUs remains strong. According to Reuters, Morgan Stanley identifies Nvidia, AMD, Intel, Arm, Micron, Samsung, SK hynix, TSMC, and ASML among potential beneficiaries.

Industry-pattern observations: Agentic AI workloads, by definition, emphasize planning, control flows, and state management across multiple steps. In comparable architectures, that increases reliance on general-purpose cores and larger memory footprints to orchestrate models, run coordination logic, and host stateful services. For practitioners, this pattern typically increases demand not just for accelerator throughput but for lower-latency orchestration layers, larger DRAM pools, and coherent CPU-GPU interconnects. That pattern also elevates the importance of system-level bottlenecks such as I/O, memory bandwidth, and kernel-level scheduling.

Editorial analysis: Brokerage estimates of incremental total addressable market in the tens of billions, as reported by Reuters, matter because they reframe where hyperscalers and cloud providers may prioritize capital spending. When public research notes or sell-side forecasts call out memory and CPU as growth drivers, procurement and procurement-adjacent teams at cloud providers and enterprises often re-evaluate capacity planning and supplier roadmaps. For hardware and infrastructure vendors, the mix shift described in the note implies potential margin and pricing dynamics for supply-constrained components, which Morgan Stanley told Reuters could translate into pricing power for constrained suppliers.

For practitioners: observable indicators to monitor include announcements of new server SKUs optimized for mixed CPU-accelerator orchestration, memory roadmap updates from Micron, Samsung, and SK hynix, and capacity guidance from foundry and equipment providers such as TSMC and ASML. Also watch cloud providers' instance-type introductions and public benchmark workloads that emphasize multistep agentic pipelines; those will show whether procurement and deployment trends align with the brokerage thesis. Separately, public analyst coverage and equity-market signals, such as the Insider Monkey summary citing expected 39% five-year EPS growth for Nvidia, reflect investor expectations but are distinct from the infrastructure demand analysis reported by Morgan Stanley.

What happened above reports Morgan Stanley's note as covered by Reuters and related outlets. The research note's underlying modeling assumptions are not fully public in the coverage; interested readers should consult the original Morgan Stanley note for line-item assumptions. Editorial analysis sections above synthesize industry patterns and do not attribute internal motives or future actions to Morgan Stanley or any vendor.