Log In Get Started
Stock Markets April 20, 2026 04:36 AM

Morgan Stanley Says Agentic AI Will Push CPU and Memory Demand Beyond GPUs

Broker projects significant CPU market upside as AI systems evolve to plan and act autonomously, broadening spending across chipmaking and memory supply chains

By Nina Shah NVDA AMD INTC ARM MU
Morgan Stanley Says Agentic AI Will Push CPU and Memory Demand Beyond GPUs
NVDA AMD INTC ARM MU

Morgan Stanley argues that as artificial intelligence shifts from simple prompt-response models toward agentic systems capable of planning and acting independently, the primary computing bottleneck will move toward CPUs and memory. The bank forecasts a sizeable increase in data-center CPU spending through 2030 and identifies a range of chipmakers, memory suppliers and equipment vendors that could benefit.

Key Points

  • Agentic AI - systems capable of planning and taking actions autonomously - is shifting compute bottlenecks from GPUs toward CPUs and memory.
  • Morgan Stanley estimates agentic AI could add $32.5 billion to $60 billion to a data-center CPU market already forecast to exceed $100 billion by 2030.
  • Broader AI spending would extend beyond GPU vendors to benefit CPU and accelerator designers, memory suppliers, and chipmaking equipment manufacturers.

Morgan Stanley said on Sunday that the evolution of artificial intelligence toward more autonomous, agentic systems could materially increase demand for central processing units (CPUs), reshape how data centers are built, and expand investment beyond the graphics processing units (GPUs) that have led the recent AI spending surge.

In a research note, the brokerage wrote: "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," while noting that GPU demand remains strong.

The firm estimates that agentic AI - defined in the note as systems that can plan tasks and take actions on their own rather than merely responding to prompts - could add between $32.5 billion and $60 billion to a data-center CPU market that it projects will already exceed $100 billion by 2030.

Morgan Stanley said the next wave of agentic AI will be driven more by coordination than by sheer raw computing power. In this model, CPUs increasingly serve as the control layer for AI systems that orchestrate and execute multistep tasks, rather than GPUs alone handling the heavy lifting of model training and inference.

Alongside CPUs, the research note flagged a sharp rise in memory requirements as agentic workloads proliferate, which would broaden AI-related spending to include a wider set of chipmakers, memory suppliers and manufacturing partners. The brokerage added that firms operating in supply-constrained parts of the ecosystem could see improved pricing power as demand intensifies.

Morgan Stanley listed a set of companies it views as potential beneficiaries if these dynamics play out: Nvidia, AMD, Intel and Arm in CPUs and accelerators; Micron, Samsung and SK hynix in memory; and TSMC and ASML across chipmaking and equipment.

Promotional product note included in the source material:

Should you invest $2,000 in INTC right now? ProPicks AI evaluates INTC alongside thousands of other companies every month using 100+ financial metrics. Using powerful AI to generate exciting stock ideas, it looks beyond popularity to assess fundamentals, momentum, and valuation. The AI has no bias - it simply identifies which stocks offer the best risk-reward based on current data with notable past winners that include Super Micro Computer (+185%) and AppLovin (+157%). Want to know if INTC is currently featured in any ProPicks AI strategies, or if there are better opportunities in the same space?

Risks

  • Concentration and supply constraints in parts of the supply chain could limit the ability of some buyers to secure capacity and may transfer pricing power to suppliers - impacting hardware procurement and capital planning in data-center and cloud operators.
  • The scale and timing of the projected CPU and memory demand depend on adoption of agentic AI; if adoption is slower or architectures evolve differently, expected market expansions may not materialize - affecting chipmakers and memory suppliers.
  • Shifts in spending across compute layers introduce execution risk for firms that must adapt product road maps and manufacturing capacity; companies unable to reallocate resources or scale production could miss potential gains.

More from Stock Markets

Kepler Cheuvreux Lowers View on U.S. Stocks, Lifts Emerging Markets as Better Opportunity Apr 20, 2026 U.S. Futures Slip After Record Rally as Iran Tensions Resurface Apr 20, 2026 Barclays lifts targets for European energy services, adopts P/E valuation and reshuffles ratings Apr 20, 2026 QXO to Buy TopBuild in $17 Billion Transaction, Becomes No. 2 U.S. Building Products Distributor Apr 20, 2026 American Airlines Shares Fall After Carrier Says It Did Not Seek Merger With United Apr 20, 2026