AI’s Shift From Thinking to Taking Action

Summary

Shawn Kim explains the shift from generative AI to agentic AI, where AI moves from passive responses to active, multi-step workflows. This shift changes computing bottlenecks, making CPUs and memory more critical, and creates investment opportunities in the AI supply chain. He estimates $60B of incremental CPU TAM by 2030 and 70% of DRAM bit shipments tied to this theme, favoring memory, foundry, and related components.

• Agentic AI moves from passive chatbots to active, multi-step workflows.
• Computing bottlenecks shift from GPUs to CPUs and memory.
• CPUs are identified as a new bottleneck with significant TAM growth.
• Memory is seen as the highest content increase area, especially DRAM.
• The supply chain including memory, foundry, substrate, and interfaces benefits.
• Incremental CPU TAM estimated at $60B by 2030.
• Up to 70% of DRAM bit shipments linked to agentic AI.
• Supply chain areas have pricing power and capacity constraints into 2027.