AI agent expansion drives demand for CPUs and enterprise SSDs

The rise of personal AI agents, specifically Meta’s Muse, is driving a shift in semiconductor demand from GPUs toward CPUs and enterprise SSDs (eSSD). Unlike traditional chatbots, AI agents like Muse utilize long-running cloud virtual machines to perform complex, multi-step tasks such as travel booking and email management. This architecture requires significant CPU resources for workload orchestration and increased storage for maintaining task states and browser caches.

This shift is impacting the global semiconductor landscape and the Taiwanese supply chain. AMD has committed over $10 billion to the Taiwanese ecosystem to expand advanced packaging and server manufacturing. Meanwhile, Arm is collaborating with Meta on data center CPU development using TSMC’s 3nm process. Intel also remains a key player, though its specific Taiwanese partner list is less publicly detailed than its competitors.

Market analysts suggest that the transition to agentic AI could lead to structural increases in CPU usage and potential supply bottlenecks. While GPUs remain essential for parallel matrix operations, CPUs are becoming critical for the sequential reasoning and tool control required by autonomous agents. Consequently, specialized ETFs focusing on CPU manufacturers have seen significant recent gains.