Taipei — TrendForce Links Meta's Muse to Rising CPU Demand in AI

Meta Platforms Inc. has built an AI agent system called Muse that draws more heavily on central processing unit (CPU) capacity relative to graphics processing unit (GPU) capacity than earlier generative-AI systems, according to an analysis from Taiwan-based research firm TrendForce. TrendForce's note said the pattern could benefit CPU and chip-architecture suppliers Intel Corp., Advanced Micro Devices Inc., and Arm Holdings, the report indicated.

What Is Meta's Muse System?

TrendForce's analysis centers on Muse, an AI agent product attributed to Meta. The material reviewed for this report did not include additional detail on Muse's specific consumer or enterprise function, so no further description can be confirmed beyond its identification as an AI agent system in the TrendForce note.

Why Would AI Agents Need More CPU Capacity?

A CPU handles sequential logic, memory management, and task orchestration. A GPU accelerates the parallel matrix math used to train and run large language models. TrendForce's report ties Muse to a higher CPU-to-GPU ratio because agent-based AI systems — which chain together multi-step decisions and call external tools rather than simply generating a single text response — lean more on the orchestration work CPUs perform, according to the firm's analysis.

Which Chipmakers Could Benefit?

TrendForce named Intel, AMD, and Arm as potential beneficiaries of the shift toward higher CPU-to-GPU ratios in agentic AI workloads. The report did not specify projected order volumes, revenue figures, or timing for when that demand might show up in company results.

How Has AI Infrastructure Spending Shifted?

Much of the AI buildout since large language models entered wide commercial use has centered on GPU purchases, with Nvidia Corp. as the dominant supplier for training clusters. TrendForce's note on Muse suggests that as AI products move from single-response generation toward agent-style systems that execute multi-step tasks, some of that spending could redistribute toward CPU and system-level suppliers — though the report frames this as a directional signal tied to Muse specifically rather than a broad, quantified market call.

What to Watch Next

• Whether Meta or TrendForce publishes additional technical detail on Muse's compute architecture and CPU-to-GPU ratio.
• Whether Intel, AMD, or Arm address agentic-AI demand in upcoming earnings calls or investor materials.
• Whether other large AI developers roll out agent products with similar CPU-heavy compute profiles, which would test whether TrendForce's observation about Muse reflects a broader industry pattern or a single-product case.

None of the companies named — Meta, Intel, AMD, or Arm — has issued a public statement responding to the TrendForce analysis in the material reviewed for this report.