Got $10,000? 3 Stocks to Own for the Rise of Agentic AI
The article argues that “agentic AI” will shift data-center demand toward more CPUs, changing the GPU-to-CPU mix from about 8:1 for LLM training to 1:1 for agentic workloads, with higher core-count CPUs. It cites Nvidia’s view that the data-center CPU market could reach $200B. It highlights AMD (Venice mass production; 256 cores), Arm (projected $25B revenue by 2031), and Palantir (AIP for structured data/ontology).
Bullish read-through to higher data-center CPU volumes and pricing as agentic AI shifts GPU:CPU ratios toward 1:1.
Article argues AMD’s Venice data-center CPU is entering mass production and should gain share as agentic AI raises CPU demand.
Near-term: modest sentiment support; longer-term: gradual re-rating if CPU share gains persist.
Background
The article frames agentic AI as shifting compute needs from training-focused GPU-heavy workloads toward more CPU-centric, tool-using, sequential-reasoning systems, and argues software needs structured data to reduce costly hallucinations.
Why it matters
For AMD and Arm, the trade is a read-through from agentic AI compute intensity to data-center CPU demand and potential share/revenue capture. For Palantir, the trade is a read-through from agentic AI orchestration requirements to adoption of its AIP platform.
Market relevance
This is a bullish sector/theme framing for agentic AI beneficiaries (CPU and orchestration software), with limited incremental, event-driven information.
Market effects
Reinforces a read-across that agentic AI increases CPU intensity and demand for data-center compute and structured-data software layers.
Mentions TSMC’s 2nm process for AMD Venice, implying ongoing Taiwan/advanced-node supply-chain relevance.
Frames agentic AI as a global capex/software spend cycle driver, potentially affecting worldwide data-center hardware and enterprise AI platforms.
Alternative perspectives
Agentic AI may not translate into immediate CPU share gains or sustained software monetization; the piece is largely a bullish thesis without new company-specific numbers.
Key risks include execution on new CPU ramps (yields, performance-per-watt), competitive dynamics with Intel/other accelerators, and whether customers actually buy orchestration platforms like AIP at scale.
Key entities
- public_companyAdvanced Micro Devices
Discussed as a data-center CPU leader benefiting from agentic AI; Venice CPU mass production is highlighted.
- public_companyArm Holdings
Discussed as moving into designing its own data-center CPUs; projections tied to agentic AI market growth are cited.
- public_companyPalantir Technologies
Discussed as an AI orchestration software winner via AIP’s ontology/structured-data approach for agentic AI.
