Advanced Micro Devices vs. Arm Holdings: Which AI CPU Stock Is the Better Buy?

Artificial intelligence (AI) infrastructure has been dominated by graphics processing units (GPUs). But the next phase of AI adoption, especially inference (the deployment of AI models in production environments) and agentic AI, is also driving increased demand for central processing units (CPUs).

In large AI systems, CPUs help coordinate data movement, networking, and orchestration across multiple AI chips. That makes Advanced Micro Devices (NASDAQ: AMD) and Arm Holdings (NASDAQ: ARM) two very different ways to invest in the AI CPU opportunity.

Let's assess which AI CPU stock is a better buy now.

Image source: Getty Images.

AMD's CPU strategy

AMD is increasingly valued as a data center infrastructure company, not just a PC or gaming chip player. In the first quarter , AMD's revenue rose 38% year over year to $10.3 billion. The company's data center segment revenue jumped 57% to $5.8 billion, driven by strong demand for EPYC server CPUs and the continued ramp of Instinct AI GPUs.

The company now expects the server CPU total addressable market (TAM) to grow at more than 35% annually and exceed $120 billion by 2030. This is a significant upward revision from the expected 18% annual growth for the next three to five years. Management also expects server CPU revenue to grow by more than 70% year over year in the second quarter.

AMD is seeing robust CPU demand from regular server computing, head nodes that help manage GPUs and other AI accelerators, and agentic AI workloads. As agentic AI workloads grow, each AI agent can create more CPU tasks for orchestration, data processing, and parallel execution. In older systems, one CPU often supported four or eight GPUs.

But as inference and agentic AI workloads grow, AMD believes some systems may require one CPU for every GPU. In highly agentic workloads, there could even be more CPUs than GPUs. Hence, EPYC CPUs are proving to be a direct beneficiary of the agentic AI build-out.

AMD's 6th Gen EPYC processor, also called Venice, has begun ramping on Taiwan Semiconductor Manufacturing's advanced 2-nanometer process. Advanced manufacturing nodes can help chips deliver more computing performance while using less power per task. Since power and cooling are becoming critical bottlenecks in AI data centers, Venice may see solid demand in the coming years.

The Venice family of CPUs includes CPUs built for different needs, such as higher throughput, better power efficiency, and better performance. It also includes Verano, AMD's first EPYC CPU designed specifically for AI infrastructure. AI data centers will use different CPUs for general purpose computing, for supporting GPUs, and for agentic AI workloads.

AMD is well positioned to target these opportunities with a wider range of CPUs optimized for each use case. But AMD is not cheap. The stock trades at nearly 75 times forward earnings, leaving very little room for execution missteps.

Arm's CPU strategy

Arm Holdings licenses CPU architecture and chip designs to companies that want to build power-efficient processors. Already a dominant presence in the smartphone market, Arm is now gaining traction in AI data centers as well. The company's fiscal 2026 (ending March 31, 2026) performance was also impressive. Revenue was up 23% year over year to $4.92 billion, comprising royalty revenue of $2.61 billion and licensing revenue of $2.31 billion.

Arm can benefit from licensing and royalty revenue as Amazon, Alphabet, Microsoft, Nvidia, and other companies use Arm-based CPUs in cloud and AI infrastructure. The company's Arm AGI CPU also gives it a more direct way to sell into AI data centers. The company's CPU compute share among top hyperscalers is now about 50%, helped by chips such as AWS Graviton, Google Axion, Microsoft Cobalt, and Nvidia Vera.

Developed with Meta Platforms as the lead partner, Arm's AGI CPU is designed for agentic AI data centers. Arm claims that the chip offers better performance at lower capital costs than x86-based platforms.

Customer demand for the AGI CPU across fiscal 2027 and fiscal 2028 had exceeded $2 billion (as of May 6, 2026), more than double the amount discussed at its launch event in late March 2026. Arm says it has backing from more than 50 companies as it expands its compute platform from intellectual property and chip designs into finished chips.

However, Arm's new strategy also adds new risks. The company's traditional licensing model is asset-light and high-margin, but selling its own chips introduces supply chain risk, execution risk, and potential tension with partners that also build Arm-based chips.

Arm shares are also trading at a very rich valuation of nearly 179 times forward earnings. Hence, while Arm is a high-quality AI CPU platform player, investors are already paying heavily for that quality.

Better buy

Both companies offer exposure to the AI CPU opportunity but in different ways. AMD is better suited for investors who want direct AI infrastructure exposure, since EPYC CPUs are already benefiting from rising demand in cloud, enterprise, and agentic AI, while also supporting AMD's broader GPU portfolio.

Arm is better suited for investors willing to pay a premium for a longer-term platform story where licensing, royalties, and the new AGI CPU could expand its role across hyperscaler AI data centers. Hence, AMD looks more attractive for investors focused on near-term AI CPU-powered revenue visibility and lower execution risk.

Manali Pradhan, CFA has no position in any of the stocks mentioned. The Motley Fool has positions in and recommends Advanced Micro Devices, Alphabet, Amazon, Meta Platforms, Microsoft, Nvidia, and Taiwan Semiconductor Manufacturing. The Motley Fool recommends Arm Holdings. The Motley Fool has a disclosure policy.