AI Chips: Why CPU Stocks Are Suddenly in Focus

For the last couple of years, graphics processing units (GPUs) have been the main area of focus when it comes to AI chips. This is because the training of AI models requires massive parallel processing capabilities. However, today, the semiconductor market is experiencing a powerful shift. Suddenly, it’s central processing units (CPUs) that are in focus, as demand from hyperscalers and AI labs is surging.

So, what’s behind this shift in the AI chip space? And which companies are leading the CPU charge?

Agentic AI Depends on CPUs

The sudden surge in demand for CPUs is being driven by the rise of agentic AI. While training massive Large Language Models (LLMs) is a GPU-heavy task, running agentic AI applications is extremely CPU-intensive.

Agentic AI systems don't just generate text; they act as autonomous agents that orchestrate complex tasks. They search databases, write code, integrate with APIs, and execute applications.

And this kind of activity requires CPUs for coordination. Without CPUs acting as “traffic controllers”, AI agents get stuck waiting on decision loops and external data inputs, leaving even the fastest GPUs idling while waiting for their next instructions.

Ultimately, we have graduated from the GPU-only phase of the AI boom. As AI models transition from simple chat assistants to active agents, frontier AI labs and hyperscalers are scrambling for heavy-duty server CPUs to power their workloads.

The Shifting GPU to CPU Ratio

It’s worth noting that the ratio of CPUs to GPUs in data centers is tightening dramatically as agentic AI takes off. Whereas the ratio may have been 1:4 or 1:8 in the AI chatbot era, we are now moving towards a ratio of 1:1¹.

In some high-agent workloads, the ratio is even tilting in favor of CPUs. This shift in the ratio of CPUs to GPUs in data centers is one of the most significant architectural changes in the history of modern computing – we are witnessing a rapid transition from a GPU-dominated landscape to a balanced, near-parity layout.

AMD: A Leader in AI CPUs

One company at the heart of this CPU boom is Advanced Micro Devices (AMD). Right now, it is seeing high demand for its EPYC CPUs, which are increasingly being utilized as the critical "orchestration brains" inside massive AI clusters. Recently, AWS, Google Cloud, Microsoft Azure, and Tencent have all announced new and expanded 5th Gen EPYC-powered cloud instances. Looking ahead, Meta will be a lead customer for the upcoming 6th Gen AMD EPYC CPUs – codenamed “Venice” and “Verano” – due for launch later this year.

Given the high demand for its AI products, AMD has experienced strong growth recently. In the first quarter of 2026², Data Center segment revenue amounted to $5.8 billion, up 57% year-over-year. Overall, total revenue was $10.3 billion, up 38%. Note that since the company’s Q1 earnings, Wall Street analysts have been aggressively raising their price targets for the chip stock.

Intel: Working Closely with Google

Another major player in the CPU space is Intel. It is currently having success with its new Xeon 6 offering, which is designed for servers, data centers, supercomputers, and high-performance workstations. Recently, Intel and Google announced a multi-year collaboration for continued deployment of Xeon processors across Google’s workload-optimized instances. Xeon 6 was also selected as the host CPU for Nvidia’s DGX Rubin NVL8 systems, reinforcing Intel’s continued role at the center of leading AI infrastructure deployments.

Given the rising demand for CPUs, Intel is seeing a rapid improvement in operational performance at present. In the first quarter of 2026³, revenue was $13.6 billion, up 7% year-over-year. Looking ahead, the company expects revenue of $13.8 billion to $14.8 billion for Q2. At the midpoint, that represents growth of 11% on last year’s Q2 revenue of $12.9 billion.

ARM Holdings: Focusing on Agentic AI

Arm Holdings has had a lot of success in the smartphone CPU space in the past. Today, its CPU technology can be found in over 99% of the world’s smartphones⁴. Recently, however, the company has been developing products for the agentic AI era. For example, in March, it introduced the Arm AGI CPU, a product that delivers more than 2x performance per rack compared with x86-based platforms and will be produced by Arm itself (Arm has traditionally been an IP licensing company – licensing its blueprints to other chip companies such as Nvidia, Qualcomm, and Apple).

Zooming in on Arm’s financials, the company recently delivered record-breaking Q4 and full-year results⁵, with Q4 revenue reaching $1.49 billion (+20% year-over-year) and full-year revenue hitting $4.92 billion (+23% year-over-year). In its Q4 earnings, it said that the customer response to the Arm AGI CPU has been strong, with more than $2 billion of orders across FY2027 and FY2028. This is more than double the amount stated at the Arm Everywhere conference in March 2026 when the product was introduced.

Amazon: A Growing Force in AI Chips

While Amazon is mainly seen as an e-commerce and cloud computing business, it has a rapidly growing chip segment today. And within this segment, it makes Graviton CPUs, which are used within its cloud computing platform, Amazon Web Services (AWS). Its latest product here is the Graviton5, which is built on an ultra-advanced 3nm manufacturing process and features a massive 192 CPU cores on a single chip, offering a 5x larger cache than its predecessor, drastically reducing data latency⁶. Note that unlike traditional server CPUs from Intel and AMD, which use the x86 architecture, Graviton CPUs are built on Arm architecture to deliver greater energy efficiency.

While Amazon doesn’t provide a breakdown of its chip sales in its quarterly earnings reports, the company recently advised in its annual letter for 2025 that its chips business is now operating at a $20 billion annual revenue run rate⁷. Notably, CEO Andy Jassy said that the chip segment is growing at a triple-digit percentage rate year-over year. Jassy also said that if Amazon’s chips business was a standalone business, and sold chips to other companies, the annual revenue run rate would be closer to $50 billion. In other words, Amazon could be one of the largest data center chip businesses globally.

CPU Stocks in the Themes Generative AI ETF

Now, this is just a few of the companies that develop CPUs; others include Qualcomm, Nvidia, Samsung, Apple, IBM, and MediaTek. So overall, there are a number of ways to gain exposure to the CPU boom.

It’s worth pointing out that there are several CPU stocks in the Themes Generative AI ETF at present. This ETF aims to track the Solactive Generative Artificial Intelligence Index, which identifies 40 companies that derive their revenues from artificial intelligence, data analytics & big data, natural language processing, and artificial intelligence-driven services.

Footnotes:

¹AMD, Agentic AI Changes the CPU/GPU Equation, as of May 7, 2026

²AMD, AMD Reports First Quarter 2026 Financial Results, as of May 5, 2026

³Intel, Intel Reports First-Quarter 2026 Financial Results, as of April 23, 2026

⁴arm, Experience the Future of AI-Powered Mobile Computing, as of May 13, 2026

⁵arm, Arm delivers record-breaking quarter and full-year results, as of May 6, 2026

⁶Amazon News, AWS introduces Graviton5: the company’s most powerful and efficient CPU, as of December 4, 2025

⁷Amazon News, CEO Andy Jassy’s 2025 Letter to Shareholders, as of April 9, 2026