Core Points

• The craze for personal intelligent agents has driven the stock prices of AMD and Intel higher. This year and over the past month, both companies have outperformed other major tech giants.
• As the leading manufacturer in the development of central processing units ( CPU ), this long-standing rival has found an important new position in the AI market dominated by NVIDIA.
• Signal65 President Ryan Schrute stated: "As more intelligent agents are developed and an increasing number of people use them to handle various tasks, the computational load will gradually shift from GPU to CPU."

For many years, NVIDIA and its GPU have dominated the AI market; now, with the rise of personal AI intelligent agents, new vitality has been injected into AMD and Intel.

The era of generative AI began with the launch of ChatGPT; NVIDIA GPU took center stage, while CPU was once the main computing unit of the servers. Now, CPU has made a comeback, driving the performance of leading companies such as AMD and Intel upwards. These chips are adept at powering intelligent agents that can operate autonomously for extended periods, capable of working independently for several hours in a row.

Last week, OpenAI followed closely on the heels of Meta by launching AI as a proxy for Dots. Meta released Muse as early as early September, and this product quickly became a hit, topping Apple's App Store download chart. Prior to this, driven by the surge in demand for CPU, the stock prices of AMD and Intel had already seen significant increases this year; the upward trend has further accelerated over the past month, with gains of 32% and 21% respectively, outpacing all major tech stocks.

AMD has grown in sync within the domains of CPU and GPU, officially entering the trillion-dollar club in market value.

AI Hardware and Cost Consulting Firm Signal65 President Ryan Schrute stated: "As more intelligent agents are developed and an increasing number of people use them to handle various tasks, the computational load will gradually shift from GPU to CPU."

Users can directly inquire with the intelligent agent to learn about the underlying computing power: Muse informs users that it runs on servers equipped with AMD chips; Dots indicates that it relies on a virtual computer driven by AMD Xuelong ( EPYC ) CPU.

A spokesperson for Meta stated that the company's system design supports various types of CPU. "Our hardware strategy is diversified, and we are not bound to any specific CPU at the design stage, which gives us maximum flexibility in scaling up our computing power," the company said in its statement.

OpenAI also uses multiple CPU suppliers, indicating that AMD is not the only option.

Super-large cloud providers such as Amazon, Google, Meta, and Microsoft mostly use servers equipped with Intel or AMD's CPU. However, almost all large cloud service providers are developing their own chips, which are based on the ARM architecture. In March of this year, ARM released its own CPU for intelligent agents, with Meta being the first customer, and since then, the stock price of ARM has more than doubled.

Earlier this year, NVIDIA launched a completely restructured central processing unit, Vera, along with its supporting software ecosystem. The chip giant stated that their development of Vera CPU was based on their optimism about the CPU market segment, and they anticipate that the CPU market will reach a scale of 200 billion US dollars by 2030.

However, analysts believe that in the CPU segment of the market, NVIDIA is not currently the preferred choice for consumers.

Sumitomo Securities analyst Jordan Klein said in an interview, "This product is just coming out and has to face competition from many rivals." He believes that AMD currently has the 'strongest product'."

CPU is an important engine for the growth of AMD. As of the financial report for the second quarter of 2026 in June, the revenue from the AMD data center business more than doubled year-on-year, reaching $6.7 billion, accounting for nearly 60% of the total sales.

Dan McNamara, Senior Vice President and General Manager of AI, stated: "Over the past 8 to 10 months, there has been a tremendous shift in the industry's perspective on intelligent agent applications." He added that CPU sales are about to see a "significant increase."

GPU is responsible for thinking, while CPU is responsible for execution.

Muse, Dots, and the code tool OpenAI rely on the AI model to plan tasks that can be executed autonomously. Daniel Newman, the CEO of Futurum Group, explained that such intelligent agents require an independent computing environment and are capable of running continuously in the background for hours or even days.

Inference (using the AI model) still relies on GPU, but intelligent agents also depend on CPU.

Newman said, " GPU is responsible for the model's thinking and calculations, while CPU executes the entire workflow."

Meta and OpenAI utilize virtual machine technology to divide a single server into multiple lightweight virtual computing instances, allowing a single server to serve a large number of users simultaneously.

Some AMD Thunder Dragon processors have up to 192 CPU cores. Benchmark tests and information disclosed by the intelligent agents themselves show that Muse virtual machines use 2 cores, while Dots uses 9 cores.

The price of CPU is lower than that of GPU.

Benchmark tests and query information show that Dots uses AMD Xeon 9V74 CPU, with dealer prices falling short of 3,000 US dollars. Meta Xeon 9D25, used for Muse, is even cheaper on the second-hand market. A single NVIDIA GPU can cost more than ten times that, and GPU usually requires hundreds or thousands of units to be deployed in clusters.

Sensor Tower data shows that since Muse went live last month, the download volume has exceeded 5 million. Investors are closely watching how Meta can support Muse users. Morgan Stanley estimates that the monthly service cost per user ranges from $3 to $130, with an average of $37, depending on the amount of reasoning usage.

Morgan Stanley believes that this intelligent agent, Meta, is expected to contribute 20% of AMD's chip sales in 2026.

CPU The track presents huge business opportunities, with industry growth forecasts continuously being raised. Futurum It is predicted that by 2027, CPU total sales will reach $118 billion, nearly twice the previous forecast made by the institution in May.

AMD predicted in July that by 2030, the overall CPU market size will reach $220 billion, higher than the original forecast of $60 billion for 2025; AMD is expected to capture more than half of the market share by then.

Mercury Research data shows that AMD currently holds approximately 46% of the market share in x86 CPU in terms of shipment volume. Intel has also seen a surge in demand.

AMD has deeply penetrated major hyperscale cloud providers, and a large number of intelligent proxy virtual machines are deployed on these cloud platforms. At the same time, AMD chips, which utilize the clear architectural advantages of x86, enable intelligent proxies to run outdated software directly without any compatibility issues.

Klein said, " AMD is very likely to be the biggest beneficiary of this round of dividends, as it has the highest market share among cloud vendors CPU. However, I wouldn't rule out NVIDIA." "Vendors that can supply both types of chips ( CPU + GPU) are in the most advantageous position."