As the artificial intelligence (AI) industry accelerates its evolution from chatbots to agentic AI, Wall Street is reassessing the long-term growth potential of the server CPU market. A prominent investment bank, Bernstein, has released a new research report stating that the agentic AI era will significantly boost data center demand for CPUs. Consequently, the firm has substantially raised its forecast for the global server CPU total addressable market (TAM) in 2030 from $137 billion to $223 billion. Bernstein has also increased its price targets for Advanced Micro Devices (AMD.US), ARM Holdings (ARM.US), and Intel (INTC.US).
Specifically, Bernstein raised its target for Arm to $500 from $300, maintaining an "Outperform" rating. The target for AMD was lifted to $600 from $525, also with an "Outperform" rating. For Intel, the firm kept its "Market-Perform" rating but raised the price target to $100 from $65. Meanwhile, the target price for SoftBank Group, which holds a controlling stake in Arm, was also increased to 11,200 yen, up from 8,200 yen.
The Shift from Chatbots to Agentic AI Elevates CPU Importance
Analysts at Bernstein, led by David Dai, note that the AI industry is transitioning from "Generative AI 1.0" to "Generative AI 2.0." If chatbots like ChatGPT represent AI 1.0, then agentic AI—which can autonomously plan tasks, utilize tools, and execute complex workflows—represents the direction of AI 2.0.
The analysts state that agentic AI requires substantial task coordination and execution, making it far more dependent on CPU resources than traditional chatbots. Bernstein points out, "Agentic AI involves highly autonomous task orchestration and execution, which will significantly increase CPU workloads compared to GPUs."
The report indicates that in traditional AI data centers, the CPU-to-GPU ratio is typically around 1:4 or even 1:8. However, with the rapid adoption of agentic AI, this ratio is expected to rise to 1:1 or higher. In other words, each new GPU added in the future may require more supporting CPU resources. This suggests that the CPU supply chain, which has been overshadowed by GPUs in recent years, could re-emerge as a key beneficiary of AI investment.
2030 Server CPU Market Size Could Be Six Times Larger Than 2025
Based on this outlook, Bernstein has significantly raised its forecast for the global server CPU TAM in 2030 to $223 billion. In comparison, the global server CPU market size in 2025 is projected to be just $37 billion. This implies the market could grow approximately sixfold over the next five years.
This forecast is based on core assumptions, including global AI data center capital expenditure reaching $3.5 trillion by 2030 and a CPU-to-GPU ratio of 1:1 for AI inference workloads. Analysts note that the previously projected $137 billion market size now represents a pessimistic scenario. That scenario assumes about $3 trillion in global AI infrastructure investment and a CPU-to-GPU ratio of 1:2. In an optimistic scenario, if global AI data center investment reaches $4 trillion and the CPU-to-GPU ratio rises to 1.5:1, the server CPU market could even reach $330 billion by 2030.
Arm Viewed as the Biggest Winner in the Agentic AI Era
Among all potential beneficiaries, Bernstein is most optimistic about Arm. Analysts believe the Arm architecture, with its superior energy efficiency, is naturally suited for data center demands in the agentic AI era. As AI systems scale, reducing energy consumption has become a critical consideration for data center operations, an area where Arm holds a distinct advantage.
Furthermore, Arm is gradually expanding from a pure IP licensing model into CPU product development, aiming to capture more value from the industry chain. Bernstein previously estimated Arm's related revenue could reach $15 billion by 2030. Following the upward revision of the server CPU market forecast, the firm has now raised this estimate to $22 billion. The report states, "Arm will be one of the most important structural beneficiaries of the CPU renaissance driven by agentic AI."
AMD and Intel Also Stand to Benefit, with AMD Holding an Edge
Bernstein also believes that AMD and Intel will benefit from the continued growth in data center server demand. However, from a product competitiveness perspective, AMD maintains a leading position. Analysts note that AMD's current server product portfolio still outperforms its competitors in both performance and energy efficiency, positioning it to continue gaining market share.
As Bernstein's previous earnings forecasts for AMD already factored in strong server market growth, the adjustments to its profit projections this time are relatively limited. In contrast, Intel's future earnings expectations saw a more significant upward revision, leading to a more pronounced increase in its price target.
Comments