The artificial intelligence (AI) market is undergoing a fundamental shift, moving focus from Graphics Processing Units (GPUs) to Central Processing Units (CPUs). Once viewed as secondary to GPU power, CPUs are now central to the infrastructure required for advanced, autonomous AI applications known as “agentic AI.” This changing demand has led industry leaders to drastically revise market forecasts, anticipating a massive surge in CPU adoption across data centers.
The Rise of Agentic AI and the CPU Bottleneck
The demand surge is primarily driven by agentic workloads, which are structurally different from simpler tasks like chatbot queries. While chatbots respond to direct human prompts, agentic systems are far more complex, autonomously handling hundreds of concurrent tasks and reasoning through problems with limited direct human intervention.
According to research from Intel and Georgia Tech, tool-dominated agentic AI workloads are significantly hindered by CPUs, which can consume up to 88% of the end-to-end latency. The paper concludes that as GPU quality improves, the performance bottleneck is likely to shift toward the CPUs. To efficiently scale agentic AI, CPU orchestration capacity must match GPU reasoning capacity, requiring a substantial increase in the CPU-to-GPU ratio within AI clusters to keep token costs low.
The role of the CPU in this context is orchestration—directing API calls and coordinating tasks between numerous independent agents. While GPUs handle the core inference reasoning, CPUs are responsible for telling the GPUs where, when, and how to allocate their resources.
Market Forecasts and Supply Constraints
The rapid growth in CPU demand has caused multiple major companies to significantly raise their market projections. AMD, for instance, recently increased its server CPU market forecast, nearly doubling its expected Compound Annual Growth Rate (CAGR) to 35%, projecting the market will surpass $120 billion by 2030. Similarly, Arm announced in March that the total addressable market (TAM) for data center CPUs is projected to exceed $100 billion by fiscal year 2031, representing a more than 4X increase from its current $24 billion estimate.
Other financial institutions corroborate this steep growth trajectory. UBS forecasts the market will grow from $31 billion in 2025 to $170 billion in 2030, a 40.6% CAGR. Bank of America forecasts a TAM expansion from $43 billion in 2026 to $125 billion in 2030, representing a 30.6% CAGR. Citi projects the overall market will grow from $29.3 billion in 2025 to $132 billion in 2030, maintaining a 35% CAGR.
In the physical supply chain, shortages are already becoming apparent. Reuters reported in February that Intel possesses a substantial backlog of unfulfilled CPU orders, with delivery times stretching up to six months. For some AMD products, delivery delays range between eight and ten weeks. KeyBanc noted in January that both Intel and AMD were nearly sold out of CPU servers for 2026 and reported anticipated Average Selling Price (ASP) increases of 10% to 15%. Electronic equipment distributor Fusion Worldwide reported that Intel distributors were only fulfilling approximately 40% of their yearly backlog allocations, with lead times of 8 to 22 weeks domestically. The situation was further highlighted by the South Korean trade publication The Elec, which reported won-denominated price increases as high as 3X for some x86 CPUs.
Competitive Strategies of Industry Leaders
The intense demand has positioned AMD, Nvidia, Arm, and Intel to compete fiercely for market share. Each company is developing unique hardware solutions to capitalize on the architectural shift.
Nvidia’s Expansion Beyond GPUs
Nvidia is actively expanding into the CPU sector, notably with the standalone Vera rack. This rack, containing 256 CPUs and 22,528 total CPU cores, allows customers to deploy a high volume of CPUs without needing proportional increases in GPUs. Nvidia’s CEO, Jensen Huang, frames CPUs as additive to GPUs, arguing that more AI agents require increased orchestration (CPU demand) alongside increased inference (GPU demand).
Critically, Nvidia indicated that the Vera rack opens up a $200 billion CPU TAM for the company. Furthermore, Nvidia reported that it expects to generate nearly $20 billion in CPU revenue this year, largely derived from standalone Vera racks.
AMD and Arm: Targeting Market Dominance
AMD is aggressively pursuing a 50% market share in the server CPU market by 2030, a goal that would imply $60 billion in annual revenue. The company is leveraging its Venice family of EPYC CPUs, with plans to launch Verano, an EPYC CPU designed specifically for AI infrastructure, in 2027.
Arm is focusing heavily on power efficiency. Through its AGI CPU, co-developed with Meta, Arm asserts that its chip provides up to 2x greater performance per watt compared to x86 CPUs. Arm also promotes the standalone Vera rack, arguing that its 4X CPU core count growth estimate is likely conservative.
Intel’s Efforts to Maintain Leadership
Intel is positioning itself to remain a strong competitor by announcing plans to deploy rack-scale CPU systems. Intel’s blueprints feature designs supporting up to 128 of its Granite Rapids Xeon 6 or Clearwater Forest Xeon 6+ chips, providing either 16,384 or 36,864 cores, respectively. Intel’s Xeon 6+ chip boasts 288 cores per chip. However, the company faces challenges, including the expected delay of its next-generation Xeon 7 ‘Diamond Rapids’ CPU, which was moved from late 2026 to the middle of 2027.
While Intel utilizes its advanced 18A node to improve efficiency, its primary advantage is core density. Meanwhile, AMD’s Venice CPU is noted for offering up to 512 threads via multi-threading, potentially increasing efficiency over Intel’s 288 threads.
Conclusion: The Structural Shift
The consensus among industry analysts is that the AI market’s growth is no longer solely defined by GPU capacity. The move toward standalone CPU racks represents a major architectural shift, allowing for increased orchestration capacity without having to increase GPU counts proportionally. The overall market is expanding rapidly, making CPUs the next major bottleneck and a critical area of investment for the coming years.
