
Bank of America raised its forecast for the 2030 server CPU market to more than $210 billion from roughly $170 billion, predicting 36% annual growth as agentic AI workloads push data centers to use CPUs and GPUs in closer-to-equal ratios.
The chipmaker AMD, named Bank of America's top CPU pick, could benefit significantly from this shift, as CPUs become additive to—rather than displaced by—broader AI infrastructure spending.
What happened
Bank of America raised its 2030 server CPU market estimate to more than $210 billion from about $170 billion, expecting 36% annual growth from roughly $35 billion in 2025. The shift stems from agentic AI workloads, which require CPUs to handle more orchestration and control tasks alongside GPUs.
Why it matters
During early AI phases, data centers used one CPU for every four GPUs; agentic AI could push that ratio toward 1:1, making CPUs additive rather than displaced by GPU growth. Bank of America named AMD its top CPU pick, citing strength in frequency and core count, suggesting AMD stands to capture significant share of this expanded market.
What to watch
The ratio shift from 1:4 (CPU:GPU) to closer to 1:1 will determine whether CPU demand truly expands as agentic AI deployments scale in data centers through 2030.
Advanced Micro Devices Inc. (AMD), which competes across CPUs and GPUs, stands to benefit from a significant expansion in the server processor market driven by agentic artificial intelligence. Bank of America has raised its 2030 server CPU market estimate to more than $210 billion from about $170 billion, reflecting expectations that the market will grow at a 36% annual rate from roughly $35 billion in 2025.
The key driver of this forecast revision is a fundamental shift in how AI workloads are evolving inside data centers. During the training-heavy phase of the AI boom, data centers operated with a ratio of roughly one CPU for every four GPUs. Bank of America expects agentic AI to push that ratio much closer to 1:1, as CPUs take on substantially more orchestration and control tasks. This shift is significant because it positions CPUs as additive to the broader AI infrastructure opportunity rather than simply being displaced by GPUs, as some earlier market analyses had suggested.
Bank of America named AMD its top CPU pick, citing the company's strength in both frequency and core count—attributes that support the orchestration demands of agentic workloads. While Nvidia remains the firm's overall semiconductor favorite and both Intel and Arm also stand to benefit from expanded CPU demand, AMD's technical positioning makes it a focal point for the anticipated market expansion. For investors, the broader takeaway is that the next phase of AI spending may significantly broaden well beyond GPUs, opening a larger addressable market for CPU manufacturers than the training-phase dynamics would have suggested.
The AI infrastructure market has historically been dominated by GPU demand during the training phase of large language models and generative AI systems. Bank of America's revised forecast reflects a recognition that the next wave of AI—characterized by agentic systems that autonomously plan and execute tasks—will require a fundamentally different balance of compute resources. Rather than data centers simply adding more GPUs, agentic AI necessitates stronger CPU performance to coordinate, orchestrate, and control these autonomous systems. This structural shift transforms CPUs from a subordinate component in the AI stack into a co-equal infrastructure requirement.
AMD's positioning as Bank of America's top CPU pick underscores the competitive dynamics at play. While the bank notes that Nvidia remains its overall semiconductor favorite and that Intel and Arm also stand to benefit, AMD's strength in both frequency and core count—key attributes for orchestration-heavy workloads—makes it well-positioned to capture share of the expanded CPU market. For investors, this signals that AI spending intensity may shift away from a GPU-centric model; the next phase of AI capex will broaden beyond accelerators to encompass a larger and more profitable CPU market.
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytime
Ask AI anything about this article. Q&As are published on this page for other readers too.
Sam Altman is demanding OpenAI's IPO valuation reach $1 trillion, but SoftBank must repay a $40 billion bridge…

DeepSeek moved its V4-Pro flagship model to production (build V4-Pro-0813), released its proprietary agent har…

Google released Gemini 3.7 Flash, its successor to Gemini 3.6 Flash (released three weeks earlier), available…

Microsoft has begun integrating its consumer and enterprise Copilot applications into a unified platform, star…

OpenAI and Cerebras announced Ultrafast Mode, a new service tier delivering up to 750 output tokens per second…

Google shut down AlphaFold, DeepMind's AI protein-structure prediction tool, two weeks before removing its co-…

The AI news that matters, in one minute each morning.
Sign up free