Bank of America sees the server CPU market quintupling to more than $170 billion by 2030, and it's betting on AMD, ARM, and Intel to ride the wave. The thesis hinges on one word: agents.
BofA analyst Vivek Arya raised his 2030 server CPU total addressable market estimate to $170 billion-plus from a prior $125 billion, representing a near-fivefold expansion from the $35 billion market in 2025. The new forecast carries a 37% compound annual growth rate through the decade.
"We view the emergence of agentic AI as a powerful demand accelerant that expands the CPU opportunity and lifts both [Intel and AMD] and [Arm-based] challengers," Arya wrote in a note following discussions at the BofA Global Tech Conference. The logic is straightforward. GPUs train and run AI models, but when an AI agent acts on a prompt, scanning an inbox for a flight confirmation, cross-referencing a calendar, executing a booking, CPUs handle those orchestration tasks.
More agents means more CPU demand, and the industry is only beginning to deploy them at scale.
Arya responded by raising price targets across the board. AMD got a $560 target (up from $500) with a Buy rating, ARM was lifted to $335 from $245, and Intel received a rare double upgrade to Buy.
Nvidia remains BofA's top semiconductor sector pick.
AMD shares traded over 2% higher in pre-market trading Thursday, while Intel surged 5% and ARM gained 2%. The moves extend a blistering run: Intel has rallied 436% over the past 12 months, AMD has climbed 280%, and ARM has gained 182% year to date.
The CPU market's expansion is drawing in new contenders. Nvidia already sells its Grace CPU as part of the Grace Blackwell superchip and has begun offering CPU-based servers.
Qualcomm is reportedly debuting its own data center CPU later this month, entering a space where x86 dominance is no longer guaranteed.
AMD is also pushing x86 deeper into adjacent territory. At its Embedded Solutions Day in Seoul on June 10, the company unveiled its Ryzen AI Embedded P100 and X100 series, single-chip designs combining Zen 5 CPU cores, RDNA 3.5 graphics, and an XDNA 2 NPU delivering up to 50 TOPS of AI acceleration. The chips target automotive, robotics, and industrial automation, markets where Arm-based rivals Nvidia and Qualcomm currently hold the advantage.
The embedded push carries weight from AMD's $49 billion Xilinx acquisition in 2022, which brought four decades of FPGA expertise and more than 400 partnerships into the fold. Lee Hee-man, head of Korea sales for AMD's Adaptive and Embedded Computing Group, said the company has secured 7,000 embedded customers worldwide across automotive, healthcare, broadcast, and satellite systems.
Whether x86 can carve meaningful share from Arm in cars and robots is a question that will take years of design wins to answer. But BofA's 37% CAGR call on a $170 billion CPU market highlights that the server chip, long overshadowed by GPUs in the AI narrative, is getting a second act from agents that don't just think, but act.
