The AI server CPU market is on track to reach $201 billion by 2030, growing at a 44% compound annual rate, according to Raymond James’ latest framework. The forecast splits the total into $33.5 billion in conventional datacenter CPUs, $83 billion in AI head-end CPUs, and $85 billion in agentic CPUs, reflecting demand from persistent AI agents that orchestrate tool execution and retrieval workloads.
Raymond James’ base-case estimate aligns closely with Nvidia’s long-term framework of $200 billion, though it sits below AMD’s $220 billion projection under scenarios involving higher agent adoption and concurrency. The firm upgraded Advanced Micro Devices to Strong Buy from Outperform, citing the company’s direct earnings leverage, expanding datacenter footprint, and accelerating market-share gains. Raymond James also extended its financial model for AMD through 2028, following the chipmaker’s second-quarter 2026 Data Center revenue growth of 107% year-over-year, which surpassed expectations.
BMO Capital initiated coverage of AMD with an Outperform rating, while Raymond James adjusted its estimates for Nvidia to account for evolving server CPU dynamics. Nvidia’s CPU business remains a smaller segment relative to its core accelerator operations, though the company plays a key role in coordinating AI workloads across servers. BMO Capital anticipates Nvidia will report second-quarter revenues well above consensus, driven by data center strength, while KeyBanc expects strong results from ramping shipments of the Rubin R200 GPU. The Rubin R200 is positioned as a high-performance accelerator for dense model computation, complementing CPU-driven orchestration and tool workloads.
