CPU shortage boosts Intel and AMD as agentic AI demand grows

CPU demand tied to agentic AI is emerging as a separate bottleneck in the broader AI infrastructure cycle, lifting Intel and AMD as buyers race to secure server capacity. Intel CEO Lip-Bu Tan said some customer agent workloads require a CPU demand ratio that reaches “4-to-1” compared with GPUs, while AMD has reported record server CPU revenue for a fourth straight quarter.

Procurement is shifting from quarterly or short-term orders toward long-term agreements as hyperscalers and server makers try to lock in multi-year supply. Intel delivery lead times have stretched to as long as six months because of order backlogs, while AMD is seeing some products extend to 8-10 weeks. Bank of America projects the server CPU market will grow from $43 billion (~66.1 trillion won) in 2026 to $1.25 trillion (~1,922 trillion won) by 2030, with a compound annual growth rate of 30.6%. Citi forecasts the agentic CPU segment alone will reach $59.4 billion (~91.4 trillion won) by 2030, growing at a 185% annual rate.

Meta’s plan to rent or sell idle AI compute resources has raised fears of GPU oversupply and weighed on semiconductor stocks, including Micron, SanDisk and Intel. Analysts cited in the report argue those concerns should be separated from CPU shortages, where delivery delays, long-term agreements and agentic AI adoption continue to support demand.