Both AMD and Intel said during the 2026 Morgan Stanley Technology, Media & Telecom Conference that demand for CPUs is seeing an uptick due to artificial intelligence. Intel CFO David Zinsner said during his question and answer (via Investing.com) that “the CPU has become cool again this year,” especially as AI agents need CPUs to orchestrate the computationally-heavy tasks that the GPUs and NPUs will execute. It has even started seeing customers who are looking at long-term agreements, ensuring that they’ll have a continuous supply of these chips needed to expand their operations. On the other hand, AMD CEO Lisa Su said during the same conference, “You know, we’re seeing a significant CPU demand, frankly, as a result of the inference demand picking up.” She also added later that “the CPU portion of the business has actually far exceeded my expectations in terms of demand.”
We felt the full swing of this crisis in the fourth quarter of last year, with pricing for RAM modules and SSDs continuing to rise through February 2026. What makes this worse than the GPU shortage is that it has a much wider impact. While the GPUs that were in short supply were mostly limited to desktop PCs and gaming laptops with a discrete graphics card, virtually every modern digital device — from consumer devices like smart TVs and smartphones to automobiles and industrial-grade equipment — needs memory and storage. And consumer-grade memory and storage is fighting for the same wafer space that enterprise-grade memory and storage could occupy, usually with a far higher price tag.
As AI advancements move forward from large language models and chatbots to agents that can observe, reason, plan, act, and learn independently, data centers require more multi-processor computing power — that means combining CPUs, GPUs, NPUs, and more — to support the entire agentic AI workflow. China is starting to see this spike in demand, with both Team Blue and Team Red reporting supply shortages for server CPUs in the region. We’re also seeing a spike in demand for high-end Mac Studios and Mac minis, especially as we see more people build their own local AI agents with the rise in popularity of the open-source Clawdbot/Moltbot/OpenClaw.
AMD and Intel are presumably talking about data center demand for their CPUs; consumer systems aren't equipped to handle the massive memory demands of agentic AI. If there is a shortage, however, it could trickle down to the consumer market, assuming the demand keeps pace.
Over the past several generations, AMD and Intel have converged their data center and consumer offerings, allowing them to maximize yields by leveraging the same microarchitecture across both client and enterprise. Some of that silicon won't be useful in the data center, so the consumer market won't evaporate. But it could put downward pressure on supply if the focus shifts toward the data center, as we've seen with RAM and SSDs.
Unlike Nvidia, which has seen exponential increases in revenue in its data center business, both AMD and Intel still see about half of their total revenue each quarter from the consumer market. It's still an important market, so although demand from data centers may increase, it shouldn't come at the cost of the consumer market, at least entirely.
Hopefully, both Intel and AMD can keep up with the future demand, so as not to exacerbate the worsening situation of the computer industry. Otherwise, some are already predicting the end of the entry-level PC by 2028 if things continue as they are.
Follow 3DTested on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.
