The global AI race is now creating pressure far beyond graphics chips, and Advanced Micro Devices says the CPU market has suddenly become much tighter than anyone expected. Speaking in Taipei on Friday, AMD CEO Lisa Su revealed that demand for processors has exploded so quickly that the company is now rushing to expand production capacity with manufacturing partners across Taiwan. According to Su, the industry underestimated how aggressively businesses would begin adopting AI-powered systems, especially newer forms of “agentic AI” capable of carrying out tasks autonomously.
Su explained that she recently met with some of AMD’s biggest customers in both China and global markets before traveling to Taiwan to focus directly on supply-chain expansion. Taiwan remains one of the most critical centers in the global semiconductor ecosystem, especially because companies like TSMC manufacture chips for nearly every major AI player including Nvidia, Apple and AMD itself. Su admitted that overall CPU demand has become “significantly higher” than forecasts made just one year ago, adding bluntly that the market is now extremely tight.
AI Is No Longer Just About GPUs Anymore
For the last couple of years, most attention inside the AI industry focused heavily on GPUs because those chips power the training of massive AI models. But now CPUs are suddenly becoming just as important because companies are deploying AI systems into real-world operations at much larger scale. Agentic AI, which can perform actions independently instead of simply generating responses, is reportedly pushing businesses to upgrade broader server infrastructure instead of relying only on GPU-heavy systems.
AMD says it is now increasing CPU production quarter by quarter throughout 2026 and already preparing for even larger expansion plans stretching into 2027 and beyond. The company also confirmed that it has begun ramping production of its new Venice CPUs using TSMC’s advanced 2-nanometer manufacturing process. That next-generation chip technology is expected to become a huge battleground as semiconductor companies compete to deliver faster and more power-efficient AI hardware.
AMD Doubles Down on Taiwan With Massive Investment Push
Alongside the production expansion, AMD also announced plans to invest more than $10 billion into Taiwan’s AI ecosystem. The investment reportedly focuses heavily on advanced packaging, manufacturing systems and infrastructure needed to support increasingly complex AI chips. Modern semiconductor designs now rely heavily on advanced packaging techniques where smaller chip components are connected together instead of building everything into one massive processor.
Lisa Su described this as one of AMD’s smartest long-term bets, especially as AI systems become more complicated and power-hungry. She said the company is now co-investing with Taiwanese partners to secure enough land, buildings and manufacturing capacity for future growth through at least 2029. AMD is reportedly working closely with companies including ASE Technology, Wistron, Inventec and several others inside Taiwan’s chip manufacturing network.
China also remains a huge part of AMD’s business despite ongoing U.S. export restrictions on advanced AI chips. Su confirmed that China still accounts for roughly 20% of AMD’s revenue and described it as an extremely important market for the company. At the same time, she emphasized that AMD would continue following U.S. export regulations while maintaining relationships with Chinese customers.
The broader message from AMD honestly feels clear now: the AI boom is no longer slowing down. Instead, it’s spreading deeper across the entire semiconductor industry and creating supply pressure almost everywhere, not just in the high-end GPU market that dominated headlines over the last two years.
