Is NVIDIA's Dominance Faltering? AMD CEO Lisa Su: AI Inference Demand Makes CPUs Key Again
AMD Chair and CEO Lisa Su stated on the 22nd that the growth of the AI infrastructure market is astonishing, and with the rise in inference demand, CPUs have returned to a critical core position. AMD aims to leverage its comprehensive product portfolio, including CPUs, GPUs, and ASICs, to capture opportunities in this massive market.
📋 Article Processing Timeline
- 📰 Published: May 22, 2026 at 13:24
- 🔍 Collected: May 22, 2026 at 13:31 (7 min after Published)
- 🤖 AI Analyzed: May 31, 2026 at 21:24 (223h 53m after Collected)
Central News Agency (Taipei, 22nd) AMD Chair and CEO Lisa Su stated today that the growth rate of the AI infrastructure market is astonishing, requiring different types of solutions. With the increase in inference demand, the Central Processing Unit (CPU) has returned to a critical core position, which is exactly where AMD is very strong.
Su attended the CommonWealth Magazine 45th Anniversary Flagship Forum this morning for an AI summit dialogue. Supply chain industry leaders in attendance included Wistron President Jim Hong, Lite-On Technology CEO Anson Chiu, and Wiwynn Chair Emily Hong.
During the media Q&A session after the dialogue, when asked how to compete with NVIDIA, which dominates the AI infrastructure field, she said that the growth rate of the AI infrastructure market is astonishing. Within the next 3 to 4 years, the data center market for AI infrastructure alone will exceed US$1 trillion (approximately NT$31.52 trillion), which is a massive growth.
Su believes that in such a huge market, different types of solutions are needed, including CPUs, Graphics Processing Units (GPUs), as well as Application-Specific Integrated Circuits (ASICs) and other technologies. This is AMD's unique advantage, as AMD's technology covers these core components.
Su said, 'Looking back 6 or 12 months ago, no one was talking about CPU shortages. But with the increase in inference demand, the CPU has returned to a critical core position, which is exactly where AMD is very strong.' As the AI infrastructure market evolves, various types of computing technologies will be seen, and AMD's technology will perform extremely well across the entire infrastructure market.
Su is also optimistic about the growth potential of CPUs in future AI racks. She said that in the past 3 to 4 years, the CPU market growth rate was only about 3% to 4%, and all focus was on GPUs. With the explosion of inference and agentic AI, the CPU market has seen significant growth since the end of last year. In the next 5 years, she expects the CPU market to grow at a rate of over 35% annually.
She emphasized that AMD's strategy is not just to launch a single CPU, but to build a complete family. AMD's new Venice CPU, which uses 2nm technology, includes cloud-optimized versions, host node-optimized versions, and agentic AI-optimized versions. 'I think this is what the market needs. It's not just a single design specification, but a full suite of products that we provide to meet different design needs,' she said.
Su attended the CommonWealth Magazine 45th Anniversary Flagship Forum this morning for an AI summit dialogue. Supply chain industry leaders in attendance included Wistron President Jim Hong, Lite-On Technology CEO Anson Chiu, and Wiwynn Chair Emily Hong.
During the media Q&A session after the dialogue, when asked how to compete with NVIDIA, which dominates the AI infrastructure field, she said that the growth rate of the AI infrastructure market is astonishing. Within the next 3 to 4 years, the data center market for AI infrastructure alone will exceed US$1 trillion (approximately NT$31.52 trillion), which is a massive growth.
Su believes that in such a huge market, different types of solutions are needed, including CPUs, Graphics Processing Units (GPUs), as well as Application-Specific Integrated Circuits (ASICs) and other technologies. This is AMD's unique advantage, as AMD's technology covers these core components.
Su said, 'Looking back 6 or 12 months ago, no one was talking about CPU shortages. But with the increase in inference demand, the CPU has returned to a critical core position, which is exactly where AMD is very strong.' As the AI infrastructure market evolves, various types of computing technologies will be seen, and AMD's technology will perform extremely well across the entire infrastructure market.
Su is also optimistic about the growth potential of CPUs in future AI racks. She said that in the past 3 to 4 years, the CPU market growth rate was only about 3% to 4%, and all focus was on GPUs. With the explosion of inference and agentic AI, the CPU market has seen significant growth since the end of last year. In the next 5 years, she expects the CPU market to grow at a rate of over 35% annually.
She emphasized that AMD's strategy is not just to launch a single CPU, but to build a complete family. AMD's new Venice CPU, which uses 2nm technology, includes cloud-optimized versions, host node-optimized versions, and agentic AI-optimized versions. 'I think this is what the market needs. It's not just a single design specification, but a full suite of products that we provide to meet different design needs,' she said.
FAQ
What is AMD's AI strategy?
It focuses on a comprehensive product family including CPUs, GPUs, and ASICs.