At an AI policy and business event in Washington, D.C., on July 23, 2025, AMD CEO Lisa Su revealed that chips manufactured for AMD by Taiwan Semiconductor Manufacturing Co. (TSMC) at its new Arizona facilities will cost between 5% and 20% more than their Taiwanese equivalents. Despite the cost increase, Su emphasized that these U.S.-based chips will play a vital role in strengthening AMD’s supply chain resilience and manufacturing diversity. She confirmed that AMD expects to receive its first batch of Arizona-made chips by the end of 2025.
The announcement reflects AMD’s continued investment in onshore production in response to increasing geopolitical risks and global supply chain disruptions. Su stated that yields from TSMC’s U.S. fab are now comparable to those from its Taiwanese plants, suggesting no compromise on chip quality. She also stressed that higher costs are an acceptable trade-off for predictable and secure production pipelines, particularly for AI applications with long-term deployment cycles.
The remarks come at a time when the demand for AI accelerators is exploding. Lisa Su projected that the total market size for AI chips could surpass $500 billion in the coming years. She referenced massive investments in the AI sector from major figures like OpenAI’s Sam Altman and xAI’s Elon Musk as signs that the ecosystem is expanding rapidly. AMD has positioned itself to compete more directly with Nvidia, which still holds the dominant share of the AI compute market.
Beyond her comments on chip sourcing, AMD has made several major moves this quarter. At its “Advancing AI” event in June 2025, the company launched its Instinct MI350 series accelerators, which deliver four times the AI performance of the previous generation and include up to 288 GB of HBM3e memory. AMD also previewed its upcoming Helios rack system, due in 2026, built around the MI400 series and next-gen Epyc “Venice” CPUs.
Institutional sentiment toward AMD is improving. HSBC recently upgraded AMD to a “Buy” rating, citing a strong roadmap and momentum in AI products. Analysts praised the MI350’s performance-to-cost ratio and noted it could pressure Nvidia’s Blackwell B200 systems in both hyperscaler and enterprise deployments. AMD stock has also rebounded sharply since April, with multiple analysts raising their price targets ahead of the company’s Q2 2025 earnings call scheduled for August 5.
Su’s remarks highlight a larger trend in the semiconductor industry: the willingness to absorb higher costs for the sake of resilience, localization, and strategic autonomy. As AMD becomes more competitive in the AI accelerator space and diversifies its supply chain, it stands poised to capture greater market share from Nvidia and other incumbents—if execution aligns with ambition.
Discover more from Semiconductors Insight
Subscribe to get the latest posts sent to your email.
