Advanced Micro Devices is committing tens of billions in global capital to expand AI chip supply through 2027, with CEO Lisa Su signaling that current capacity remains insufficient to meet multi-year demand growth.
AMD is raising its Taiwan supply chain investment beyond the $10 billion commitment announced in May. Su indicated the company intends to deploy tens of billions globally, citing sustained growth in AI computing hardware demand.
The scale of demand prompted AMD to extend its planning window with suppliers from one to two years to three to five years—a structural shift that coordinates capacity expansion across the chip manufacturing ecosystem. Su's visits to Taiwan suppliers on Oct. 5, including stops at Taiwan Semiconductor Manufacturing Co. Acer, Asus, Foxconn and Quanta, focused on ensuring partners could expand capacity for CPUs, GPUs and AI infrastructure products.
Later that day, Su traveled to South Korea for talks with memory chipmakers Samsung and SK Hynix, underscoring memory components as a critical input to AMD's AI infrastructure supply strategy.
AMD's Helios AI platform began shipments in the third quarter on schedule, with volumes expected to grow in subsequent quarters. Second-quarter data center revenue reached $6.7 billion, up 107 percent year-over-year, driven by EPYC processor and Instinct GPU sales.
Memory supply remains the binding constraint. Su said AMD is coordinating with suppliers and customers on memory allocation, including high-bandwidth chips for AI servers and conventional memory for CPUs and personal computers.
When asked about a potential TSMC investment in Texas, Su declined direct comment, noting that TSMC's Arizona facility has performed well for AMD.
