
Taiwan Semiconductor Manufacturing Company (TSMC) captured roughly 72% of the global semiconductor foundry market in the third quarter of 2025, as artificial intelligence chip demand drove record revenue, according to Counterpoint Research.
Global semiconductor foundry revenue increased approximately 17% year-over-year to $84.8 billion in Q3 2025. This growth stemmed from strong demand for AI applications that rely on advanced manufacturing processes.
Manufacturer Market share Key technology focus TSMC 72.0% 3nm / 5nm / CoWoS Packaging Samsung 11.5% 2nm GAA / HBM4 Base Die SMIC 5.7% 7nm / Domestic AI Infrastructure UMC 2.3% 22nm / 28nm Specialty Nodes GlobalFoundries 2.0% Automotive / PhotonicsTSMC’s foundry revenue rose more than 40% annually during the quarter. This performance outpaced the broader market and expanded the company’s lead over competitors.
The revenue increase resulted from higher output of TSMC’s 3-nanometer manufacturing process. Utilization rates remained robust for 4-nanometer and 5-nanometer capacity. These processes serve AI accelerator customers such as Nvidia, Advanced Micro Devices, and Broadcom.
Advanced technologies with nodes below 7 nanometers represented nearly three-quarters of TSMC’s wafer revenue in Q3 2025. The 3-nanometer process accounted for 23% of that revenue. The 5-nanometer process contributed around 37%.
Rival foundries experienced slower growth. Non-TSMC players increased revenue by only about 6% collectively. This reflected softer orders following earlier demand linked to tariffs. China subsidy programs provided some support.
Samsung Electronics, TSMC’s closest competitor, maintained a mid-single-digit market share. Texas Instruments, Intel, and Infineon held similar positions. Analysts noted capacity limits at advanced nodes and constraints in advanced chip-packaging technology known as CoWoS. These factors may limit sequential growth in the fourth quarter of 2025.
Full-year 2025 foundry revenue growth reached an estimated 15%. The pure-play foundry segment grew at a faster pace. This occurred through continued shipments of AI GPUs and custom AI chips.