Odaily Planet Daily report: Analyst Jukan from Citrini posted on X that Samsung Electronics' foundry division is expected to reach 100% capacity utilization in the second half of this year. Current capacity utilization is estimated at 70% to 80%, and given the backlog of orders and progress in contract negotiations, both internal and external stakeholders believe achieving this target is highly likely. The division has been operating at below 50% capacity utilization since 2024, but after about a year, it is nearing full capacity, showing signs of a turnaround in its long-term loss structure.
Industry insiders say the rise in capacity utilization is driven by high-bandwidth memory (HBM) substrate chips and growing demand for advanced products centered on major U.S. Big Tech customers. As 4nm technology is applied to the sixth-generation HBM, HBM4, memory demand is directly boosting foundry capacity utilization. Meanwhile, orders for 2nm nodes from major cloud service providers, AI, and high-performance computing customers are expanding, with ongoing discussions continuing with large clients such as Broadcom.
Customer diversification is also progressing. While Samsung’s foundry business has long been criticized for its reliance on specific clients, major tech companies such as Qualcomm, AMD, and Google have recently begun engaging in negotiations. Tesla is also reportedly assigning its next-generation chip orders to 2nm production, helping to expand the medium- to long-term order pipeline. However, the market generally believes that converting these orders into large-scale production contracts requires 2nm yield to stabilize in the 70% range; current yield is estimated at around 60%, and the pace of subsequent improvement remains a key focus. The business structure is also evolving. Revenue from advanced nodes is expected to exceed 50% this year, while AI and high-performance computing revenue is projected to rise from nearly 20% last year to over 30% this year. In the second half of the year, mass production of the second-generation 2nm SF2P mobile products will begin. Texas Taylor Fab 1 is on track for commissioning this year, and Taylor Fab 2 is scheduled to break ground within the year, targeting mass production by 2030. As demand for 1.4nm process technology grows, Samsung is also evaluating plans to add additional fabs.
At its Q2 earnings call on July 30, Samsung stated that utilization rates across all nodes improved year-over-year, with advanced nodes at 8nm and below reaching peak levels by focusing on high-growth-demand products. Samsung expects the number of 2nm project wins this year to more than double compared to last year. Regarding the timeline for returning to profitability, Samsung said it is difficult to specify an exact date but indicated that profitability should be achieved soon. Analysts expect non-memory businesses, including foundry and System LSI, to turn profitable as early as Q3 and no later than Q4. Some analysts believe that if wafer price increases are combined with normalized utilization rates, profitability improvements could outpace market expectations. Others in the industry suggest that the key threshold for major customers like Qualcomm and AMD to place full-scale orders is not achieving 100% capacity utilization, but rather whether 2nm yield can exceed 70%.