At the close of the first quarter of 2025, major industry sources revealed that TSMC had completed milestone trial production runs of their leading-edge 2 nm (N2) fabrication process across premier facilities. This achievement sets the stage for mass production, expected to commence at multiple sites by year´s end, marking a significant leap for the world´s top contract chip maker. TSMC has also begun laying groundwork for an even more advanced 1.4 nm process at its ´P2´ Baoshan plant, though commercial readiness is not anticipated before 2028.
Yield improvements have played a pivotal role in TSMC´s confidence to advance these next-generation technologies. Insiders report that internal teams and state-of-the-art manufacturing tools have recently pushed yields for ´memory product´ silicon past a remarkable 90%—a threshold that secures both profitability and manufacturability for high-stakes customers. Earlier, hitting a 70% yield was considered sufficient to launch full-scale production, so these latest figures highlight rapid progress. The leap in yield aligns with public statements from TSMC executives who noted an unprecedented surge in demand for 2 nm wafers, exceeding interest observed for previous node transitions.
Meanwhile, competition between TSMC and Samsung remains fierce. As of mid-May, Samsung´s comparable 2 nm ´SF2´ process is still reported to be in the testing phase, with yields surpassing the 40% mark for gate-all-around (GAA) transistor trials. While this is seen as a substantial step for Samsung Foundry, it underscores a significant gap versus TSMC´s 2 nm advancements. Amid reports that TSMC may impose higher-than-expected prices on crucial customers, Samsung Semiconductor executives are rumored to be courting industry giants such as NVIDIA and Qualcomm in an effort to close market share gaps and attract clients away from the leader. With rapid advances and cross-industry maneuvering, the stage is set for an intense battle to supply the next generation of chips for applications ranging from mobile devices to Artificial Intelligence.
