TSMC 3nm Reaches Full Capacity, 2nm to Ramp Quickly

According to a recent report by South Korean media outlet ZDNet Korea, TSMC is rapidly increasing the proportion of its advanced process node production, fueled by the surging global demand for AI chips. Among the highlights are the 3nm process, which has already entered mass production, and the upcoming 2nm process—both seen as key indicators of the semiconductor industry's current health.

Research from Counterpoint reveals that TSMC's 3nm node has reached full capacity utilization for the first time, following five consecutive quarters of ramp-up. This milestone is largely driven by increased demand for x86-based PC central processing units (CPUs) and various application processors used in high-performance computing and flagship smartphones. Notably, Apple's A17 Pro and A18 Pro chips, built for its latest devices, are among the major contributors.

Looking ahead, next-generation AI chips such as NVIDIA's upcoming Rubin GPU, Google's in-house TPU v7, and Amazon AWS's Trainium 3 are expected to sustain this momentum. These AI accelerators highlight the increasing need for cutting-edge semiconductor processes to meet the computational demands of modern PCs, mobile devices, and AI infrastructure.

While the 3nm node shows robust growth, TSMC's more mature technologies—such as 7nm and 6nm nodes—have experienced different market dynamics. Primarily designed for the smartphone sector, these nodes peaked in utilization around 2020 amid a mobile device boom but have seen more gradual capacity growth since then.

The 5nm and 4nm nodes, although also smartphone-oriented, are currently seeing a resurgence. Starting mid-2023, utilization rates began to recover—not solely due to smartphone demand but significantly fueled by the explosive growth of AI accelerator chips. High-performance AI chips like NVIDIA's H100, B100, B200, and the newly launched GB200—key components in AI data centers—are predominantly manufactured using 5nm and 4nm technologies. This AI-driven momentum is now revitalizing even what were previously considered mature process nodes.

TSMC's forthcoming 2nm technology is poised to achieve full capacity at a record pace. Market forecasts suggest that the 2nm node could reach optimal utilization within just four quarters of volume production—faster than any previous generation. This accelerated ramp-up reflects the simultaneous, powerful demand from both the smartphone and AI sectors.

In its Q1 2025 earnings call, TSMC confirmed that it expects more new chip design projects for the 2nm process in its first two years of mass production than seen in earlier nodes like 3nm, 5nm, or 4nm. These new designs are expected to be driven by both mobile and high-performance computing applications. In addition to core partner Apple, major IC design leaders such as Qualcomm, MediaTek, Intel, and AMD, along with leading integrated device manufacturers (IDMs), are actively evaluating or planning to adopt TSMC's 2nm technology. Their engagement is projected to be a crucial factor in maintaining high utilization rates for the next-generation process.