TSMC shifts past 3nm to cement 2nm dominance, Samsung Electronics’ catch-up faces headwinds
Input
Modified
TSMC’s 2nm node pulls in AI chip demand, tightening its grip on the market With 3nm capacity tapped out, TSMC pivots to a major 2nm investment expansion Samsung bets on 2nm, but yield and customer gaps remain clear constraints

Taiwan’s TSMC, the world’s largest foundry, is strengthening its presence in the market with its next-generation 2-nanometer (nm) process. With strong technology and a deep roster of customers, TSMC is absorbing much of the surging demand for advanced semiconductors fueled by the AI boom. Samsung Electronics, the industry’s No. 2 player, is also stepping up efforts to bolster its 2nm competitiveness, but the prevailing view in the market is that it will be difficult for Samsung to catch TSMC in leading-edge process technology in the near term.
TSMC secures an edge in the 2nm foundry race
On the 7th (local time), tech outlet WCCFTech, citing semiconductor industry sources, reported that demand for TSMC’s 2-nanometer (N2) process is surging faster than expected. According to the report, the number of tape-outs for TSMC’s 2nm technology is 1.5 times higher than for its 3nm technology. A tape-out refers to the stage at which a fabless (design-focused) company completes a chip design and hands off the final layout to a foundry—effectively the last gate before mass production. A rise in tape-outs implies that more types of chips, and larger volumes, are likely to move into mass production on TSMC’s 2nm lines.
The trend is being driven by the rapid expansion of the AI semiconductor market. WCCFTech said TSMC is expected to maintain a dominant 95% share in the AI accelerator market, and projected that, powered by the AI boom, TSMC’s monthly wafer output target for its 2nm process could reach a record 140,000 wafers by the end of 2026. The revenue mix is also expected to shift sharply. The industry forecasts that in the third quarter (July–September), TSMC’s 2nm revenue will surpass the combined revenue from its current mainstay 3nm and 5nm nodes. That would signal a faster-than-ever transition toward the most advanced processes.
More than 50% of TSMC’s initial 2nm capacity has reportedly been allocated to Apple. Once again, Apple—TSMC’s largest customer—is said to have secured the earliest 2nm volume. Most of that capacity is expected to go toward producing the A20 and A20 Pro chips for the next iPhone 18 lineup, as well as the M6 chip for OLED MacBook Pros. Qualcomm and MediaTek are also cited as major customers for TSMC’s 2nm mobile chips.
Backing the leading edge over 3nm
As 2nm emerges as a key battleground for market leadership, TSMC is gradually shifting emphasis away from its 3nm node. Taiwan’s Commercial Times reported on the 8th that supply shortages for TSMC’s 3nm process have recently persisted. With existing 3nm capacity struggling to meet surging demand, TSMC has moved to raise 3nm pricing and temporarily halt the launch of new 3nm projects—effectively accelerating customers’ roadmaps toward 2nm. TSMC’s 3nm lines are said to be essentially fully booked, driven by orders for AI GPUs, ASICs for cloud data centers, and high-end mobile application processors (APs).
Investment in 2nm capacity is also accelerating. Based on foreign media reports, TSMC said at a meeting with government agencies including the National Science and Technology Council (NSTC) last November that it plans to build three additional 2nm fabs in Taiwan to respond to a sharp rise in AI chip demand. The fabs are widely expected to be located in the Southern Taiwan Science Park being developed by Tainan City, with total investment estimated at about $3.1 billion. The industry expects TSMC’s monthly 2nm wafer output to reach 80,000–90,000 wafers by the end of next year, roughly doubling from the current 40,000 wafers.
TSMC is also pushing aggressively into post-2nm development. Taiwanese media recently cited supply-chain sources saying TSMC could accelerate construction of a new fab in the Central Taiwan Science Park to support its A14 node. The plant is expected to complete risk production by the end of 2027 and begin mass production in 2028. A14 refers to TSMC’s 1.4nm foundry process. Previously, TSMC said A14 is designed to drive AI-led innovation with faster computing and better power efficiency, and could also enhance on-device AI capabilities in smartphones. The company said development is proceeding smoothly and yields are running ahead of plan.

The reality of Samsung Electronics’ 2nm process
As TSMC rapidly strengthens its lead in advanced-node competition, rival Samsung Electronics is also doubling down on 2nm, led by mobile application processors such as the Exynos 2600. Samsung, the world’s second-largest foundry, began mass production of 3nm chips in 2023—the first in the industry—but failed to gain the upper hand in the market. Difficulties in securing stable yields drove its market share lower instead. If development of the 2nm process proceeds smoothly, Samsung could secure a turning point after its setback in the 3nm race.
Still, whether Samsung can gain a real edge in market competition remains uncertain. The industry estimates that initial yields for Samsung’s 2nm process will be around 50–60%, well below TSMC’s level of over 80%. Another headwind is that TSMC has already locked in the market by pulling its existing 3nm customers into 2nm. Semiconductor publication EE Times reported that TSMC’s 2nm customer list includes Nvidia, AMD, Apple, Qualcomm, Google, Microsoft, and Amazon, adding that experts expect TSMC to maintain its advantage over Samsung and Intel for years to come. Research firm TechInsights told EE Times that TSMC’s 2nm node is likely to become the biggest “hit” process in history, noting that more than 15 customers are already lined up.
Analysts also point out that as foundry technology grows more complex, orders inevitably concentrate with TSMC, which offers higher yields and proven stability. For large customers, designing advanced chips typically costs about $800 million, while annual production requires $20–30 billion. If a foundry fails in mass production, customers are left to absorb massive losses. EE Times said these market dynamics have further reinforced TSMC’s appeal over Samsung and Intel, both of which have previously struggled with advanced-node supply.