2nm chip process war upgrade! Qualcomm (QCOM.US) returns to Samsung to switch from single-source TSMC to a dual foundry chain

The Zhitong Finance App learned that some media quoted Cristiano Amon (Cristiano Amon), CEO of Qualcomm in the US, as reporting that Qualcomm (QCOM.US) will probably use Samsung Electronics' 2nm chip foundry process to manufacture its next generation mobile application processor.

The media quoted information revealed by Amon in an interview as saying that the smartphone and PC chip leader is currently discussing contract OEM manufacturing using the most advanced 2nm chip manufacturing process; Amon said that most of Qualcomm's next-generation core chip design work for OCs, smartphones, and even data centers has been completed to achieve large-scale OEM and full commercialization in the near future.

The report added that this potential deal will also mark Qualcomm's return to Samsung Electronics' cutting-edge chip manufacturing node after relying almost entirely on TSM.US for the most advanced chip manufacturing process for many years.

On Monday, TSMC's stock price once soared 6.9%. The stock price once again reached a record high, pushing the Taiwan Stock Market Benchmark Index to break through the 30,000 mark. This rise occurred after Wall Street financial giant Goldman Sachs sharply raised TSMC's target price by 35% to NT$2,330 within 12 months, further strengthening the market's confidence in the long-term strong growth in demand for artificial intelligence-related computing power infrastructure. TSMC shares recently closed at NT$1,675. In the US stock market, TSMC's US stock ADR has risen sharply by 8% since the beginning of 2026, with a market value close to 2 trillion US dollars.

AI GPUs, which have been in high demand since 2023, and AI ASICs, which have recently exploded in demand, are inseparable from TSMC. TSMC is currently the world's largest contract chip foundry. As the fanatical wave of AI layout continues to cool down and continues to take the world by storm, its customers Nvidia, and chip giants such as AMD and Broadcom continue to benefit from the market's surge in AI chips, the core infrastructure for AI. These chip giants have surged in the size of chip foundry contracts with TSMC, which in turn has driven TSMC's performance to continue to exceed expectations and expand strongly since last year. This is also an important logical support for TSMC's shares and ADR shares that have repeatedly reached new highs in recent years.

TSMC and Intel's advanced chip process layout

TSMC, which has the title of “King of Global Chip Foundry,” also fully focuses on the most advanced chip manufacturing capacity of 2nm and below. The company clearly stated in the process description on its official website that 2nm (N2) “began mass production as planned in the fourth quarter of 2025 (4Q25),” and emphasized that N2 uses first-generation nanosheet (nanosheet) transistors (that is, GAA/surround gate direction). TSMC's official website also indicates that Fab 20 and Fab 22 are 2nm advanced process chip production facilities. Recently, the media further revealed that TSMC's initial mass production/climbing focus was more biased towards Kaohsiung Fab 22.

Compared with the N3E process (equivalent to the 3nm advanced process), N2 can achieve performance of +10% to 15% under the same power consumption, or 25% to 30% reduction in power consumption, and +15% to 20% transistor density (depending on the chip architecture design).

Another chip manufacturing giant, Intel, skipped the 2nm integer process and focused on the 1.8nm process (that is, 18A). Intel said that 18A has moved from “process node commitment” to “terminal product release and mass production climbing”, while 14A is still in the R&D and roadmap/customer introduction stage. Intel heavily launched/showcased Panther Lake (Core Ultra Series 3) based on the 18A process at CES 2026, and confirmed it as the 18A “first batch of product platforms”. The key process features of the 18A, namely GAA/RibbonFET+ back-powered PowerVia, were also highlighted by the media and Intel's official materials as a selling point for this generation of products and processes, but specific production yield indicators have not yet been announced.

What does Qualcomm's return to Samsung Electronics mean for TSMC and Intel?

Gao passed “almost completely relied on TSMC” in terms of leading manufacturing processes in the past few years, and if some advanced node orders for the next generation mobile AP were to be transferred to Samsung 2nm this time, it would mean that Qualcomm would launch a dual supply chain strategy or decentralization on the “most core and largest mobile phone SoC”, putting slight pressure on TSMC at the level of share and bargaining power, but there is no major disturbance to TSMC's fundamental growth prospects.

Qualcomm's choice of Samsung OEM mobile processors is more like “customer risk hedging and production/cost rebalancing”, which does not necessarily mean that TSMC's competitiveness at 2nm is being questioned; TSMC may still hold a very large share of Qualcomm (such as AIPC core chips from Qualcomm and the upcoming launch of AI inference chips in data centers), and overall high-end production capacity usually doesn't worry about capacity digestion under the almost endless strong demand for AI/HPC. In addition, Apple has already locked in part of TSMC's 2nm production capacity, and TSMC will even need to expand in 2026 Zhang advanced process chip production capacity.

This could be a major blow to Intel's foundry prospects. At a stage where Intel is striving to use 18A and more advanced nodes to acquire major external customers and strengthen the chip foundry growth narrative, if Qualcomm prioritizes returning to Samsung over Intel, it means that it will still be difficult for Intel to obtain substantial verification in the short term for “the iconic customer of high-end mobile SoC.”

18A and more advanced 16A technology can be said to be a key point in whether Intel can turn over, so for the growth prospects of Intel, which has been underperforming for a long time, it is essential to master advanced process chip manufacturing technology that is leading the world and is extremely suitable for mass production. This established American chip giant didn't even hesitate to buy a more expensive high-NA EUV lithography machine — this is also the world's first High-NA delivered by Asmack, and Intel strives to create the most advanced chip manufacturing technology at 2nm and below, ahead of TSMC and Samsung Electronics.