Samsung Electronics' foundry division is accelerating efforts to improve the yield of its 2-nanometer manufacturing process, and it is using Tesla's orders as a strategic springboard to enter the market of major technology customers.

On September 28th, according to South Korean media ChosunBiz, which cited industry insiders, the yield rate of Samsung Electronics' 2-nanometer manufacturing process has further improved from the previously widely recognized level of around 55% to nearly 60% at present.

Against the backdrop of stable yield rates, Samsung is focusing its engineering resources on optimizing power consumption and heat dissipation performance in order to build a foundation of trust during technical negotiations with potential major customers.

At the same time, the second-generation 2-nanometer process SF2P, which is designed to address the shortcomings of existing manufacturing processes, is also being accelerated, with the goal of achieving mass production within this year.

Analysis suggests that this move has a clear strategic purpose: to use next year's supply of 2-nanometer chips to Tesla as an opportunity to fully break into the orders of large technology companies.

Yield rate approaches 60%, power consumption efficiency becomes a key variable

According to reports citing industry sources, the improvement trajectory of Samsung's SF2 process yield is clear: it was around 30% in the middle of last year, rose to the 50% range in January this year, reached around 55% around August, and is now approaching 60%.

SF2 is Samsung's first 2-nanometer mass-production process that adopts the full surround gate (GAA) architecture. The first mass-produced product is the self-developed mobile application processor Exynos 2600, which has been incorporated into some models of the Galaxy S26 series launched at the beginning of the year. It is manufactured at the S3 production line in Hwaseong, Gyeonggi Province.

Samsung has adopted copper heat dissipation modules in the chip packaging, reducing the thermal resistance by 30%.

However, after its launch on the Exynos 2600 platform, some overseas tests revealed that it consumed more power when performing the same tasks than Qualcomm's latest Snapdragon products, which raised doubts about the energy efficiency of this manufacturing process. Some argue that this reflects the inherent limitations of the SF2 manufacturing process itself.

In response to the aforementioned issues, Samsung is preparing for the second generation of 2-nanometer process SF2P, with the goal of mass production within this year. It is reported that SF2P offers a 12% improvement in performance, a 25% reduction in power consumption, and an 8% reduction in chip area compared to SF2.

Samsung's current focus in engineering is on ensuring the consistency of product performance during the mass production phase. Specifically, this involves controlling the heat dissipation and power consumption variations after the chips are integrated into customer products within an acceptable range. This capability will serve as a key technical advantage for Samsung when negotiating with large chip design companies ( Fabless ).

TSMC dominates the top spot, while Samsung and Intel compete for second place

Currently, the global advanced process foundry market exhibits a clear pattern of 'one leader and two strong players'.

TSMC, with core customers such as Apple, AMD, and NVIDIA, recorded a 72.5% market share in the second quarter of this year, while Samsung's manufacturing share was only 5.9%, further widening the gap between the two.

TSMC's 2-nanometer wafer prices have stabilized at around $30,000, which is 50% higher than the 3-nanometer price of about $20,000, and they also maintain a 'no discount' policy for major customers. Industry insiders generally believe that TSMC's 2-nanometer production capacity has been booked until 2028.

On the Intel side, their 2-nanometer process 18A has been mass-produced at Fab in Arizona, USA. The produced Panther Lake processors have helped Intel gain a 56% market share in AI PC, achieving certain results in their own product line.

However, there has been no significant progress in recruiting external customers for 18A. It is reported that Intel internally hopes to rely on the next-generation 14A process, which is planned to be mass-produced after 2027, to break through in attracting external customers.

In this context, there is a general expectation in the industry that the market landscape will be dominated by TSMC, with Samsung and Intel competing for the second tier. Regarding the break-even point of Samsung's foundry business, the mainstream view in the market is that it must be observed in conjunction with the capacity utilization rate of the Taylor wafer factory in Texas.

A person in the semiconductor industry stated:

A yield rate of nearly 60%, along with the initiation of power consumption optimization efforts, are signs of a meaningful rebound.

But at the same time, it is pointed out that:

The gap in market share with TSMC is too large, and it will still take more time to achieve a real transformation in the market landscape.