Recent remarks by Nvidia’s founder about the possibility of switching from TSMC to other contractors and news of product quality problems at a new factory built in Texas have prompted South Korean media to generalize these problems. According to them, Samsung is not doing well with the level of yield of suitable products not only for 4-nm technology, but also for 3-nm and 2-nm.
Image Source: Samsung Electronics
In fact, as The Korea Times notes, the 4nm process technology, as the most mature of this sample, suffers the least from problems with product quality, but it still does not help attract customers for contract manufacturing. The situation at Samsung’s Texas facility is now forcing the company to choose between the 4nm process, which it could master next year, and the 2nm process, which it won’t try to introduce at the site until 2026.
Even in its native South Korea, third-party customers are in no hurry to use Samsung’s 4nm process, so the company even revised the equipment installation procedure at the P4 plant in Pyeongtaek. Initially, it planned to combine the production of different types of memory and contract logic products on it, but may abandon the latter in favor of producing exclusively memory chips. Moreover, the same HBM is in good demand, and it will be economically justified.
Korean sources claim that Samsung has problems increasing the yield of products produced using 3-nm and 2-nm technologies. At the same time, the 3nm technological process is already in the second generation, and the company had to draw certain conclusions based on the results of mastering the first. Representatives of Eugene Investment & Securities believe that Samsung’s contract division is suffering from unprofitability. In the first half of the year alone, it brought the company about $1.12 billion in operating losses, although such statistics are not officially disclosed. Because of this, operating profit for Samsung’s entire business could drop to $4.15 billion in the third quarter, experts say.
The Korea JoongAng Daily reports that Samsung will transfer a large group of specialists who were previously involved in research and development to the relevant departments to solve problems with chip production. Up to two-thirds of the staff currently employed in research could undergo such a reorganization. Those researchers who specialized in more mature technical processes will be affected, so that the development of advanced lithographic standards does not suffer. So far this is only a preliminary plan, and it is difficult to say that such transformations will be implemented in practice.