Samsung strives to change the customer composition structure and breakthrough process technology

Previously, Samsung expected that the 4nm process will be mass-produced in 2021, and the 3nm GAAFET process may be launched in 2023. Compared to TSMC’s N3 process, Samsung is probably 6 to 12 months late. Samsung not only needs to catch up with TSMC in process research and development but also follows closely behind in foundry orders. It is the largest foundry company outside TSMC. Due to the high cost of research and development of advanced technology, even if Samsung’s own chip orders are not low, it still needs a large number of orders from other customers to share costs and maintain operations.
According to DigiTimes, Samsung’s foundry profit in the second quarter of 2021 was only $268 million, far less than TSMC’s $13 billion. Not even as good as SMIC or United Microelectronics (UMC), currently relying on Samsung’s own orders and low-cost competition to maintain its position as the second-largest foundry manufacturer. Samsung has fallen behind in the competition for orders with TSMC. In addition to the level of craftsmanship, part of the reason is that many chip designers have concerns about using Samsung foundries because they are likely to be direct competitors of Samsung chips.

Samsung has considered splitting its wafer production business to reduce third parties’ suspicion of using Samsung’s foundry, but it has not taken action. In order to remain competitive, Samsung still has a lot of work to do, to convince potential customers to choose them instead of competitor TSMC. Without sufficient order support, it is difficult to maintain technological competitiveness. More troublesome is that Intel intends to compete for the foundry market. Although there are rumors that Tesla intends to shift orders from TSMC to Samsung, and Google is also cooperating with Samsung to develop Tensor SoC, the number of orders from these customers is far from enough.

TSMC has received more than 60 EUV lithography machines by 2021. Samsung will have less than 20 by the end of 2020, and it will be difficult to have 30 even by the end of 2021. Samsung will still find it difficult to challenge TSMC for some time to come, but it may seize some business opportunities in the high-end market. As Intel joins the battle, there may be some subtle changes in the high-end foundry market after 2023.