Samsung originally planned to achieve mass production of the 1.4nm process in 2027, but has adjusted its strategy to prioritize improving 2nm yield. Now it has once again put the commercialization of processes below 1.4nm on the agenda, with the latest timetable pointing to 2029. As Apple is rumored to be hoping to jump directly from 2nm to 1.4nm in just two years to ease the capacity and cost pressures brought about by the AI wave, Samsung is likely to become one of the important options in the American company's future dual supply strategy for wafer foundry.
According to Korean media The Bell, the outside world once believed that Samsung had abandoned the development of the 1.4nm node (internal code name SF1.4), but in fact, Samsung only made a scheduling adjustment: the mass production node originally scheduled for 2027 was postponed to 2029. Even so, judging from current information, Samsung will still lag behind TSMC by about a year in terms of mass production time - TSMC has previously announced plans to start mass production of its 1.4nm technology in 2028.
This delay is directly related to Samsung’s “increased priority” on 2nm GAA (SF2) and second-generation 2nm GAA (SF2P) yields. According to previous reports, Han Jinwan, president of Samsung's DS division, said that with the improvement of the yield rate of the 2nm process, the company's profitability has improved to a certain extent, which has also provided Samsung with confidence and space to re-promote the commercialization of 1.4nm. Against the background that AI chips are currently mainly concentrated on the 3nm process and are expected to gradually shift to 2nm in the future, Samsung's strengthening of the layout of SF2 and SF2P will help win high-value orders from companies such as NVIDIA.
However, from a longer-term perspective, Samsung is obviously not only focusing on AI customers, but also evaluating potential large-scale system customers, of which Apple is the most representative. It is reported that driven by the AI boom, Apple does not want to get into a situation of "grabbing production capacity" in wafer supply, so it plans to quickly switch to the 1.4nm node after only using TSMC's 2nm process for two years. Currently, TSMC’s 3nm monthly production capacity is about 175,000 wafers, but it is still in short supply driven by AI demand. It is widely expected that similar tensions will continue to the 2nm process.
At the cost level, Apple is also under pressure from TSMC’s 1.4nm process quotations. It is estimated that the price of a 1.4nm wafer is about $45,000, while a 2nm wafer is about $30,000, a difference of about $15,000. For Apple, which needs a lot of high-end chips, this means manufacturing costs will increase significantly during the jump from 2nm to 1.4nm. At the same time, rising storage and flash memory prices have also pushed up the price of Apple's entire machine: the price of a 12GB LPDDR5X memory module has climbed from about $39 to $145, and the price of 256GB NAND flash memory has risen from about $13 to $51. Against this background, Apple has recently raised prices on some Mac and other product lines.
It is precisely because of the dual pressures of cost and production capacity that Apple does not stick to a "single supplier" in its foundry strategy, but gradually moves towards diversification. In addition to TSMC, there are rumors that Apple has contacted Intel and plans to use Intel's 18A-P process in future products for the upcoming M7 chip; in the next step, Intel's 14A node is regarded as a potential candidate process for Apple's iPhone chips. This layout reflects Apple’s accelerating attempts at “multi-party manufacturing” on high-end process nodes to balance risks, production capacity and costs.
Under such a situation, Samsung's restart of 1.4nm commercialization will undoubtedly provide Apple with additional options. If Samsung can continue to improve yield and performance on the 2nm GAA process and successfully achieve mass production of SF1.4 in 2029, its competitiveness in the field of high-performance mobile and AI chips is expected to be significantly enhanced. For Apple, which hopes to take into account cutting-edge processes, cost control and supply security, it is not an impossible path to choose to partially use Samsung's 1.4nm node in a certain future generation of products, and use TSMC and Intel as a "multi-source combination".
From a more macro perspective, the 1.4nm process node is becoming the focus of the next round of competition in the semiconductor industry. TSMC continues its time lead in advanced processes, Samsung is trying to catch up through strategic adjustments and yield improvements, and Intel is trying to return to the forefront with processes such as 18A and 14A. In the context of AI and high-performance computing continuing to drive up the demand for computing power, this competition for 1.4nm and below nodes is not only about technical routes, but also about business models and cooperation ecology. In this process, top customers like Apple are both promoters and beneficiaries. Every adjustment in the selection of its foundry partners may set off new waves in the global semiconductor supply chain.