Samsung Electronics сотрудничает с Tesla по AI-чипам, посылая сигнал о восстановлении в полупроводниковой промышленности

Samsung Electronics’ semiconductor business is driving overall performance with the good momentum of the memory division, but some observations suggest that it will take more time for its foundry business to break even. However, as cooperation with Tesla on artificial intelligence chips expands from current production to next-generation products, this has become a key variable in judging the timing of a foundry business rebound.

According to industry news on the 26th, Tesla CEO Elon Musk publicly revealed the upgrade plan for its AI chip “AI4” at the earnings release on the 22nd. He预计 mass production around mid-2027, and stated that Samsung Electronics is working on design modifications for this chip, with whether it can ultimately transition to mass production being crucial. Industry interpretations suggest that Tesla has effectively indicated that the production of the improved AI4 product will be outsourced to Samsung. This product is a higher capacity and computational performance model than the AI4 mass-produced in 2023, possibly called “AI4+” or “AI4.1”.

The cooperation between the two companies has expanded to the next-generation chips. Samsung Electronics has decided to participate in the production of Tesla’s “AI5” and “AI6”, with the AI6 order officially confirmed last July, reportedly amounting to about 23 trillion Korean won. This is the largest contract for a single customer in Samsung’s semiconductor division. Currently, the existing AI4 is produced using 7-nanometer process on the foundry line in Pyeongtaek, Gyeonggi Province, South Korea, and this upgraded product is also very likely to be produced in Pyeongtaek. AI5 and AI6 are planned to be produced using 2-nanometer process at the Texas factory in the United States. It is reported that AI5 will share production with TSMC, while AI6 will be exclusively produced by Samsung Electronics. These chips are AI semiconductors自主研发用于自动驾驶的人工智能半导体，用于实现车辆的完全自动驾驶功能。

The background for Samsung’s foundry business to continuously secure Tesla orders is believed to be the trust accumulated during the current AI4 production process, as well as the yield and performance of advanced processes gradually entering a stable range. Yield refers to the proportion of semiconductor products on a wafer that are actually functioning normally; if this value is low, even with factory operation, profitability is difficult. Recently, Musk also announced on X that AI5 has completed tape-out, and expressed gratitude to Samsung Electronics and TSMC. Tape-out is the first step in moving from semiconductor design to pilot production. Correspondingly, the Taylor factory held an equipment entry ceremony on the 24th, with initial operation expected to start as early as the end of this year, and正式开始生产特斯拉相关订单 from next year.

The issue lies in the time gap. Samsung Electronics’ preliminary operating profit for Q1 2026 announced earlier this month was 57.2 trillion Korean won, but the market estimates that about 95%, or roughly 54 trillion Korean won, came from the memory division. The non-memory division, including foundry, is observed to have a loss of about 1 trillion Korean won. The securities industry believes that the annual loss of the non-memory division this year will reach 3 to 4 trillion Korean won. Therefore, the Tesla effect is more reflected in the actual shipment and revenue recognition timing rather than in the contract signing itself. Based on this, industry experts believe that the watershed for performance improvement will be after 2027, when the Taylor factory begins operation and customer shipments are combined. Samsung Electronics will announce detailed performance for each division in the first quarter on the 30th, and questions about the specific operation schedule of the Taylor factory and current customer cooperation are expected to be concentrated at that time. This trend may become a litmus test for whether Samsung’s foundry can surpass mere cost reduction and, by leveraging large clients, restore its competitive edge in advanced process technology.