Samsung Electronics encounters manpower shortage and plans to outsource some chip back-end design
In the context of the surge in AI demand, semiconductor production bottlenecks not only appear in the wafer foundry link, but also extend to the design link. According to TheElec, Samsung Electronics is considering outsourcing the back-end design work of Google's TPU (Tensor Processing Unit) I/O chip based on the 2nm process due to related labor shortages caused by the increase in chip factory production. The TPU consists of a computing processor and an I/O chip. The latter is responsible for data transmission between the computing processor and high-bandwidth memory (HBM).
In the context of the surge in AI demand, semiconductor production bottlenecks not only appear in the wafer foundry link, but also extend to the design link. According to TheElec, Samsung Electronics is considering outsourcing the back-end design work of Google's TPU (Tensor Processing Unit) I/O chip based on the 2nm process due to related labor shortages caused by the increase in chip factory production. The TPU consists of a computing processor and an I/O chip. The latter is responsible for data transmission between the computing processor and high-bandwidth memory (HBM). Currently, Samsung Electronics is considering cooperating with three local Korean semiconductor design service companies, AD Technology, Gaonchips, and Alphachips. According to reports, these companies may undertake some back-end design services in order to ensure a successful record in the 2nm projects of large technology companies. The so-called back-end design means that before manufacturing chips, DSP (digital signal processing) engineers will convert their designs into a format suitable for wafer foundry production. The work includes steps such as arranging and connecting logic circuits, designing test circuits, and verifying the design. In addition to Google TPU, Samsung Electronics is also responsible for the back-end development of Tesla's self-driving chips, Anthropic's self-developed AI chips, and Korean startup DeepX products. These chips are all produced based on the 2nm process. An industry insider said: "Because TSMC's production capacity has reached its limit, some orders that cannot be processed have also flowed to Samsung Electronics." This has exacerbated the shortage of back-end design capacity to a certain extent. Currently, the semiconductor industry continues to face production capacity shortages. As far as the 2nm process is concerned, TSMC is the only semiconductor manufacturer that can stably achieve high-yield mass production and support corresponding packaging technologies, resulting in AI infrastructure needs that cannot be fully satisfied. Previously, TSMC President Wei Zhejia said that TSMC’s global chip supply will not be able to meet the demand driven by AI in the next few years. In this regard, Nomura warned that the AI semiconductor cycle is far from peaking, and there may be an "epic-level" supply chain mismatch in the second half of 2026. As cloud vendors' capital expenditures continue to expand, the structural shortage of semiconductors will directly exacerbate short-term market price fluctuations, but it also confirms the long-term sustainability of this cycle. (