On July 16, Samsung Electronics is considering outsourcing the backend design work for Google's 10th generation TPU (codenamed Icefish). This TPU, based on a 2nm process, consists of a Compute processor and an I/O chip—the former is expected to be manufactured by TSMC using a 1.4nm process, while the latter will be produced by Samsung using a 2nm process, responsible for data transmission between the Compute processor and HBM. Google is co-designing the chip with MediaTek, aiming for mass production as early as 2028. However, a recent influx of 2nm orders at Samsung—beyond Google and Tesla, it has also secured clients Anthropic and DeepX—has led to a tight internal workforce situation, especially since Tesla's 2nm autonomous driving chip's backend design is still being handled in-house by Samsung. Potential outsourcing partners mentioned include ADTechnology, Gaonchips, and Alphachips. The first two have each invested in large projects—ADTechnology is focusing on the 2nm CPU project ADP620, targeting annual revenues exceeding 1 trillion won between 2028 and 2029; Gaonchips is preparing to participate in the South Korean Ministry of Trade, Industry and Energy's approximately 800 billion won K-On-Device AI project, collaborating with Hyundai Motor to develop 5nm ADAS chips. Due to the backend design being essentially a lower value-added service contract (typically in the hundreds of billions of won), far less lucrative than ASIC projects that can reach trillions of won, both companies are not very enthusiastic, preferring to take on limited work to establish performance in advanced process projects. In contrast, Alphachips views the Google TPU project as a growth driver and is more willing to participate. Industry insiders point out that TSMC's inability to handle orders in the 2nm market due to capacity constraints is also a deeper reason for Samsung's workforce strain.