- Published on
South Korean Semiconductor Engineer Accuses Tesla of Pressuring Samsung to Lower AI Chip Prices, Seeking Sensitive In...
- Authors
- Name
- DeepHorus
A Korean semiconductor engineer recently publicly accused Tesla on the social platform X of trying to secure extremely favorable terms while negotiating a chip foundry agreement with Samsung. The engineer, whose account is @TTHUEONE, said Tesla wanted both access to sensitive manufacturing process information and audit rights, while also demanding lower prices. He also said Tesla's Terafab project had poached a large number of employees from Samsung's Taylor, Texas, fab. Neither Tesla nor Samsung has confirmed or denied the engineer's allegations.
The dispute emerged after Tesla CEO Elon Musk had just finished a busy week. He publicly praised Micron Technology, saying it "has a rational attitude toward future memory chip pricing, while other manufacturers are engaging in price gouging." Those remarks were widely interpreted as an implicit reference to Samsung. Musk also suggested TSMC might participate in the Terafab project, but ruled out the possibility of TSMC holding equity or taking part in operations. Terafab is an ambitious semiconductor project between Tesla and SpaceX.
Claims about talent being poached are not new. A Business Post report in March this year pointed out that the two factories are geographically close, posing a structural risk to Samsung, and mentioned that Terafab would compete for labor and supplier resources in the same region. At that time, Tesla was already recruiting semiconductor R&D engineers for the project. Recently, reports said Samsung had dispatched advanced-process yield engineers from South Korea to the Taylor fab to add manpower, in preparation for 2-nanometer chip mass production planned to start in 2027.
Regarding the chip foundry contract at the center of the controversy, Tesla's AI5 chip has already entered trial production at Samsung's Taylor fab, with a target of mass production in 2027. The chip's design is roughly half-mask size, with a large compute die in the center and 12 memory modules around the periphery, targeting 2,000 to 2,500 TOPS of computing power. Musk said the chip's performance is about three times that of comparable Nvidia chips, while costing only one-tenth as much. However, these figures have not yet been independently verified by third parties.
In addition, the AI6 and AI6.5 chips are expected to use TSMC's and Samsung's 2-nanometer processes, respectively. Related reports say the on-chip SRAM capacity of these two chips has been greatly increased, while memory capacity has been reduced to 144GB. This arrangement shows that Tesla, while advancing its own AI chips, is relying on multiple wafer foundry partners at the same time. For Samsung, the Taylor fab is involved in both AI5 trial production and preparations for 2-nanometer mass production.
From an industry perspective, this allegation highlights that, amid the AI computing power race, the conflict between chip designers and wafer foundries is intensifying. The sensitive manufacturing process information, audit rights, and pricing mentioned in the allegation are precisely the focal points of the dispute between the two sides. At the same time, competition for talent and regional supply chain resources is also heating up. Tesla and Samsung have neither confirmed nor denied the relevant allegations, making it difficult for outsiders to judge their authenticity, but it has already drawn attention to the terms of cooperation and competitive relationship between the two sides.
Looking ahead, whether Tesla's AI5 can enter mass production as planned in 2027, the progress of Samsung's 2-nanometer process yield ramp at the Taylor fab, and TSMC's actual role in the Terafab project will all affect the direction of this supply chain relationship. If AI6 and AI6.5 use 2-nanometer processes as expected, Tesla's demand for advanced processes and memory chips will continue to expand. The game over pricing, technical information, and talent may become the norm in the AI chip race. Whether Tesla and Samsung respond later will become an important marker for observing how the situation develops.
