Samsung Electronics Nears Mass Production of Tesla 'AI5' Chip... Accelerating Operations at U.S. Taylor Plant

Introduction of 2nm Process and Entry into Prototype Production Stage

Foundry Division Expected to Return to Profitability Next Year

Samsung Electronics has begun preparations for the mass production of Tesla's next-generation artificial intelligence (AI) semiconductor, the "AI5."

According to the semiconductor industry on July 13, an insider from Samsung Electronics' Foundry Business Division recently announced via LinkedIn, a social networking service, that "the Tesla-Samsung AI5 chip has completed tape-out (initial batch production)." The insider further stated, "AI5 will be produced at Samsung's Taylor plant in Texas, USA, using the company’s 2-nanometer (1nm = one billionth of a meter) process, and it will soon be installed in Tesla’s latest products." This is the first time that the specific production schedule for Samsung Electronics' Tesla AI5 chip has been disclosed externally.

Samsung Electronics Chairman Jay Y. Lee (third from left) poses for a commemorative photo with Tesla CEO Elon Musk (fourth from right) at Samsung Electronics North America Semiconductor Research Center in Silicon Valley, USA, on May 10, 2023 (local time). Photo by Yonhap News

Tape-out refers to the stage in which a fabless (semiconductor design specialist) company completes its final design and delivers the design files to the manufacturing plant. As this is the final step before mass production, the ramp-up speed at the Taylor plant is also expected to accelerate. The plant is expected to begin initial operations at the end of this year and start full-scale mass production of products for major clients, including Tesla, from next year.

Previously, Tesla CEO Elon Musk shared an update on the progress of AI5 through his social networking service X (formerly Twitter) this April, expressing gratitude to both Samsung Electronics and TSMC for their assistance in producing the chip. At that time, it was confirmed that the tape-out was for the volume handled by Taiwan's TSMC. Tesla plans to equip its self-designed AI semiconductor series (AI4 Improved, AI5, AI6, AI6.5, etc.) in robots, autonomous vehicles, and data centers, with manufacturing responsibilities split between Samsung Electronics and TSMC.

As a result, profitability at Samsung Electronics' foundry business division is expected to improve. Earlier this month, Samsung Electronics announced a provisional operating profit of 8.94 trillion won for the second quarter, with the memory division estimated to have contributed over 90 percent of the company’s total profit. In contrast, the system semiconductor and foundry division is estimated to have recorded an operating loss of around 600 billion won. However, it is widely expected that the division will return to profitability after Tesla’s product supply begins in earnest next year.

Meanwhile, Samsung Electronics signed a foundry supply contract with Tesla last year valued at approximately 22.7 trillion won for AI semiconductor production, and is also collaborating in the manufacturing of Nvidia’s autonomous driving chips and Groq’s AI semiconductors. Chairman Jay Y. Lee of Samsung Electronics, together with Han Jinman, President of the Foundry Business Division, reportedly attended the global Sun Valley Conference and discussed AI semiconductor and contract manufacturing cooperation plans with major clients.

This content was produced with the assistance of AI translation services.