Tesla will put about 1.1 trillion won, roughly $830 million, into a new Samsung Electro-Mechanics production line in Sejong, South Korea, Maeil Business Newspaper reported on 22 October 2026, citing industry sources. The line will make FC-BGA substrates for the AI chips in Tesla's self-driving cars and its Optimus robot. In return, Tesla gets guaranteed volumes under a five-year supply contract running from 2029 to 2033. Reuters carried the report the same day. Neither Tesla nor Samsung Electro-Mechanics has confirmed it.

What Maeil reports

According to the paper, Tesla's money covers part of the building extension and equipment at Sejong. In late September, Samsung Electro-Mechanics said it would invest 4.27 trillion won there through May 2028, its largest investment in a single product. It also said unnamed big-tech customers would fund part of the cost and guarantee medium- to long-term purchase volumes, the Korea Herald reported at the time. Maeil now names two of those customers.

Tesla Nvidia
Reported investment ₩1.1 trillion (~$830M) ₩2.6 trillion (~$1.97B)
Plants Sejong Sejong and Busan
Substrates for AI chips in self-driving cars and Optimus AI compute chips and data-centre networking
Supply period 2029–2033 Multi-year, dates not given
Officially confirmed No Partnership yes, amount no

Nvidia announced its part at GTC Berlin on 21 October. It said the collaboration supports new factories in Sejong and Busan but gave no figure; Maeil says its own reporting confirmed the 2.6 trillion won. The Sejong building is due to be finished by the end of 2027, with equipment and trial runs completed in the first half of 2028 and volume production from the second half.

What an FC-BGA substrate is

A chip die is not soldered straight onto a circuit board. It is bonded to a substrate, a layered board that fans its thousands of connections out to the board and carries power in. The FC-BGA type is used for AI accelerators, graphics chips and CPUs. Only a few companies make the high-end version: Japan's Ibiden, which holds more than 70% of that market by industry estimates cited by Maeil, Taiwan's Unimicron and Samsung Electro-Mechanics. Nvidia has substrate agreements with all three and plans to raise its substrate capacity 2.5 times over the next three years.

Why Tesla would pay for someone else's factory

Tesla designs its own AI chips and has others manufacture them. Its AI5 is in trial production at Samsung's fab in Taylor, Texas and is also due from TSMC in Arizona. Musk has said AI5 will go into Optimus and Tesla's training clusters rather than its cars, so a 2029–2033 contract points to later chips, although Maeil does not name one.

The reason to commit now is scarcity. Maeil calls the substrate the AI industry's new supply-chain bottleneck, and Nvidia has paid capacity-reservation advances to Ibiden and the chip tester KYEC for the same reason. Tesla knows what a bottleneck costs. Its third-quarter letter said battery availability limited vehicle output while the order backlog was the largest in four years. On the same 21 October earnings call, Tesla cut its 2026 capital spending guidance to about $20 billion.

It is the second report in a week tying Tesla's chip plans to Korean suppliers, after The Elec's account of Terafab staff lining up Korean equipment makers.

What is not known

Tesla has not commented, and Samsung Electro-Mechanics has not named its customers. Maeil does not say when Tesla pays or whether the money is a prepayment, a deposit or something else. Nothing changes for cars on the road today, which run on HW3 or AI4. A disclosure by either company, or a line in Tesla's next quarterly filing, would settle it.