Samsung Electro-Mechanics just got two of the most demanding customers in tech to help pay for its factories. Tesla and Nvidia are putting a combined 3.7 trillion won, approximately $2.8 billion, into the South Korean company’s chip substrate business.
The money is going into a very specific product: advanced flip-chip ball grid array substrates, known as FC-BGA. These are the parts AI semiconductors need, and they are increasingly hard to come by.
Who is paying for what
The split is lopsided. Nvidia is contributing 2.6 trillion won, while Tesla is putting in 1.1 trillion won.
Tesla’s money is tied to substrates for its upcoming autonomous vehicles and its Optimus humanoid robots. The deal comes with a long-term FC-BGA supply contract covering AI chips from 2029 through 2033.
Nvidia’s larger check is attached to a multi-year partnership. That arrangement targets substrates for AI accelerators and data center networking switches, the hardware that keeps massive AI clusters running and talking to each other.
Nvidia publicly backed the announcement at its GTC 2026 event in Germany on October 21.
Most of the funding will support dedicated production lines at Samsung Electro-Mechanics’ facility in Sejong. The company also plans to expand capacity in Busan.
The timeline is long, on purpose
Construction of the new facilities is expected to finish by the end of 2027.
Mass production is anticipated in the second half of 2028. Tesla’s first supplies are set to arrive in 2029, with the contract running to 2033.
The investment also fits into a bigger plan. Samsung Electro-Mechanics had already disclosed a 4.27 trillion won investment program aimed at expanding production capacity to meet rising demand for high-end substrates.
The Tesla and Nvidia commitments line up with that earlier plan. Rather than funding all of the expansion alone, the company now has two anchor customers sharing the bill.
Why customers are funding suppliers
Large tech companies are no longer content to simply place purchase orders and hope the parts show up. Instead, they are putting capital directly into suppliers in exchange for guaranteed capacity.
For Tesla, the logic is tied to bets that have not fully materialized yet. Optimus robots and next-generation autonomous vehicles both depend on serious onboard AI computing, and that computing depends on chips sitting on substrates like these.
Locking in supply through 2033 suggests Tesla expects those products to need meaningful chip volumes.
What this means
For Samsung Electro-Mechanics, the deal does two things at once. It brings in outside capital for expensive new production lines, and it secures long-term customers to fill them.
There are risks baked into the timeline. Construction has to finish on schedule by the end of 2027, and mass production has to ramp in the second half of 2028 for the supply commitments to land as planned.
The milestones to watch are straightforward: progress at the Sejong site, the Busan expansion, and whether production begins on time in 2028. Tesla’s first deliveries in 2029 will be the clearest sign of whether this long-range bet is paying off for all three companies.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.