0
Samsung Electro-Mechanics Suwon business site. / Samsung Electro-Mechanics

Nvidia is participating in the semiconductor package substrate capacity expansion investment pursued by Samsung Electro-Mechanics. Samsung Electro-Mechanics secures a global AI chip leader as a long-term customer to elevate investment stability, while Nvidia reinforces its supply base for advanced substrates necessary for expanding AI infrastructure.

On the 21st, Samsung Electro-Mechanics announced that it signed a strategic business partnership with Nvidia to supply package substrates for next-generation AI semiconductors. The two companies will build dedicated production lines at the Sejong and Busan plants, supplying package substrates for AI chips and network switches over multiple years starting this year.

This joint investment takes the form of Nvidia participating in the production capacity expansion plan previously announced by Samsung Electro-Mechanics. In July, Samsung Electro-Mechanics announced plans to invest 8 trillion won in its Sejong plant and 15 trillion won in its Busan plant from this year through 2040. Under the plan, Sejong will pursue the expansion of package substrate production facilities for AI servers and technology development, while Busan will drive the enhancement of production and R&D capabilities for high-performance package substrates and multi-layer ceramic capacitors (MLCCs).

Last month, it also announced an FC-BGA (flip-chip ball grid array) capacity expansion plan injecting a total of 6.78 trillion won, including 4.27 trillion won into the Sejong plant and 2.51 trillion won into its Vietnamese subsidiary. At the time, Samsung Electro-Mechanics explained that it would ease the burden of facility investments and secure stable demand based on investment funding support and medium- to long-term volume commitments from a global customer.

The collaboration with Nvidia is a move that fleshes out this customer-linked investment strategy. Samsung Electro-Mechanics can scale up its high-value substrate business by linking production facility expansions with long-term supplies, while Nvidia preemptively secures substrate suppliers and production capacity.

The supply targets include substrates for Nvidia's computing chipsets and substrates for network switches that handle ultra-high-speed connectivity between chips within data centers. Samsung Electro-Mechanics anticipated that starting this year, supply volumes bound for Nvidia will grow by double digits annually on average, accounting for a substantial portion of its total package substrate production capacity by 2028.

Substrates for AI chips face higher manufacturing difficulty as their surface area enlarges and layer counts increase. Key technologies include fine circuitry that reduces signal loss and methods to suppress substrate warpage caused by heat. Samsung Electro-Mechanics will also apply technology embedding MLCCs and silicon capacitors inside the substrate to reduce power noise and elevate power efficiency.

Chang Duck-hyun, CEO of Samsung Electro-Mechanics, said, "We will combine our accumulated manufacturing competitiveness in advanced FC-BGA with our unrivaled high-reliability MLCC technology." He added, "We will fully execute our role as a total solution provider so that Nvidia's AI platforms can be supplied to the market in a timely manner with top-tier performance and quality."

Debora Shoquist, executive vice president of operations at Nvidia, said, "Responding to the pace of AI infrastructure buildouts requires expanding production capacity on a global scale." She added, "Through our collaboration with Samsung Electro-Mechanics, we have broadened our partner network of substrate suppliers and secured additional production capacity."

                                                                                                                Lee Ji-sun