Samsung Electro-Mechanics and NVIDIA will expand South Korean production of ABF package substrates for AI computing and networking equipment, adding capacity to NVIDIA’s AI infrastructure supply chain.
Samsung Electro-Mechanics said Oct. 21 that the multiyear agreement covers dedicated manufacturing at its sites in Sejong and Busan. The substrates will serve NVIDIA compute chipsets and high-speed networking switches.
Package substrates connect semiconductor chips to circuit boards and carry signals between them. Samsung said the partnership supports its expansion in the high-value AI semiconductor package-substrate market and will help NVIDIA secure a stable supply for AI chip production. Samsung expects shipments to NVIDIA to post double-digit annual growth over several years, beginning in 2026. By 2028, it projects NVIDIA-allocated production will account for a significant share of its annual package-substrate capacity, but it did not give an exact share.
CNA reported Oct. 22, citing a Maeil Business Newspaper report based on industry sources, that Tesla and NVIDIA would contribute 1.1 trillion won and 2.6 trillion won, respectively, toward production lines in Sejong. The report put the combined figure at 3.7 trillion won ($2.8 billion). The companies have not confirmed the reported amount for the Samsung project or Tesla’s role.
CNA’s report described production lines in Sejong. Samsung’s and NVIDIA’s announcements identify dedicated facilities in both Sejong and Busan, but do not say whether the reported investment covers one site or both.
NVIDIA said its broader substrate-capacity plan, involving Samsung Electro-Mechanics, IBIDEN and Unimicron, is expected to raise capacity for its products to 2.5 times current levels over the next three years. NVIDIA CEO Jensen Huang said, “Demand is extraordinary. Capacity is the constraint.” Deliveries under the Samsung agreement are scheduled to begin in 2026.
