Announcement made at Q3 earnings call
Additional support follows earlier funding for Sejong, Busan scale-up
Collaboration expands to substrate and MLCC supply
Nvidia will provide additional funding to support Samsung Electro-Mechanics' expansion of its semiconductor package substrate factories. The move is part of a strategy to secure supply ahead of a deepening shortage driven by surging AI demand, with Nvidia actively backing the Korean supplier's production buildout.
For Samsung Electro-Mechanics, the deepened partnership with Nvidia — one of the biggest spenders in the AI market — opens the door to broader growth in high-value AI semiconductor package substrates and a stronger foothold in the global market.
According to industry sources Wednesday, Colette Kress, Nvidia's chief financial officer and a widely cited candidate to succeed Jensen Huang as CEO, spoke on Huang's behalf at the company's fiscal year 2027 third-quarter earnings conference call held Tuesday (local time).
Kress said Nvidia is "working closely with partners to rapidly expand production capacity for key components, including wafers, substrates and advanced packaging," and specifically named Samsung Electro-Mechanics as a substrate partner.
The core of the announcement is a revision to the long-term AI semiconductor substrate supply agreement with Samsung Electro-Mechanics disclosed last month, further reinforcing the supply chain.
Under the updated arrangement, Nvidia plans to provide additional funding for the production facility expansions Samsung Electro-Mechanics is carrying out at its Sejong and Busan sites. Nvidia had already signaled large-scale financial support for dedicated AI semiconductor substrate production lines at both locations; the new commitment adds to that pledge.
The additional funds will go toward expanding production infrastructure and introducing cutting-edge manufacturing equipment at the two sites. Samsung Electro-Mechanics plans to build out a full mass-production system starting in 2029. For Nvidia, the arrangement allows it to secure advanced substrates — essential for AI semiconductor production — ahead of schedule and ease supply bottlenecks.
Samsung Electro-Mechanics last month broke ground on a package substrate factory for AI servers at its Sejong site, committing about 4.03 trillion won ($3.02 billion) to the project. The company is compressing the construction schedule to meet supply requests from global big-tech clients and is targeting mass production in September 2028.
Samsung Electro-Mechanics, which has planned a total investment of 8 trillion won in Sejong, also signaled it would go further, saying it will "secure additional nearby land and proceed with investment on a larger scale than originally planned" given sustained demand.
At present, only Samsung Electro-Mechanics and Japan's Ibiden are capable of supplying high-performance AI semiconductor substrates to Nvidia. High technological barriers to entry and stringent quality requirements make a rapid increase in market supply unlikely in the near term.
Nvidia had previously relied exclusively on Ibiden for AI semiconductor substrates but turned to Samsung Electro-Mechanics as the supply crunch worsened. With Samsung Electro-Mechanics now part of Nvidia's supply chain, direct competition with Ibiden is expected to intensify.
Samsung Electro-Mechanics plans to leverage its long-accumulated packaging process technology to support Nvidia's expanding AI semiconductor production.
Samsung Electro-Mechanics President Jang Deok-hyeon said the company will "combine world-class substrate microfabrication technology with high-reliability multilayer ceramic capacitor technology to ensure Nvidia's AI platform achieves peak performance and quality and reaches global markets on time."
Samsung Electro-Mechanics had earlier signed a separate supply agreement with Nvidia for multilayer ceramic capacitors for AI servers valued at 1.07 trillion won, covering a one-year period from January through December next year. The deals signal that Samsung Electro-Mechanics is cementing its position across Nvidia's broader supply chain, spanning both AI semiconductor substrates and components.
joze@heraldcorp.com
