Nvidia has signed multiyear agreements with Samsung Electro-Mechanics to supply advanced semiconductor substrates for its AI processors and networking chips, with LG Innotek separately working on sensing systems for Nvidia’s humanoid robots. Both were reported by The Korea Herald.
Substrates provide the electrical connection between a semiconductor chip and a circuit board. They are a routine component, but for high-performance AI processors they are one of the parts that constrain how fast production can scale.
What the deals actually say
Samsung Electro-Mechanics said it will supply substrates for Nvidia’s AI computing processors and high-speed networking chips, and that the two companies will jointly invest in expanding substrate production facilities in Sejong and Busan. Supply begins this year, with additional capacity planned through factory upgrades and new facilities.
LG Innotek said on October 23 that it is working with Nvidia to develop sensing systems for humanoid robots, applying its existing camera and sensing expertise to a new product category.
Nvidia said agreements with Samsung Electro-Mechanics, Japan’s Ibiden and Taiwan’s Unimicron will increase substrate production capacity for its products by 2.5 times over the next three years. Nvidia-related production is expected to account for a significant portion of Samsung Electro-Mechanics’ annual package substrate capacity by 2028, though specific volumes and financial terms were not disclosed.
Why component suppliers are suddenly strategic
The supplier deals show where the bottleneck sits. Nvidia designs the chips, but the packaging and substrate capacity that turns a design into a shipping product lives with a small number of component makers. Adding that capacity is the work that makes a chip launch possible.
For consumers the practical question is different: this kind of deal is about future supply, not about any product you can buy today. It does not change the price or availability of existing graphics cards or AI hardware.
It also follows a pattern. Nvidia has been broadening where its hardware runs, and our coverage of RTX Spark PCs looked at the push to run local AI on ordinary Windows machines.
What we did not do
We did not independently confirm the 2.5-times capacity figure or the 2028 timeline beyond Nvidia’s and Samsung Electro-Mechanics’ own statements as reported. The LG Innotek sensing announcement was reported on October 23, a day after the Samsung agreement.
Sources
- The Korea Herald — Nvidia taps Korean component makers for AI chips, robots
- DIGITIMES — Nvidia lines up six suppliers to expand substrate, packaging, and testing capacity
- Honest Fred — Nvidia RTX Spark PCs Bring Local AI to Windows
- Honest Fred — OpenAI Raises $30 Billion as Anthropic Races Toward IPO
