Nvidia Sets 2028 Capacity Targets Across Six Chip Supply Partners

Nvidia outlined capacity expansion targets with six supply-chain partners on Oct. 21, covering substrates, advanced packaging, testing and optical engines to meet demand for accelerated computing and AI factory platforms.

The partners are 欣興, 京元電子, 矽品, Ibiden, Samsung Electro-Mechanics and Amkor. Their projects span the United States, Taiwan, Japan and South Korea.

Nvidia aims to increase substrate capacity 2.5 times over three years. 欣興’s Taiwan expansion is expected to begin adding capacity in the second quarter of 2027 and reach full utilization by mid-2028. Ibiden plans to expand capacity at its Ono plant to three times current levels, with new capacity expected in the first quarter of 2028. Samsung Electro-Mechanics is upgrading facilities and building new ones in Sejong and Busan.

The plan also includes a U.S. testing plant backed by a $1.1 billion prepayment to 京元電子. The U.S. testing plant is part of Nvidia’s wider chip-testing buildout. Mass production there is targeted for April 2027.

In Taiwan, 矽品 is targeting optical-engine capacity 15 times current levels by mid-2027 and about 30 times current levels by the end of 2028. The expansion involves co-packaged optics (CPO), which packages optical components together with switch chips instead of relying on traditional pluggable optical modules.

Amkor is targeting an increase of more than 300% in South Korean interposer wafer capacity by the end of 2028.