BAKU, Azerbaijan, November 18. NVIDIA is expanding manufacturing capacity through agreements with six supply chain partners - Amkor Technology, Foxconn, Kinsus Interconnect Technology, Samsung Electro-Mechanics, Siliconware Precision Industries (SPIL) and Taiwan Semiconductor Manufacturing Company (TSMC) - as it prepares for growing demand for its artificial intelligence (AI) and accelerated computing platforms.
According to NVIDIA, the commitments cover several stages of semiconductor production, from package substrates and advanced packaging to wafer manufacturing and contract production. They include both new agreements and expanded cooperation with existing suppliers.
One of the largest capacity increases concerns Kinsus, which plans to expand its advanced ABF package substrate manufacturing capacity dedicated to NVIDIA by more than 3.5 times current levels. The expansion will take place at Kinsus' Youth campus in Taoyuan, Taiwan. Production is expected to begin in the second half of 2027, with the additional capacity targeted for the third quarter of 2028.
NVIDIA is also deepening its cooperation with Samsung Electro-Mechanics. The companies have amended their existing substrate supply agreement to improve production capabilities at facilities in Sejong and Busan, South Korea, supporting the manufacture of substrates for NVIDIA's computing and high-speed networking technologies.
Advanced packaging is another major area of investment. NVIDIA's agreements with Amkor, announced in July and October 2026, provide for the expansion of manufacturing capacity in Arizona and South Korea. The commitments total nearly $3 billion and include prepayments to support Amkor's investment in facilities serving NVIDIA's long-term production needs.
In Taiwan, NVIDIA and SPIL are expanding their advanced packaging operations under a multiyear agreement backed by multibillion-dollar funding from NVIDIA. The project will add production lines for bumping, interposer wafer fabrication, assembly and testing. The additional contracted capacity dedicated to NVIDIA is expected to become available by mid-2029.
The cooperation with TSMC is focused on manufacturing technologies for NVIDIA's next-generation platforms. The Vera Rubin NVL72 systems and Vera CPUs, designed for agentic AI, use TSMC's 5-nanometer and 3-nanometer process technologies as well as its Chip-on-Wafer-on-Substrate (CoWoS) advanced packaging. NVIDIA said the first volume shipments of these products took place in the third quarter.
Foxconn, meanwhile, is working with NVIDIA on next-generation GPU and networking platforms. Its manufacturing network across Asia, North America and Europe is expected to support the ramp-up of production for future AI infrastructure.
NVIDIA Executive Vice President of Operations Debora Shoquist described the supply chain as a strategic advantage, saying the expanded cooperation would strengthen manufacturing capacity across components, advanced packaging and system production.
The latest agreements add to a broader expansion plan announced in October 2026, which also involved IBIDEN, King Yuan Electronics (KYEC) and Unimicron, alongside several partners included in the latest commitments.
That earlier package envisaged a 2.5-fold increase in substrate capacity over three years, expanded advanced packaging operations and a doubling of semiconductor testing capacity between 2026 and 2027. It also included plans to increase SPIL's optical-engine capacity 15-fold by mid-2027, with a further doubling targeted by the end of 2028.
