NVIDIA and Partners Boost Manufacturing for AI Computing Systems
NVIDIA is working with several manufacturing partners to add capacity for the substrates, advanced packaging, testing, and optical components used in its AI systems. The company plans to increase substrate capacity for its products by 2.5 times over three years, with partners including Ibiden, Samsung Electro-Mechanics, and Unimicron expanding plants. NVIDIA says demand for its accelerated computing and AI factory platform is outpacing available production capacity.
NVIDIA’s partners span substrate production, packaging, testing and optical engines. Ibiden’s Japanese plant is slated to reach three times its current capacity, aided by a multibillion-dollar advance, with added output starting in early 2028. Samsung Electro-Mechanics brings existing South Korean capacity and plans upgrades plus new factories in Sejong and Busan.
Unimicron’s Taiwan substrate expansion is expected to add capacity from spring 2027 and reach full output by mid-2028. Amkor aims to more than triple interposer wafer capacity in South Korea by late 2028. Testing capacity is projected to double from 2026 to 2027, while SPIL targets a 15-fold optical-engine increase by mid-2027.
The buildout could affect cloud providers, AI developers and manufacturers by making advanced AI systems more available, though actual effects depend on demand and execution. Regions hosting new plants may see investment and specialized jobs, while local infrastructure and energy needs may rise. Consumers and businesses could eventually see broader access to AI services, but capacity gains may not directly translate into lower prices or immediate availability.