NVIDIA Secures More Substrate and Packaging Capacity for AI Platforms

Unite.AI
Read full story· 3 min
NVIDIA has secured expanded substrate and advanced packaging capacity through new and amended multiyear agreements with Kinsus, Samsung Electro-Mechanics, and Siliconware Precision Industries. These expansions, focused in Taiwan and South Korea, aim to support NVIDIA's AI infrastructure platforms with production scaling from late 2027 through 2029. The deals build on longstanding partnerships and include multibillion-dollar funding commitments to enhance manufacturing capabilities for AI compute

What happened

On November 17, 2026, NVIDIA announced new and updated long-term supply agreements with Kinsus, Samsung Electro-Mechanics, and Siliconware Precision Industries. These deals will expand substrate and advanced packaging capacity in Taiwan and South Korea for NVIDIA's AI infrastructure platforms. Kinsus will increase its NVIDIA-dedicated substrate manufacturing by over 3.5 times, with production starting in the second half of 2027. The agreements include multibillion-dollar funding commitments and are designed to support production scaling from late 2027 through 2029.

Key details

  • Samsung Electro-Mechanics will upgrade existing factories in Sejong and Busan, South Korea, to support more NVIDIA accelerated-computing platforms.
  • Siliconware Precision Industries (SPIL) will add advanced-packaging capacity in Taiwan, with new lines dedicated to NVIDIA by mid-2029.
  • NVIDIA is providing multibillion-dollar funding to support these expansions.
  • Vera Rubin NVL72 and Vera CPUs began volume shipments in Q3 2026, using TSMC's 5nm and 3nm process technologies.

Why it matters

NVIDIA and its manufacturing partners will be able to produce more AI hardware, supporting customers who rely on NVIDIA's compute platforms for data centers and AI workloads. Companies in the supply chain, such as Kinsus, SPIL, and Samsung Electro-Mechanics, will receive significant investment and increased business. Organizations planning to deploy large-scale AI systems will have greater access to NVIDIA's next-generation infrastructure as production ramps up from 2027 to 2029.

More on this story


More in Chips & Compute