NVIDIA expands global supply chain to scale AI infrastructure
NVIDIA is securing multi-year agreements with key manufacturing partners to boost capacity for next-generation AI chips and hardware.
The short version
- Kinsus will expand ABF package substrate capacity for NVIDIA by more than 3.5x at its Taiwan campus.
- NVIDIA is supporting Amkor's expansion of advanced packaging in Arizona and South Korea with nearly $3 billion in funding.
- SPIL will increase contracted advanced packaging capacity for NVIDIA by mid-2029 through multibillion-dollar funding.
- Vera Rubin NVL72 and Vera CPUs began volume shipments in the third quarter using TSMC's 3nm and 5nm technologies.
Why it matters
These investments ensure that NVIDIA can meet the massive hardware demand required for the next wave of AI development. By securing long-term capacity now, the company aims to prevent supply bottlenecks as 'agentic AI' becomes more prevalent.
Securing the hardware foundation
NVIDIA recently announced a series of expanded commitments with its primary supply chain partners to support long-term growth. The company is working with several major manufacturers to increase capacity across substrates, advanced packaging, and wafer production. These moves are designed to align the supply of hardware components with the surging global demand for AI infrastructure.
The strategy involves deepening relationships with established names like Amkor Technology, Foxconn, and TSMC. By securing these partnerships, NVIDIA aims to build a manufacturing foundation that can sustain the production of its current and next-generation accelerated computing platforms. These platforms are the core engines used by companies to train and run large-scale artificial intelligence models.
Debora Shoquist, executive vice president of operations at NVIDIA, noted that the company's supply chain serves as a strategic advantage. She explained that the work with these partners builds on a manufacturing foundation to deliver the platforms customers need. This proactive approach helps ensure that the physical components required for AI do not become a bottleneck for technological progress.
Expanding substrate and packaging
A significant portion of the new agreements focuses on substrate capacity, which provides the physical base for semiconductor chips. NVIDIA reached a deal with Kinsus Interconnect Technology to expand advanced ABF package substrate manufacturing in Taiwan. This expansion is expected to increase capacity by more than 3.5x over current levels from the company's Youth campus.
Production for this new Kinsus capacity is slated to begin in the second half of 2027, with the additional capacity targeted for the third quarter of 2028. This multiyear agreement is intended to support NVIDIA’s long-term supply planning for its high-performance computing technologies. It builds on a decade-long relationship between the two companies to ensure a steady flow of essential components.
Additionally, NVIDIA and Samsung Electro-Mechanics have amended their existing agreement to improve operational capabilities at factories in South Korea. These facilities will support the production of NVIDIA’s compute and high-speed networking technologies. By combining fine-processing technology with high-reliability MLCC technology, the partners aim to supply the global market with high-performance AI platforms on time.
Multi-billion dollar investments
NVIDIA is also investing heavily in advanced packaging, which is the process of connecting multiple chips into a single functional unit. The company is supporting Amkor’s expansion of advanced packaging capacity in both Arizona and South Korea. These agreements, totaling nearly $3 billion, include prepayments to help Amkor scale its manufacturing for NVIDIA’s long-term needs.
In Taiwan, NVIDIA is providing multibillion-dollar funding to SPIL to expand its advanced packaging operations. This project will increase the contracted capacity dedicated to NVIDIA by mid-2029 and includes new production lines for various stages of fabrication. These capabilities are necessary for products that require high bandwidth, power efficiency, and complex system integration.
These investments are designed to serve multiple generations of AI platforms rather than just immediate needs. By providing upfront funding and long-term planning, NVIDIA allows its partners to advance their own packaging technologies. This collaborative growth ensures that the infrastructure can handle the increasingly demanding requirements of future AI systems.
Next-generation AI hardware
The practical results of these supply chain efforts are already appearing in the market. During the third quarter, NVIDIA began volume shipments of the Vera Rubin NVL72 and Vera CPUs. These specific products are designed for agentic AI, which refers to AI systems capable of taking independent actions to achieve complex goals.
These new products leverage TSMC’s 5nm and 3nm technologies, as well as CoWoS advanced packaging. TSMC and NVIDIA have maintained a longstanding partnership to plan for future technology requirements as new platforms come to market. This collaboration allows for the rapid transition of new designs into high-volume manufacturing environments.
Furthermore, Foxconn is deepening its collaboration with NVIDIA to support the volume ramp-up of next-generation GPU and networking platforms. Foxconn will use its integrated global production network across Asia, North America, and Europe to facilitate this growth. This global footprint is essential for distributing the massive amount of hardware required to power the next era of computing.
Scaling for the AI era
The broader scope of NVIDIA's expansion includes a 2.5x increase in substrate capacity over the next three years. The company is also doubling its semiconductor testing capacity between 2026 and 2027. These efforts are complemented by a 15x increase in SPIL’s optical-engine capacity by mid-2027, with plans to double that again by the end of 2028.
Partners like IBIDEN and KYEC are also playing critical roles in this expansion. IBIDEN is expanding its Ono Plant to meet growing demand for high-performance IC package substrates. Meanwhile, KYEC is expanding its testing operations into the United States to bring its capabilities closer to NVIDIA’s growing U.S. manufacturing supply chain.
Unimicron has also highlighted the importance of these trusted partnerships in the context of how AI reshapes industries. By investing in advanced substrate technologies, these partners are positioned to help customers unlock new opportunities. The collective goal of these multi-company commitments is to support the rapid growth of AI infrastructure on a global scale.
Sources and links
First reported by nvidianews.nvidia.com. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.