Amkor Technology has entered a multiyear agreement with Nvidia worth $1.5 billion to expand advanced semiconductor packaging and test capacity in the United States. Under the arrangement, Nvidia will make a prepayment to support the expansion of Amkor’s US advanced packaging operations, including capacity in Arizona. The companies will jointly develop packaging and testing technologies for Nvidia’s AI and accelerated-computing platforms, with a focus on combining different types of chips in a single package.
Advanced packaging has become a critical bottleneck in the AI infrastructure supply chain. As AI accelerators grow more complex, chipmakers increasingly depend on packaging technologies that can connect compute dies, memory, interconnects and supporting components into dense, high-performance modules. The capacity required for these products is different from conventional semiconductor assembly and is closely tied to the ability of AI hardware vendors to ship data-centre processors at scale. Amkor already supplies advanced packaging for Nvidia’s product portfolio, including data-centre processors, and the expanded deal is aimed at bringing new packaging technologies to market as AI infrastructure demand grows.
Amkor’s shares rose 17% in extended trading after the announcement. The deal follows Amkor’s 10-year partnership with Taiwan Semiconductor Manufacturing Co., signed in June, to enhance semiconductor packaging capabilities in the United States. Amkor is also working with Advanced Micro Devices to package the company’s chips. Taken together, these relationships show how outsourced semiconductor assembly and test providers are becoming strategic partners in the AI hardware race, not merely downstream manufacturing vendors.
The Nvidia-Amkor agreement adds another layer to the global reconfiguration of semiconductor capacity around AI demand, supply-chain resilience and geography. For Indian cloud operators, data-centre developers, AI startups, GCCs and enterprise buyers, the near-term impact is indirect but important: accelerator availability, delivery timelines and pricing are increasingly shaped by packaging capacity as much as by chip design. India’s AI infrastructure buildout remains dependent on global semiconductor supply chains for high-end GPUs and related systems. Expanded advanced packaging capacity can help relieve constraints over time, but it also shows how control over AI compute supply is concentrating around a small number of vendors, foundries and packaging partners. That concentration will continue to influence procurement planning for Indian organisations scaling large-model training, inference and enterprise AI workloads.
