NVIDIA and Amkor Technology have entered a multiyear strategic agreement valued at USD 1.5 billion to develop advanced semiconductor packaging and testing technologies for next-generation artificial intelligence and accelerated-computing systems. Under the agreement, NVIDIA will provide an advance payment to support the expansion of Amkor’s semiconductor packaging capacity in the United States, including its operations in Arizona. The two companies will also coordinate their technology plans around high-density interconnects, heterogeneous integration and other packaging methods required for increasingly complex AI processors.
Key Facts
NVIDIA and Amkor have signed a USD 1.5 billion multiyear agreement.
NVIDIA will provide an advance payment to support additional U.S. packaging capacity.
The companies will jointly develop packaging and testing technologies for AI and accelerated-computing platforms.
The partnership will support Amkor’s planned expansion in Arizona.
Amkor already provides packaging services for NVIDIA data-centre processors, networking chips and accelerated-computing systems.
What Happened
Amkor announced the strategic partnership on July 23, 2026. The agreement covers the development and production of advanced chip-packaging and testing technologies that will be used in future NVIDIA computing platforms. NVIDIA’s advance payment will help Amkor increase manufacturing capacity in the United States. The investment will particularly support the company’s Arizona expansion while complementing its established semiconductor manufacturing operations across Asia.
The companies plan to align their long-term technology roadmaps. Their work will include high-density electrical connections and heterogeneous integration, a process through which different processors, memory devices and specialised chip components are combined within one package. Amkor already supports several NVIDIA products, including data-centre processors and networking chipsets. The expanded relationship is intended to bring new packaging technologies into high-volume production as demand for AI infrastructure increases.
Why It Matters
Advanced packaging has become an important part of AI chip development. Modern AI processors frequently combine computing dies, high-bandwidth memory, networking components and specialised accelerators in closely connected configurations.
These components must communicate quickly while controlling power consumption and heat generation. Advanced packaging can improve data-transfer speed, system performance, energy efficiency and the number of components that can be placed within a single computing module.
The agreement also addresses a significant weakness in the U.S. semiconductor supply chain. Although new wafer-manufacturing plants are being built in the United States, many advanced chips must still be sent abroad for packaging and final testing.
Expanding domestic packaging capacity could allow more production stages to be completed within the United States. This may reduce transportation requirements, shorten supply chains and improve manufacturing coordination between chip fabrication, packaging and system assembly.
Company Statements
NVIDIA Executive Vice President of Operations Debora Shoquist said Amkor’s global capabilities and U.S. investment are important for building resilient AI infrastructure and supporting the development of next-generation technologies.
Amkor Chief Executive Officer Kevin Engel said the partnership highlights the central role of advanced packaging in AI development. He added that the agreement will accelerate Amkor’s technology plans and support the expansion of complete packaging and testing services.
Technology Explained
What Is Advanced Semiconductor Packaging?
Semiconductor packaging protects a chip and connects it to memory, power systems, circuit boards and other processors. Traditional packaging commonly handled a single major chip, but AI systems increasingly require several different components to operate together. Advanced packaging allows these components to be placed closer to each other. This reduces the distance that information must travel and can improve computing speed while limiting energy use.
High-Density Interconnects
High-density interconnect technology creates a large number of very small electrical connections between processors, memory devices and other components. More connections can support faster data movement, which is particularly important for AI model training and inference. These workloads require large amounts of information to move continuously between computing processors and memory.
Heterogeneous Integration
Heterogeneous integration combines different types of chips within a single package. These may include graphics processors, central processing units, memory, networking components and specialised accelerators. Each component can be produced using the manufacturing process most suited to its function. The components are then integrated into a coordinated computing system through advanced packaging.
Industry Impact
AI Chip Manufacturing
The agreement could provide NVIDIA with greater access to advanced packaging capacity as it develops more complex AI computing platforms. Packaging availability has become almost as important as access to chip fabrication. A processor cannot be delivered to a customer until its computing dies, memory and supporting components have been assembled, connected and tested.
Closer cooperation between NVIDIA and Amkor may also allow packaging requirements to be considered earlier in the chip-design process. This could reduce development delays and improve the transition from engineering samples to commercial production.
High-Bandwidth Memory
AI accelerators depend heavily on high-bandwidth memory because large AI models require rapid access to data. This memory is commonly positioned close to the main computing processor through advanced packaging. The NVIDIA-Amkor collaboration is therefore expected to influence demand for memory integration, thermal-management materials, substrates, interposers and precision testing equipment. This is an industry inference based on the technical requirements of heterogeneous AI packages.
Semiconductor Testing
Testing becomes more difficult when several high-value components are combined in one package. Manufacturers must verify electrical performance, data-transfer reliability, thermal behaviour and long-term durability before a completed device can be shipped. The partnership is expected to support the development of testing methods designed specifically for next-generation AI and accelerated-computing systems.
U.S. Semiconductor Supply Chain
Amkor’s planned facility in Peoria, Arizona, will focus on advanced packaging and testing for AI, high-performance computing, communications, automotive and industrial applications. The first phase is expected to employ more than 1,300 workers.
The location was selected partly because of its proximity to semiconductor manufacturers and suppliers in Arizona. Amkor said the site will allow it to work more closely with nearby wafer-production operations and provide faster manufacturing feedback.
Market Implications
Packaging Capacity Is Becoming Strategically Important
Competition in AI semiconductors is no longer focused only on processor design or wafer manufacturing. Memory integration, advanced substrates, interconnects, cooling and final testing are increasingly determining how quickly new products can be supplied. Long-term capacity agreements may become more common as chip companies seek dependable access to these specialised production processes.
AI Infrastructure Supply Chains May Become More Regional
The agreement supports a broader move toward geographically diversified semiconductor production. NVIDIA will continue to depend on a global network of manufacturers, but additional U.S. capacity may reduce reliance on a limited number of overseas packaging locations. A more regional supply chain could improve resilience, although costs, skilled-worker availability, manufacturing yields and construction schedules will remain important factors.
Chiplet-Based Designs Could Gain Wider Adoption
Heterogeneous integration allows semiconductor companies to combine smaller specialised chiplets instead of producing every function on one large piece of silicon. This approach can support greater design flexibility and may allow different components to be upgraded independently. However, it also increases the importance of precise packaging, system-level testing and reliable interconnection technologies.
Equipment and Materials Suppliers May Benefit
The expansion is expected to create additional demand for packaging equipment, testing systems, substrates, interposers, bonding materials, thermal-management products and factory-automation systems. These opportunities will depend on the timing of Amkor’s capacity expansion and the specific production technologies selected by the companies.
Expert Analysis
The most important part of the agreement is not only its USD 1.5 billion value. Its wider importance comes from the close connection between chip design, packaging technology and manufacturing capacity.
AI processors are becoming larger and more complex, but technical progress cannot be achieved through computing dies alone. Performance increasingly depends on how effectively processors, memory and networking components are connected inside the final package.
The advance-payment structure also indicates that NVIDIA is prepared to support manufacturing capacity before it becomes fully available. This approach can provide Amkor with greater financial certainty when purchasing equipment, developing processes and preparing production lines.
For Amkor, the agreement provides a major customer commitment and supports the expansion of its U.S. operations. For NVIDIA, it provides greater influence over packaging development and additional capacity for future AI products.
However, the agreement should not be interpreted as immediate production capacity. Facility construction, equipment installation, worker training, process qualification and production-yield improvement will determine when the additional capacity becomes commercially available.
Background and Context
Amkor has been expanding its relationships with major semiconductor companies as advanced packaging becomes more important for data-centre and AI processors. The company announced a long-term U.S. packaging partnership with TSMC in June 2026 and has also been working with AMD on advanced packaging for its processors.
Amkor’s Arizona facility is being developed as part of efforts to strengthen domestic semiconductor production. The site will provide packaging and testing services for chips used in AI, high-performance computing, mobile devices, communications, automotive systems and industrial equipment. The NVIDIA agreement strengthens the commercial role of the Arizona expansion and connects the facility directly with future AI infrastructure development.
