Nvidia is pouring $1.5 billion into one of the least visible but increasingly critical stages of AI chip production. The company has signed a multiyear agreement with Amkor Technology to expand advanced semiconductor packaging and testing capacity in the United States, a move aimed at easing a stubborn bottleneck in the AI hardware supply chain.
Announced in late July 2026, the partnership will support Amkor's Arizona expansion and fund new packaging and testing technologies for Nvidia's future AI and accelerated computing platforms. For engineers and executives tracking the latest AI hardware developments, the deal is a reminder that the race to build AI infrastructure is no longer just about designing faster silicon — it is about assembling that silicon into finished systems at scale.
Why packaging is the new bottleneck
Packaging is the process of enclosing semiconductors in metal and plastic casings. It sounds mundane, but it is a critical step in chip manufacturing: it protects the fragile silicon from damage and allows it to be integrated with other computer parts, including the memory and networking components that modern AI accelerators depend on.
As AI chips have grown more complex, packaging has become a decisive constraint. Today's top accelerators combine logic dies, high-bandwidth memory, and interconnects inside a single package using techniques like high-density interconnects and heterogeneous integration — the practice of combining different types of chips and components within one module. Getting that assembly right, at volume, is now one of the hardest problems in the industry, and the majority of advanced packaging today is performed offshore.
What the deal covers
Under the agreement, Nvidia will provide a prepayment that allows Amkor to expand its U.S. capacity while the two companies align their technology roadmaps. According to TechRepublic, which reported the deal, the partnership will focus on high-density interconnects and heterogeneous integration for Nvidia's future AI platforms.
Amkor already provides packaging services for Nvidia products, including data center processors, networking chipsets, and accelerated computing systems. The expanded partnership is intended to bring future packaging and test technologies into large-scale production as demand for AI infrastructure keeps climbing.
Neither company has detailed a precise timeline for when the added capacity will come online, but the direction is clear: more of the work that turns raw Nvidia silicon into deployable AI systems will happen on U.S. soil.
An Arizona anchor
The deal brings one of the AI industry's most consequential players directly to Amkor, a Tempe, Arizona-based company that is already building a major footprint in the Phoenix area. As KJZZ reported, the partnership centers on Amkor's Arizona operations, including a planned packaging and test facility in north Peoria — a roughly $2 billion complex whose conceptual designs have been public for months.
For Arizona, the agreement deepens the state's emergence as a semiconductor hub, adding high-value packaging and testing work to a corridor already anchored by nearby fabrication investment. For Amkor, Nvidia's prepayment offers a measure of demand certainty as it commits capital to capacity that can take years to bring online.
The onshoring angle
The partnership is also a concrete step toward onshoring more of the chip manufacturing process in the United States. Because the bulk of advanced packaging is currently performed overseas — concentrated in East Asia — domestic capacity has lagged even as billions of dollars have flowed into front-end fabrication.
Nvidia's bet is that closing that gap matters. AI accelerators are only useful once they are packaged, tested, and integrated into the servers that fill data centers, and any shortage in those downstream steps can throttle the entire pipeline regardless of how many chips come off a fab line. Amkor says bringing packaging and testing to Arizona will help build a more resilient supply chain for AI chips.
The bigger picture
The Amkor agreement fits a broader pattern. As generative AI workloads have surged, Nvidia has moved aggressively to secure every link in its supply chain, from memory and networking to the substrates and assembly that turn its designs into shipping products. Packaging has drawn particular attention because it is where raw manufacturing capacity is scarcest relative to demand.
For now, the companies are presenting the deal as a long-term alignment rather than an overnight capacity boost. Nvidia's $1.5 billion prepayment funds expansion that will roll out over multiple years, and the benefits to supply will accumulate as new lines come online. But the strategic signal is unambiguous: the part of the AI chip pipeline that most needs shoring up is finally getting serious American investment.
Stay Ahead of AI
The Nvidia-Amkor partnership will roll out new packaging and testing capacity in Arizona over the coming years, backed by a $1.5 billion prepayment. For more on this story and the wider artificial-intelligence landscape, keep up with breaking AI news as it happens.
Read more AI news →
