Enterprise & Industry

Nvidia invests $1.5B in Amkor to boost US AI chip packaging capacity

The deal targets a growing packaging bottleneck that threatens AI hardware supply chains.

Deep Dive

Nvidia is pouring $1.5 billion into a multiyear agreement with Amkor Technology to expand advanced semiconductor packaging and testing capacity in the United States. The partnership focuses on high-density interconnects and heterogeneous integration — combining different chip types (processors, memory, networking) into a single package — critical for next-generation AI accelerators like Nvidia's data center GPUs. Amkor already packages Nvidia's data center processors and networking chipsets. The prepayment from Nvidia allows Amkor to ramp up its Arizona facility while aligning technology roadmaps. Amkor shares surged 17% in after-hours trading following the announcement. Notably, Amkor also signed a separate 10-year US packaging deal with TSMC in June and serves AMD as well.

The chip packaging stage — assembling, connecting, and protecting individual dies into a finished system — has become a major supply bottleneck as demand for AI hardware explodes. Without sufficient packaging capacity, finished AI servers can be delayed even if chips themselves are available. Nvidia's investment helps secure dedicated US capacity for its platforms and gives Amkor funding to build facilities before customer demand fully materializes. However, enterprise buyers should not expect an immediate boost in AI server availability; building packaging lines takes time, and Amkor warned its Arizona campus could face changes in timing, costs, or specifications. The longer-term impact is a more geographically diverse supply chain, complementing Amkor's Asian operations and providing Nvidia with US-based production for its most advanced AI systems.

Key Points
  • Nvidia is providing a $1.5B prepayment to Amkor for US packaging and testing expansion, focusing on high-density interconnects and heterogeneous integration.
  • Amkor shares jumped 17% after the announcement; the company also has a 10-year US packaging deal with TSMC and works with AMD.
  • Advanced packaging is a key supply bottleneck for AI accelerators, and the expansion adds US capacity but won't immediately increase server availability.

Why It Matters

This investment secures dedicated US chip packaging capacity, easing a critical AI hardware supply bottleneck for enterprises.

📬 Get the top 10 AI stories daily