13 Views

NVIDIA and SPIL to Triple Advanced Packaging Capacity in Taiwan by 2029

Publish Date: | Author:

CoWoS Package feature

Nov-18-2026 — NVIDIA and Siliconware Precision Industries (SPIL) have announced a multibillion-dollar, multi-year agreement to significantly expand advanced semiconductor packaging capacity in Taiwan, as demand for AI accelerators continues to place pressure on the global semiconductor supply chain.

 

SPIL, a subsidiary of ASE Technology Holding, plans to triple its contracted advanced packaging capacity dedicated to NVIDIA by mid-2029 under the new agreement.

 

The investment will establish additional high-volume production capacity supporting NVIDIA’s current and next-generation accelerated computing platforms.

 

The agreement highlights the growing strategic importance of advanced packaging as AI processors become larger, more complex and increasingly dependent on high-bandwidth memory, silicon interposers and advanced chip integration technologies.

 

NVIDIA Secures More Advanced Packaging Capacity

Demand for AI infrastructure has created significant capacity requirements throughout the semiconductor manufacturing chain, extending beyond leading-edge wafer fabrication into substrates, semiconductor testing and advanced packaging.

 

NVIDIA has therefore been working with manufacturing partners to expand capacity across multiple stages of its supply chain.

 

The agreement with SPIL represents one of NVIDIA’s largest commitments to advanced packaging capacity in Taiwan.

 

Under the multi-year agreement, NVIDIA is providing multibillion-dollar funding supporting SPIL’s expansion.

 

SPIL plans to develop new production lines providing turnkey manufacturing capabilities covering several critical packaging processes, including:

  • Wafer bumping

  • Interposer wafer fabrication

  • Semiconductor assembly

  • Advanced packaging

  • Semiconductor testing

 

The company will also upgrade manufacturing facilities and increase automation. These investments are intended to increase production throughput, shorten manufacturing lead times and improve supply-chain resilience.

 

Advanced Packaging Becomes Critical for AI Chips

Advanced packaging has evolved from a relatively downstream manufacturing process into one of the most strategically important technologies in high-performance semiconductor production.

 

Modern AI accelerators increasingly combine several semiconductor dies within a single package.

 

These systems can include GPUs or AI accelerators, I/O dies, silicon interposers and multiple stacks of high-bandwidth memory.

 

Integrating these components efficiently is essential for achieving the enormous computing performance and memory bandwidth required by AI training and inference systems.

 

As a result, advanced packaging capacity has become one of the major constraints affecting the industry’s ability to increase production of high-performance AI processors.

 

The rapid growth of AI infrastructure has consequently triggered substantial investment from foundries, OSAT companies, substrate manufacturers and semiconductor equipment suppliers.

 

CoWoS Capacity Remains a Major AI Supply-Chain Constraint

One of the technologies at the center of this expansion is chip-on-wafer-on-substrate (CoWoS), the advanced packaging architecture widely associated with high-performance AI processors.

 

CoWoS allows multiple advanced semiconductor components to be integrated within a single package while providing extremely high-bandwidth connections between processors and memory.

 

The technology has become particularly important for AI accelerators using high-bandwidth memory.

 

Industry demand for CoWoS and related packaging technologies has increased dramatically as companies including NVIDIA, AMD and other AI semiconductor suppliers increase production of accelerator platforms.

 

The industry is consequently investing billions of dollars in additional advanced packaging infrastructure in an effort to prevent packaging capacity from becoming a limiting factor for AI chip production.

 

Taiwan Remains at the Center of AI Semiconductor Manufacturing

The SPIL investment also reinforces Taiwan’s position at the center of the global AI semiconductor manufacturing ecosystem.

 

NVIDIA works with several Taiwanese semiconductor companies across its supply chain, including TSMC for semiconductor manufacturing and advanced packaging, as well as suppliers covering substrates, testing, electronics manufacturing and system assembly.

 

SPIL’s expanded facilities will provide additional advanced packaging capacity alongside this broader semiconductor manufacturing ecosystem.

 

The company said the expansion is designed specifically for high-performance semiconductor products requiring demanding levels of computing performance, bandwidth, power efficiency and system integration.

 

This should allow SPIL to support increasingly complex generations of AI processors as packaging architectures continue to evolve.

 

NVIDIA Expands Its Semiconductor Supply Chain

The SPIL agreement forms part of a much broader NVIDIA supply-chain expansion.

 

NVIDIA has recently announced additional manufacturing commitments involving semiconductor companies across packaging, substrates, testing and system manufacturing.

 

The company is working with suppliers including SPIL, Amkor Technology, TSMC, Kinsus, Samsung Electro-Mechanics and Foxconn to increase manufacturing capacity for accelerated computing platforms.

 

For NVIDIA, securing wafer capacity alone is no longer sufficient.

 

The enormous scale of expected AI infrastructure deployment means the company must secure capacity throughout the manufacturing chain — from advanced wafers and substrates to packaging, testing and complete AI systems.

 

Advanced packaging is increasingly one of the most important elements of that strategy.

 

SPIL Expands Its Role in the AI Semiconductor Supply Chain

The latest agreement also significantly expands SPIL’s role within NVIDIA’s manufacturing ecosystem.

 

SPIL is already involved in semiconductor packaging and testing and has separately been expanding its manufacturing capabilities for optical engines used in co-packaged optics.

 

In October 2026, SPIL and NVIDIA announced plans to increase SPIL’s optical-engine manufacturing capacity by approximately 15 times by mid-2027, followed by another planned doubling by the end of 2028.

 

Those optical technologies are expected to become increasingly important as AI data centers require higher-bandwidth and lower-power communication between GPUs, switches and computing systems.

 

Combined with the latest advanced packaging investment, SPIL is positioning itself across several critical manufacturing technologies required for future AI infrastructure.

 

Advanced Packaging Becomes a Strategic Semiconductor Technology

The NVIDIA-SPIL agreement illustrates a broader transformation taking place within semiconductor manufacturing.

 

Historically, much of the industry’s competitive focus centered on transistor scaling and access to increasingly advanced process nodes.

 

While leading-edge wafer manufacturing remains critical, AI computing is increasingly shifting attention toward another challenge: efficiently connecting multiple advanced semiconductor dies within a single system.

 

Advanced packaging therefore plays an increasingly important role in determining system performance, memory bandwidth, power consumption and ultimately the scalability of AI computing platforms.

 

As AI accelerators continue to become more complex, investment in technologies including interposers, high-bandwidth memory integration, wafer bumping, advanced substrates and chiplet packaging is likely to continue increasing.

 

For NVIDIA, the multi-year SPIL agreement provides additional manufacturing capacity required to support this growth.

 

For SPIL and parent company ASE Technology Holding, it represents an opportunity to strengthen their position within one of the fastest-growing areas of the global semiconductor industry.

Logo Image
Privacy Overview

This website uses cookies so that we can provide you with the best user experience possible. Cookie information is stored in your browser and performs functions such as recognising you when you return to our website and helping our team to understand which sections of the website you find most interesting and useful.