NVIDIA backs major manufacturing expansion to fuel AI growth
Subscribe to our free newsletter today to keep up to date with the latest manufacturing news.
NVIDIA is expanding its global manufacturing network as surging demand for artificial intelligence technology puts pressure on the companies responsible for producing its hardware.
The chipmaker announced agreements with six supply chain partners on October 21, including a $1.1 billion commitment to support new semiconductor testing facilities in the United States. Other suppliers are planning major production increases across Asia, covering components and processes essential to manufacturing advanced AI systems.
The scale of the planned expansion is substantial. NVIDIA expects its semiconductor testing capacity to double between 2026 and 2027, while agreements with three suppliers aim to increase the availability of chip substrates by 2.5 times over the next three years. Another partner is targeting a 15-fold increase in production capacity for optical components by mid-2027.
Speaking at the company’s GTC Berlin event, NVIDIA founder and CEO Jensen Huang identified the challenge facing the industry.
“Demand is extraordinary. Capacity is the constraint.”
The announcement illustrates how the rapid growth of AI is creating manufacturing pressures far beyond the factories that produce semiconductor chips.
Why NVIDIA needs more than just chips
Much of the attention surrounding the AI boom has focused on semiconductor manufacturers and their ability to produce increasingly powerful processors. However, manufacturing those processors is only one stage in producing a finished AI computing system.
Specialist suppliers provide the materials that connect individual components, package them into working assemblies and test them before shipment. Each process requires dedicated equipment, technical expertise and manufacturing capacity.
A shortage at any stage can restrict how many complete systems NVIDIA can deliver, even if semiconductor production itself continues to increase.
Chip substrates are one example. These specialised components provide the electrical connections between semiconductor packages and the systems in which they operate.
NVIDIA is working with three manufacturers to expand substrate production: Japan’s IBIDEN, South Korea’s Samsung Electro-Mechanics and Taiwan’s Unimicron Technology Corporation.
Under an agreement announced earlier this year, NVIDIA is providing IBIDEN with a multibillion-dollar capacity reservation advance to support expansion at its Ono Plant in Japan. The project aims to triple the plant’s existing capacity, with additional production expected to begin ramping up in the first quarter of 2028.
Samsung Electro-Mechanics is also joining NVIDIA’s substrate supply chain, providing capacity from its existing South Korean operations while supporting factory upgrades and new facilities in Sejong and Busan.
Meanwhile, Unimicron is expanding substrate manufacturing in Taiwan, with additional capacity expected from the second quarter of 2027 and full capacity targeted for mid-2028.
Together, the agreements are expected to contribute to a 2.5-fold increase in substrate capacity for NVIDIA’s products over the next three years.
Manufacturing capacity is also being expanded further along the production process. Amkor Technology plans to increase its advanced packaging capacity in South Korea, targeting an increase of more than 300% in interposer wafer capacity by the end of 2028.
Interposers allow multiple semiconductor components to communicate within advanced packages, making them an important part of high-performance AI hardware.
NVIDIA commits $1.1 billion to expand US chip testing
One of the most significant developments for American manufacturing is NVIDIA’s agreement with King Yuan Electronics (KYEC).
The company is providing a $1.1 billion capacity reservation advance to support new semiconductor testing facilities and equipment in the US, with high-volume production targeted for April 2027.
Semiconductor testing is an essential part of production, helping manufacturers identify defects and verify that components meet performance and reliability requirements before they are integrated into finished systems.
NVIDIA expects its overall testing capacity to double between 2026 and 2027, supported by expansions in the US, Taiwan and Singapore. Additional testing operations in Singapore are expected to begin high-volume production in February 2027.
The American investment is particularly notable as companies across the semiconductor industry seek to strengthen domestic manufacturing capabilities.
However, the wider announcement also highlights the international nature of NVIDIA’s supply chain. Most of the planned substrate, packaging and optical component expansions are concentrated in Asia, where established manufacturers already possess the specialist facilities and expertise required.
Rather than representing a wholesale shift toward US production, NVIDIA’s strategy involves increasing capacity across multiple countries while adding new American testing capabilities.
Why NVIDIA is financing its suppliers
The agreements also reveal how NVIDIA is responding to manufacturing constraints.
Rather than simply ordering more components and waiting for suppliers to increase production, the company is committing substantial funding to secure future manufacturing capacity.
Several agreements involve capacity reservation advances, through which NVIDIA provides financial commitments to suppliers in return for access to future production.
This approach can help manufacturers justify major investments in equipment and facilities that may take years to become operational.
For suppliers, financial backing from a major customer can reduce some of the uncertainty associated with expanding production. For NVIDIA, the agreements provide greater confidence that the capacity required for future products will be available.
They could also improve supply-chain resilience by increasing production across several manufacturers and locations, reducing reliance on individual facilities.
The strategy reflects a wider challenge facing industries experiencing rapid demand growth.
When existing factories cannot produce enough to meet customer requirements, expanding capacity requires time, capital and confidence that demand will remain strong enough to justify the investment.
A customer with substantial financial resources can help accelerate that process by supporting its suppliers directly.
Optical component production set for major expansion
Another notable element of NVIDIA’s announcement concerns the technology used to move information between powerful computing systems.
As AI infrastructure grows, connecting large numbers of processors efficiently is becoming increasingly important. Moving data between components and systems requires considerable bandwidth and energy.
One potential solution is co-packaged optics, which brings optical communication components closer to the processors they serve. The technology uses light to transmit data and is intended to improve the efficiency of connections within large-scale computing infrastructure.
NVIDIA is working with Siliconware Precision Industries (SPIL) to expand manufacturing capacity for optical engines used in these systems.
SPIL is targeting a 15-fold increase in optical-engine production capacity in Taiwan by mid-2027, with plans to double that expanded capacity again by the end of 2028.
Although these targets remain subject to successful implementation, their scale demonstrates how NVIDIA is preparing for future manufacturing requirements beyond conventional semiconductor production.
It also suggests that the ability to produce specialised connectivity components could become increasingly important as AI computing systems grow larger and more complex.
What the expansion means for manufacturers
NVIDIA’s announcement offers a wider lesson for manufacturers beyond the semiconductor industry.
Rapid growth in demand does not necessarily translate into higher output if suppliers lack the equipment, production capacity or specialist expertise needed to keep pace.
Similar challenges can arise in aerospace, automotive manufacturing and other industries that depend on complex networks of component suppliers.
NVIDIA’s approach demonstrates how major customers can respond by making financial commitments that encourage suppliers to expand capacity before shortages become more restrictive.
However, significant uncertainties remain.
Much of the additional manufacturing capacity announced by NVIDIA is not expected to become available until 2027 or 2028. The production targets are forward-looking plans rather than guaranteed increases.
The company and its partners must also manage the risks associated with constructing new facilities, installing equipment and scaling technically demanding manufacturing processes.
Demand is another consideration. Suppliers committing to major expansions must balance current requirements against the possibility that future market conditions could change.
For now, the agreements show how demand for AI computing is reshaping investment decisions across the semiconductor supply chain.
NVIDIA is no longer relying solely on the ability of chip manufacturers to increase production. It is supporting expansion throughout the manufacturing processes needed to turn advanced processors into functioning AI systems.
Whether those investments deliver capacity quickly enough to meet demand will be an important test of the industry’s ability to sustain its growth.
Source:
NVIDIA News
