Nvidia expands supplier roster to accelerate AI infrastructure output
Nvidia is stepping up its efforts to boost manufacturing capacity to meet the explosive growth of artificial intelligence infrastructure. The chipmaker has inked deals with six suppliers to increase the supply of key components for systems that train and run AI models, according to a press release on Oct. 21.
The agreements cover various parts of the supply chain including semiconductor substrates and advanced chip packaging and testing. Some projects are about multiplying existing production capacity over the next three years and some are about making large investments in new facilities and equipment in Asia and the United States.
Nvidia has signed agreements with Ibiden, Samsung Electro-Mechanics and Unimicron Technology Corp. to expand capacity. In three years, the three suppliers are expected to increase their manufacturing capacity by a factor of 2.5 overall.
The company has also signed agreements with Amkor Technology, King Yuan Electronics Co. (KYEC) and Siliconware Precision Industries to boost the production capacity needed for AI computing systems.
In its announcement, Nvidia founder and CEO Jensen Huang put the scale of the challenge in perspective. He said the current infrastructure expansion is the largest in human history and that extraordinary demand has made manufacturing capacity a major constraint.
Semiconductor substrates to be prioritized for growth
Nvidia has three partners that produce semiconductor substrates: Ibiden, Samsung Electro-Mechanics and Unimicron. Specialized circuit boards that hook up processors to a computer’s main circuit board are key pieces in making advanced chips.
Samsung Electro-Mechanics said it will upgrade existing facilities and build new plants in Sejong and Busan in South Korea. Under the agreement, the company expects to dedicate a significant part of its total annual package substrate production capacity to Nvidia by 2028.
Unimicron to expand in Taiwan Manufacturing expansion work is planned to start in the second quarter of 2027, with the additional capacity expected to be fully operational by mid-2028.
Nvidia is helping Ibiden build out in Japan with a multi billion dollar capacity reservation. The 2026 accord signed earlier is designed to triple the capacity of the plant involved in the project. Production at the expanded facility is expected to start ramping up in the first quarter of 2028, said Ibiden.
The commitments are intended to secure a more consistent supply of key components as Nvidia and its customers expand their AI computing infrastructure.
Advanced packaging and testing capacity is also expected to grow
Nvidia’s vision isn’t just about the substrates, but also the processes for assembling and testing more complex semiconductor systems.
Amkor Technology will boost output at its South Korean plants that combine chips into connected units using advanced packaging. Amkor has agreed to triple its wafer processing capacity for interposer-containing wafers by the end of 2028, according to a deal with Nvidia.
An interposer is a connecting layer that enables chips within the same package to communicate with each other. The technology is the key to bringing together the components required for advanced AI computing.
The King Yuan Electronics Co. deal also has Nvidia funding more testing capacity in the United States. The chipmaker said it will spend $1.1 billion on new test facilities and equipment, with mass production expected to begin in April 2027.
Siliconware Precision Industries is also planning to expand its manufacturing capacity in Taiwan by 15 times by mid-2027. The company aims to double that capacity again by the end of 2028.
This expansion will allow us to build advanced optical engines that will be used to meet the bandwidth and power efficiency requirements of traditional artificial intelligence systems. These components are becoming increasingly relevant, because computing infrastructures need faster and more efficient data transmission.
But supply constraints still cloud Nvidia’s outlook
Nvidia is trying to keep up with demand for its processors and the infrastructure around them, and the new deals are part of that.
In August, Huang told investors on the company’s earnings call that supply constraints were impacting Nvidia’s outlook for the next fiscal year. The constraints echoed through the company’s supply chain, a product of the global growth of data centers designed to support AI workloads.
Nvidia’s high-end processors are designed to run big AI models. Its customers include big tech companies like Microsoft, Google and Amazon. These companies operate huge networks of data centers that require vast amounts of computing power to train and run artificial intelligence applications.
The speed of deployment is increasingly dependent on the availability of specialized components, packaging services and testing facilities with investment in AI infrastructure accelerating.
Nvidia is working to secure manufacturing commitments in a number of countries and production stages to relieve these bottlenecks and increase its ability to deliver the hardware that its customers require.
The projects will come online gradually between 2027 and 2028 so it will take a while to see the full benefit. Still, the size of the deals underscores how chipmakers and their partners are reshaping their industrial capabilities to meet the demands of the growing AI market.
