Nvidia has reached agreements with supply chain partners at the GTC Berlin conference to expand production capacity for AI accelerators and computing systems. The company will provide funding for new facilities and equipment to address current bottlenecks.
Over the next three years, Nvidia aims to increase substrate production capacity by 2.5 times. Agreements involve manufacturers in Japan, South Korea, and Taiwan, with Nvidia providing advance payments to secure future output.
Nvidia is investing several billion dollars in the Ibiden plant in Ono, Japan, to triple its capacity by 2028. Samsung Electro-Mechanics will also contribute through its existing and expanded facilities in South Korea.
In partnership with Amkor, Nvidia plans to more than triple its interposer wafer production by late 2028 and double semiconductor testing capacity between 2026 and 2027.
Nvidia is providing King Yuan Electronics with an advance of $1.1 billion for US production capacity. Testing is scheduled to begin in Singapore in February 2027 and in the US in April 2027.
Nvidia and Siliconware Precision Industries are expanding optical module production in Taiwan for co-packaged optics. Capacity is set to increase 15-fold by mid-2027 and double again by late 2028.
