Nvidia unveiled a sweeping series of multi-year supply-chain expansions to relieve bottlenecks in AI computing hardware, announcing the plans at its GTC Berlin event. The chipmaker said substrate capacity for its products will increase 2.5 times over three years via agreements with Japan’s Ibiden, South Korea’s Samsung Electro-Mechanics and Taiwan’s Unimicron. Packaging partner Amkor aims to lift interposer wafer capacity by more than 300% by 2028. Testing specialist KYEC will add new U.S. facilities backed by a $1.1 billion prepayment from Nvidia, part of a buildout that Nvidia says will double its overall testing capacity from 2026 to 2027. SPIL plans to raise output of optical engines used in co-packaged optics 15-fold by mid-2027, with another doubling by late 2028.
CEO Jensen Huang framed the moves as essential to meet “extraordinary” demand for AI systems, broadening geographic redundancy with upgrades and new sites across Japan, South Korea, Taiwan, Singapore and the U.S. The company cautioned the initiatives remain subject to execution, market and regulatory risks.
Related article:
Co-Packaged Optics: The Next Big Thing in Data Center Networking?
