Nvidia and Google select Innolux for AI server optical interconnect manufacturing
News

Nvidia and Google select Innolux for AI server optical interconnect manufacturing

Monday, March 9, 2026 at 12:10 AM

Nvidia and Google have reportedly selected Taiwan-based Innolux to produce optical interconnect components for AI servers. This transition from copper to optical interconnects aims to enhance data speeds within server architectures. The partnership highlights a strategic move by US technology firms to diversify their supply chains away from China, which currently controls approximately 50% of the global optics manufacturing market.

Context

Tech giants Nvidia and Google have reportedly selected Taiwan-based Innolux to manufacture optical interconnect components for their next-generation AI servers. This partnership marks a critical architectural shift as the industry moves from traditional copper to optical-for-copper solutions to eliminate data bottlenecks within high-performance compute clusters. By integrating these components into platforms like the Vera Rubin GPU and TPU systems, the companies aim to achieve the higher bandwidth and lower energy efficiency required for large-scale AI training. This strategic move is also a significant step in de-risking global electronics supply chains. Since China currently controls roughly 50% of the global optics market, sourcing from Innolux allows US firms to reduce geographic dependency. For Innolux, the contract validates its transition from a traditional display panel maker to a high-value semiconductor packaging and photonics player, leveraging its existing fab infrastructure to compete in the burgeoning advanced packaging sector.

Related Companies

3
Innolux
3481
Nvidia
Nvidia
NVDA
US
Google
Google
GOOGL
US