Wedoany.com Report on Feb 6th, Tower Semiconductor recently announced that it is collaborating with NVIDIA to jointly develop 1.6T data center optical modules for next-generation artificial intelligence infrastructure, leveraging its silicon photonics manufacturing platform. As hyperscalers and various organizations continue to drive AI clusters towards higher bandwidth and lower power consumption interconnect solutions, this collaboration aims to achieve high-speed optical connectivity solutions that comply with NVIDIA's networking protocols.
Tower Semiconductor stated that its silicon photonics platform can achieve data transfer rates up to twice that of previous silicon photonics solutions, providing higher throughput for AI training and inference architectures. The company positions this foundry technology as the foundation for optical transceivers and network components in large-scale data center deployments, where optical I/O is gradually replacing short-range electrical connections.
NVIDIA is collaborating with Tower to expand its networking ecosystem to support rapidly growing artificial intelligence workloads. Both companies view silicon photonics technology as a key enabler for efficiently scaling AI infrastructure, especially as data rates evolve towards 1.6T per module in next-generation optical interconnects.
The core objective of the collaboration is to design 1.6T data center optical modules that comply with NVIDIA's networking protocols. Tower's silicon photonics platform aims to double the data rate compared to previous-generation technologies, focusing on high-speed optical interconnects for AI training and inference clusters, and supporting the broader transition towards optical connectivity within AI-driven data centers.
"Tower Semiconductor is proud to provide advanced, high-speed technologies that support the demanding requirements of data centers and AI," said Russell Ellwanger, CEO of Tower Semiconductor. "We continue to make significant investments in our SiGe and silicon photonics platforms to support the ecosystem for next-generation data center architectures with performance, scalability, and manufacturability."
This collaboration reflects the accelerating industry trend towards 1.6T-class optical modules as AI architectures scale beyond 800G. Foundry-ready silicon photonics technology is becoming a crucial link in the supply chain. NVIDIA increasingly emphasizes networking and optical I/O as key elements of AI infrastructure, while foundries like Tower position silicon photonics as a technological advantage that runs parallel to advancements in traditional logic and packaging.
