Tower Semiconductor said it is working with NVIDIA to support 1.6T data center optical modules aimed at next-generation AI infrastructure, using its silicon photonics (SiPho) manufacturing platform. The collaboration focuses on enabling high-speed optical connectivity aligned with NVIDIA’s networking protocols, as hyperscalers and enterprises push to scale AI clusters with higher bandwidth and lower power interconnects.
Tower Semiconductor stated that its SiPho platform can deliver up to twice the data rate of prior silicon photonics solutions, targeting higher throughput for AI training and inference fabrics. The company positions its foundry technology as a foundation for optical transceivers and networking components used in large-scale data center deployments, where optical I/O increasingly replaces electrical links at shorter reaches.
NVIDIA is collaborating with Tower as it expands its networking ecosystem to support rapidly growing AI workloads. The companies framed silicon photonics as a key enabler for scaling AI infrastructure efficiently, particularly as data rates move toward 1.6T per module across next-generation optical interconnects.
- Collaboration targets 1.6T data center optical modules designed for NVIDIA networking protocols
- Tower’s silicon photonics platform aims to double data rates versus prior SiPho generations
- Focus on high-speed optical interconnects for AI training and inference clusters
- Supports broader shift toward optical connectivity inside AI-driven data centers
“Tower Semiconductor is proud to deliver advanced, high-speed technologies that support demanding data center and AI requirements,” said Russell Ellwanger, CEO of Tower Semiconductor. “We continue to invest significantly across our SiGe and silicon photonics platforms to support the ecosystem with performance, scalability, and manufacturability for next-generation data center architectures.”
🌐 Analysis
The announcement reflects accelerating industry momentum toward 1.6T-class optical modules as AI fabrics scale beyond 800G, with foundry-ready silicon photonics becoming a critical supply-chain element. NVIDIA has increasingly emphasized networking and optical I/O as bottlenecks for AI infrastructure, while foundries like Tower position SiPho as a differentiator alongside traditional logic and packaging advances.
🌐 We’re tracking the latest developments in networking silicon. Follow our ongoing coverage at: https://convergedigest.com/category/semiconductors/
