Sivers, O-Net and Enablence partner on ELS for AI data centres
Sivers Semiconductors has announced a partnership with O-Net Technologies and Enablence Technologies to develop an advanced external light source (ELS) module with Sivers' DFB laser arrays to support Co-Packaged Optics (CPO) roll-out in AI data centres and HPC systems.
O-Net Technologies will serve as the original design manufacturing partner, integrating Sivers' laser arrays and Enablence’s NxN Star Coupler to deliver an ELS module for scale-out and scale-up optical systems.
CPO integrates most optical elements into GPU and Switch packages thereby delivering high speed optical interconnects with 80 percent lower energy consumption compared to copper. According to IDTechEx (December 2025 report), the CPO market is projected to exceed $20 billion by 2036, growing at a robust CAGR of 37 percent from 2026 to 2036, with ELS solutions constituting 10 percent of this market.
“ELS is a crucial companion to CPO because it separates temperature-sensitive lasers from the extreme heat of high-power processors, ensuring wavelength stability and significantly enhancing system reliability and serviceability,” said Alex McCann, managing director for Sivers’ Photonics business. “Our high-performance DFB laser arrays deliver the reliability and wavelength stability required to accelerate the deployment of AI data centres with CPO based architectures.”
“O-Net Technologies is pleased to partner with Sivers and Enablence to enhance our advanced external light source module,” said Austin Na, CEO and chairman of O-Net Technologies. “Together, we’re enabling a practical, scalable ELS aligned with the needs of High-Performance Compute (HPC) and AI data centers.”
“We are excited to partner with Sivers and O-Net to help redefine how optical interconnects scale for the AI era,” said Todd Haugen, CEO of Enablence Technologies. “Our NxN Star Coupler enables efficient wavelength distribution at scale, and together we are unlocking an effective optical pathway for CPO based architectures to keep pace with exponential growth in AI data centre compute.”
