The 2025 China International Optoelectronic Exposition (CIOE) opened its doors in Shenzhen from 10–12 September, bringing together global leaders in optical communications and intelligent computing. Under the theme “ Computing, Network, Connectivity”, Ruijie Networks showcased a powerful lineup of innovations designed to meet the soaring demands of AI-driven computing and high-speed interconnectivity.

Setting New Benchmarks for Optical Interconnect

At the centre of Ruijie’s exhibit was the 51.2T CPO (Co-Packaged Optics) Switch Commercial Interconnect Solution, powered by Broadcom’s Bailly 51.2Tbps CPO chip. This cutting-edge switch integrates 128 x 400G FR4 optical ports within a compact 4RU form factor. By co-packaging optical engines with switching chips, the solution dramatically shortens electrical interconnect paths, reducing signal loss and power consumption while achieving energy efficiency and scalability. The technology provides a clear pathway to future 800G and 1.6T upgrades, ideal for massive AI training and hyperscale computing clusters.

Energy-Efficient LPO Optical Modules

Ruijie also introduced a new series of 1.6T/800G LPO (Linear Pluggable Optics) modules, covering speeds from 400G to 1.6T. Leveraging silicon photonics integration, these modules remove traditional DSP chips to lower system power consumption by up to 38% per module and reduce end-to-end latency by as much as 200 nanoseconds in multi-hop scenarios. Beyond performance, the energy savings translate to significant cost reductions and lower carbon emissions—benefits that resonate strongly with operators of AI data centres and high-frequency trading environments.

AI-FlexiForce: Driving Lossless AI Networking

Drawing strong attention from attendees, the AI-FlexiForce Lossless Network Solution demonstrated Ruijie’s commitment to AI-era connectivity. Designed for large-scale AI training clusters that demand ultra-high throughput, ultra-low latency, and zero packet loss, AI-FlexiForce leverages a DDC 2.0 distributed architecture to enable flexible, open, and elastic scaling. Supporting clusters of up to 72,000 GPUs interconnected at 400G speeds, the solution delivers seven layers of high-availability protection, ensuring stability and uninterrupted training even in mission-critical environments.

Sharing Insights on the Future of AI Optical Technology

During the concurrent “Next-Generation Optical Technology for Ultra-Scale AI Clusters Forum”, Ruijie optical systems engineer Dr. Liu Jingwei delivered an in-depth keynote analysing the evolution of AI interconnect technologies. Highlighting the explosive growth of global AI computing power, doubling nearly every two years, Dr. Liu underscored the importance of optical interconnect as the key to overcoming bandwidth bottlenecks. He detailed the complementary roles of LPO, NPO (Near-Packaged Optics), and CPO, emphasising that these technologies will coexist and evolve to meet diverse cost, power, and scalability needs in the AI era.

Leading the Future of Intelligent Connectivity

From components to full-scale system architectures, Ruijie’s presence at CIOE 2025 showcased its end-to-end innovation strength across optical networking and intelligent computing. With breakthroughs in CPO switching, energy-efficient LPO modules, and next-generation lossless AI networks, Ruijie continues to drive the future of high-performance connectivity, empowering industries worldwide to unlock the full potential of AI and next-generation digital infrastructure.

As global demand for computing power accelerates, Ruijie remains committed to advancing optical and AI networking technologies that deliver not only performance, but also energy efficiency and long-term sustainability.