Ciena has rolled out a suite of new innovations designed to meet surging AI-driven bandwidth demands, including recently acquired co-packaged/near-packaged optics and electrical redriver technologies from Nubis Communications.

“Ciena is building on our established leadership in high-capacity, low-latency optical networking with innovations that address the AI-driven transformation of networks,” said Dino DiPerna, Senior Vice President of Global Research & Development at Ciena. “Our latest advancements in hyper-rail photonics, 1600ZR/ZR+ pluggables, and agentic AI enabled automation drive high-speed connectivity and scalable capacity to support the increasing demands of next-generation AI workloads.”

Ciena’s new data center innovations featured at the OFC 2026 optical networking and communications conference in Los Angeles included:

Hyper-rail photonics: new amplifier configurations in collaboration with hyperscalers and service providers, that deliver up to 32x the density of today’s solutions or 128 fiber pairs in a single rack, reducing power consumption by up to 75% and space requirements by 85%
Next generation coherent technology: 2nm silicon designed to enable single-carrier 1.6 Tb/s coherent (1600ZR/1600ZR+) pluggables and full spectrum transponders that scale, simplify, and automate networks with higher capacity, more compact solutions
AI-driven network automation: Agentic AI capabilities powered by data-rich telemetry, featuring new agents for assurance, routing, and network digital twin validation, deliver exceptional operational efficiency and reliability.

“The surge in AI workloads is creating unprecedented pressure within and between data centers, forcing a rethink of network architecture. According to our findings, demand for purpose-built optical DCI equipment will nearly double in the next five years,” said Jimmy Yu, Vice President at Dell’Oro Group. “In that period, cloud providers will implement new architectures such as scale-across to overcome space and power bottlenecks. Ciena’s new innovations in hyper-rail photonic systems, 1600ZR/ZR+ based on 2nm design, and full-spectrum transponders will help bridge the gap between the existing infrastructure and the higher-capacity, lower-latency demands of AI-driven applications.”