- Ciena is seeing strong demand for 1.6T wavelengths to support AI workloads
- Nokia says 400G is now mainstream, and it expects 800G to ramp in 2026
- Ribbon’s Apollo platform is delivering 800G capabilities in live deployments
Some of the top optical vendors in the world are saying that the surge in AI and data center traffic is increasing demand for higher optical connectivity, such as 400G and 800G. Fierce Network touched base with executives from Ciena, Nokia and Ribbon to get their predictions on the optical market for 2026.
Jürgen Hatheier, VP of Business Development & CTO Global Partnerships at Ciena, said the vendor is seeing a clear shift toward much higher-capacity optical connectivity.
“We also see strong demand for 1.6T wavelengths as service providers, cloud and neoscaler customers upgrade their networks to support AI workloads,” said Hatheier. “Service providers like Lumen in the United States, e& in the UAE and Cirion in Latin America (just to name a few) have all adopted our WaveLogic 6, 1.6 Tbps coherent solution to support the explosive growth in cloud and AI traffic. We expect this demand to continue in 2026.”
Rob Shore, head of Portfolio Marketing for Optical Networks at Nokia, said the company is seeing strong and sustained demand for 400G and growing momentum toward 800G especially from cloud and content providers who are interconnecting AI training sites and data centers.
“400G is now mainstream, while 800G deployments are beginning and expected to ramp rapidly in 2026,” said Shore. “An emerging application in the market driving adoption for 800G coherent optics is scale across networking where a GPU cluster is spread across multiple geographically dispersed data centers and connected with high-speed optics. The market for these types of high-speed interfaces is projected to reach more than one million units per year by 2030.”
In fact, the vendor AOI said in October that it plans to ramp its production of 800G transceivers for data centers by constructing a new 210,000-square-foot facility at its Sugar Land, Texas headquarters. The plant will also be used to make 1.6 Tbps transceivers once the company sees more orders come in.
“The manufacturing process that we have developed and the design of the 800G and subsequently 1.6-terabit transceivers is very similar, meaning AOI can produce either on the same production line,” said AOI CFO and Chief Strategy Officer Stefan Murry.
Jonathan Homa, senior director of Solutions Marketing at Ribbon, told Fierce, “Interest in 400G and 800G connectivity is strong, and our Apollo platform is already delivering 800G capabilities in live deployments to help carriers and hyperscalers scale quickly.”
Ribbon recently partnered with North Georgia Network Cooperative (NGN), a key contributor to regional broadband development, to modernize and expand its optical network with Ribbon’s Apollo 800G optical transport solution.
Asked for a prediction of the optical market in 2026, Ciena’s Hatheier said, “The next wave of AI innovation will focus on efficient data movement between GPUs, with interconnects becoming the key factor for scaling.” He said in 2026 AI fabrics — with silicon, optics and link technologies — will become a competitive advantage and be as critical as GPUs.
Nokia’s Shore predicts that 2026 will see 800G coherent pluggables becoming the standard optical connectivity solution for AI networks and that there will be an evolution of short reach intra-data center optical solutions optimized around power efficiency.
A service provider's perspective
Wayne Lotter, head of International Networks for Australia's Telstra International, sent Fierce the following prediction:
“Looking ahead to 2026, we’re going to see a further shift in how carriers, enterprises and hyperscalers consume high-bandwidth capacity for their cloud and data center needs. The industry is moving towards 'capacity as a service' models, where customers subscribe to flexible pools of capacity that can be deployed wherever needed across subsea and terrestrial routes.
“This is a significant change from the traditional model of placing individual capacity orders. Instead, customers will work under outcome-based agreements designed to meet the demands of AI and cloud services. By shifting capacity dynamically across different networks, enterprises will start to get the flexibility and speed-to-market they need.”