Spanish connectivity provider Telxius is deploying Nokia’s 800G coherent optical technology across terrestrial networks in Europe and the Americas, adding capacity to infrastructure linking data centres, cloud services, subsea cables, and major internet hubs. The project uses pluggable optical modules in an IP-over-DWDM architecture, allowing Telxius to put higher-capacity wavelengths into compatible routing systems without relying on a separate transport appliance for every connection. The deployment follows an earlier long-distance trial over the BRUSA submarine system.

Nokia’s ICE-X family supports 800ZR and ZR+ operating profiles alongside automation software for monitoring, fault management, and service provisioning. Telxius says the architecture should reduce power, physical space, and expansion cost while improving visibility across the optical network. Those benefits will depend on the routes and equipment configurations used in production, but the broader shift towards coherent pluggables is already changing the way carriers design high-capacity networks.

The architectural change is more useful than the headline 800G figure alone. Traditional long-distance optical networks often use dedicated transponders to convert router traffic into signals suitable for fibre transport, whereas coherent pluggables package more of that capability into compact modules that can sit directly inside compatible equipment. Fewer separate layers can reduce hardware count and simplify some deployments, although operations teams still have to manage the optical performance of the underlying fibre.

Different optical profiles allow operators to trade capacity against reach and fibre conditions, making the same technology useful for data-centre interconnection, metro links, and longer terrestrial routes. The economics depend on amplifier design, available spectrum, route length, and reliability requirements rather than the maximum line rate printed on a module. Telxius therefore gains another tool for increasing capacity across an estate with varied infrastructure rather than one universal 800G configuration.

AI traffic becomes a transport-network problem

AI infrastructure is frequently discussed through processors, electricity, and data-centre construction, yet computing facilities also have to exchange huge quantities of information. Training clusters need storage and external data sources, cloud platforms operate across several sites, and inference services require connections between applications, users, and the facilities running models. Additional compute therefore creates network demand outside the building as well as power demand inside it.

Telxius operates more than 100,000 kilometres of subsea and terrestrial fibre alongside landing stations, points of presence, and data-centre connectivity across Europe and the Americas. Its infrastructure sits beneath many of the cloud and content services generating traffic, making optical upgrades partly a response to growth elsewhere in the digital economy. A new AI campus can attract attention, but its usefulness depends on the routes connecting it with other facilities and users.

The company has already been modernising that estate through 400G routing and other optical upgrades. Moving towards coherent 800G connections continues a wider industry trend in which packet routing and optical transport become more closely integrated. That convergence can reduce equipment and power consumption, although it also creates a greater need for management tools able to provide visibility across layers that were historically operated by separate teams.

Efficiency accumulates across large networks

Energy efficiency has become a more explicit requirement as operators add capacity without proportionally increasing power and cooling. An individual optical module consumes far less electricity than an AI accelerator, but carrier networks contain large numbers of interfaces and supporting systems. Eliminating dedicated equipment can therefore create meaningful savings when the architecture is repeated across many sites and routes.

Higher-capacity optics also improve the economics of fibre already in the ground. New long-distance routes require civil engineering, permits, or subsea construction, whereas carrying more data over existing fibre can defer some physical expansion. Operators have a strong incentive to increase spectral efficiency and equipment density where reliability remains acceptable.

The move towards 800G is not unique to Telxius, and competing optical suppliers are pursuing similar architectures as data-centre and cloud traffic increases. Customers ultimately buy capacity, availability, latency, and predictable cost rather than optical specifications. Telxius’ deployment will therefore be useful insofar as the new technology lets it deliver those outcomes with fewer infrastructure layers and less power per unit of traffic.