Adding capacity to existing subsea cable routes does not necessarily reduce risk and, in some cases, can concentrate it further. According to FLAG, changing traffic patterns, distributed compute environments, and AI workloads are prompting operators to rethink how global networks are designed, with greater emphasis on route diversity and resilience rather than capacity alone.
In this Q&A, Carl Grivner, CEO of FLAG, discusses how network design is evolving, the growing importance of resilience, and the implications of AI-driven traffic patterns for global connectivity.
What factors now shape how subsea cable routes are designed?
The way subsea routes are designed has changed significantly. It is no longer simply about finding the shortest path between two points or adding capacity where traffic demand is highest. The nature of demand itself is changing.
AI is a major factor. It is not only increasing traffic volumes but also changing how data moves, where it flows, and the level of consistency networks are expected to deliver. The focus is shifting from efficiency alone to ensuring that networks can handle sustained and sometimes unpredictable demand.
At the same time, some established routes in the Northern Hemisphere carry a disproportionate share of global traffic. Simply adding more capacity to those routes does not necessarily solve the problem and can further concentrate risk. As a result, operators are increasingly exploring alternative paths that distribute traffic more effectively.
The location of data centres and the distribution of computing resources are also becoming central considerations. Route design is now part of a broader infrastructure strategy that links connectivity, compute and data storage.
How are operators engineering resilience and continuity into cable infrastructure today?
Resilience is increasingly being built into network architecture rather than treated as a feature of individual cable systems. Operators are designing networks that can continue carrying traffic when conditions change, making route diversity a key consideration.
There is also greater recognition that disruptions do not occur only in the deep ocean. Landing stations, shore-end segments, and terrestrial links connecting cable systems to data centres and interconnection hubs are also potential points of failure. As a result, resilience must now extend across the wider network environment, including how these connections are designed and how quickly traffic can be rerouted when disruptions occur.
These efforts are taking place against a backdrop of geopolitical uncertainty, regulatory change and fluctuating demand patterns. In this context, resilience is not only about keeping systems operational but also about maintaining performance and adapting quickly when circumstances change.
How is global bandwidth demand evolving with AI workloads?
AI is changing both the scale and nature of network demand. While traffic volumes are increasing rapidly, the more significant shift is the growing need to move large datasets between regions, cloud environments and processing locations.
Networks are supporting more distributed traffic flows as data moves between multiple locations simultaneously. This differs from the more traditional pattern of traffic flowing primarily between end users and centralised data centres.
As a result, existing routes are being supplemented by investments in multiple additional paths to support more distributed network architectures. The discussion is also shifting from overall capacity levels to the ability to deliver reliable and consistent performance under increasingly demanding conditions.
What assumptions about global networks no longer hold up?
One common assumption is that simply adding more capacity solves network challenges. In practice, adding capacity along the same routes can increase the concentration of risk rather than reduce it. The key issue is not only how much capacity exists, but where it is located and how it is deployed.
Another assumption is that networks operate in relatively stable environments. Demand patterns are becoming less predictable, while geopolitical and operational risks continue to evolve. Infrastructure planning increasingly needs to account for these uncertainties.
The industry’s understanding of value is also changing. Historically, ownership of cable systems or routes was often viewed as a key differentiator. Today, the ability to provide diverse routing options and integrate multiple systems is becoming equally important.
What is becoming possible in global connectivity that wasn’t realistic before?
One significant development is the growing integration of subsea networks, terrestrial infrastructure, and data centres into a more connected ecosystem.
This allows networks to support greater flexibility, enabling traffic to be rerouted across multiple paths when necessary and reducing dependence on any single route. It also creates more options for route diversity and redundancy.
At the same time, network infrastructure is becoming more closely aligned with where data is processed and stored. As AI applications place greater emphasis on speed and proximity, connectivity strategies are evolving accordingly.
Perhaps the most notable shift is that customers are increasingly focused on outcomes such as performance, consistency, and reliability rather than infrastructure assets alone. This reflects a broader move towards designing networks around what they need to deliver, rather than solely around the infrastructure itself.
