Submarine communication cables, the unsung heroes of global connectivity, have been silently weaving together continents for decades. With nearly 560 of these cables deployed across the world, they carry an astonishing 99% of intercontinental data traffic. As we stand on the brink of an AI-driven future, the implications for these vital conduits are profound. The surge in demand for processing power and the associated electrical consumption from AI applications could reshape how we design and operate the infrastructure that feeds data into these cables.
Reports are already rolling in, painting a picture of a looming crisis. Countries like Singapore and Ireland have hit the brakes on data center projects due to concerns over their power grids’ capacity. Yet, the industry remains in the dark about how AI will impact transport networks, including the submarine links. The cloud computing revolution offers some insight, as it experienced a significant evolution in traffic patterns over the last decade. Initially, “user-to-network” traffic, or north-south traffic, dominated the landscape. However, as companies transitioned to cloud applications, the real game-changer emerged: east-west traffic. This intra- and inter-data center communication skyrocketed, driven by virtualization and the need for constant server interaction.
Today, as AI takes center stage, we can expect a similar shift. The transition to AI data centers is already beginning to boost east-west traffic, particularly during the training of large neural networks. This process often exceeds the capacity of a single data center, necessitating communication across multiple locations. While inference tasks require low latency and typically occur within a single data center, the north-south traffic will inevitably rise as AI applications evolve from simple text responses to high-quality video content.
Submarine cables, with an engineering lifespan of 25 years, offer limited opportunities for upgrades post-deployment. However, the submarine line-terminating equipment (SLTE) can be enhanced to adapt to evolving demands. AI holds the key to this enhancement, primarily through operational and network-based functionalities. For instance, automating network planning and optimizing capacity can streamline the labor-intensive process of characterizing fiber pairs. AI can also assist in spectrum allocation and optical power planning, ensuring stable service levels.
Moreover, enhanced network monitoring through AI can help manage the deluge of telemetry data from coherent transponders. By extracting patterns and correlations that often go unnoticed, AI can revolutionize how we monitor network health. Predictive maintenance powered by AI will enable operators to identify potential issues before they escalate, keeping the cables humming along smoothly.
On the network side, AI-enhanced transponders are paving the way for advanced features like threat monitoring and proactive protection. The development of seismic anomaly detection systems showcases how existing submarine cables can contribute to tsunami warning systems, while also addressing the pressing need for enhanced sensitivity to threats like seabed trawling.
The future of submarine communication cables and AI intertwines in complex ways. As the demand for data continues to surge, the maritime industry must adapt to the challenges and opportunities presented by these technological advancements. The marriage of AI and submarine cable technology could lead to a more resilient, efficient, and responsive network, but it also raises questions about sustainability and the long-term viability of our power infrastructure. As we navigate these uncharted waters, one thing is clear: the maritime industry must brace itself for a transformation that could redefine how we connect and communicate across the globe.
