AI has ushered in a new era of technological possibilities and challenges, yet the full impact on networks is still largely unknown. The signs suggest that it is going to add significantly to the huge amounts of data traversing networks, also requiring them to densify and become more intelligent. Most service providers have identified this and are looking to fortify their networks, particularly in the metro where a lot of AI data is created and transported.
Capacity upgrades in optical and passive optical networks
One of the primary challenges networks face in the wake of AI is the need for increased capacity. As AI applications become more sophisticated, the large language models used in technology such as GenAI generate massive amounts of data that must be transmitted and processed efficiently. This necessitates upgrades in both optical and passive optical networks to accommodate the surge in data traffic as this data traverses the network.
Operators are investing in advanced coherent optical technologies, that can deliver 400G and 800G, which can greatly improve network capacity, transferring up to 800 billion bits per second. Technology such as this can enable service providers to transfer more data with less power, helping to reduce their carbon footprints.
Service providers are ramping up the deployment of high-capacity fibres as part of fibre-to-the-premises (FTTP) rollouts, which can better meet the demand for speed, latency and reliability that AI and other next-gen applications demand.
In addition, they are ensuring that their metro networks are closely connected to subsea networks and the cable landing stations connecting global data centres that are powering AI applications.
By bolstering these foundational elements, networks can lay the groundwork for seamless integration with the evolving landscape of AI.
Connecting the edge
GenAI will go beyond low latency to deliver real-time responses. AI also needs to act on local intelligence to make better-informed decisions. To meet these requirements, we will see the strategic deployment of distributed edge data centres across a metro network.
These mini data centres bring computational resources closer to the end-users and AI applications, reducing latency, and improving the overall responsiveness of applications and the intelligence they can deliver.
The network, in turn, must be robust enough to connect these distributed edge resources seamlessly, ensuring a cohesive and responsive infrastructure for AI.
The infrastructure must be capable of efficiently transmitting data between edge data centres, a central cloud and end-users. Network route diversification to a local edge or central data centre is crucial for service providers to ensure that they can provide redundancy in case of any problems with a fibre line.
Network operators are now focusing on enhancing this data centre-to-data centre connectivity backbone to ensure seamless communication across the distributed elements of the network.
Software Intelligence: Navigating Traffic Flows and Mitigating Bottlenecks
As GenAI applications become more widespread, software intelligence within the network becomes increasingly important. Operators are integrating advanced software platforms that gain insights into major traffic flows and identify potential bottlenecks.
With AI and machine learning underpinning these software platforms, they can understand the patterns and dynamics of data traffic, automating the allocation of resources and optimising routing in real time. This proactive approach ensures a smooth and efficient operation, even as the demands on the network fluctuate with the evolving AI landscape.
