From the world of technological advancements, there’s no doubt that artificial intelligence (AI) poses as the reigning king. AI, like a relentless tide, continues to engulf businesses, investors and even governmental entities. This is not an exception for the telecom sector, as evident in the published ‘Top 10 Telco Trends 2024’ by Juniper Research, where AI graced three slots.
Despite the spotlight on AI’s capabilities, from network intelligence to voice bots, a significant element still remains underappreciated. The spotlight here, is on the one question not many seem to recognize: How equipped are telecom networks to anchor AI and its colossal hype? While NVIDIA and its GPU chips have been in the limelight for the enabling infrastructure, it’s important to note that you cannot maximize your NVIDIA chips without good connectivity.
Forecasts from IDC anticipated global spending on AI-oriented systems to scale up to $154 billion by 2023, and then to rise further to $300 billion by 2026. This indicates that we have barely begun to tap into the potential growth of AI. As the UK government set its sights on being a forerunner in this space and is two years into a ten-year plan to position the UK a ‘global AI powerhouse,’ it is essential not to disregard securing telecom networks capable of delivering on all these promises.
According to Gartner, generative AI, which stands less than 1% today, will constitute 10% of all data produced globally by 2025. It is anticipated that with the rising deployment and construction of AI applications and generative AI tools like ChatGPT and DALL-E, the volume of data coursing through networks to data hubs will surge. IDC speculates that generative AI alone will generate zettabytes of technology in the coming five years.
While considering all this, the subsequent rise of data centres cannot be overlooked. Telecom networks will have to shoulder all this extra data flow back to these hubs, before even factoring in the continued growth of other data-consuming technologies such as 5G or IoT.
To handle this, the construction of a dynamic computational infrastructure will be pivotal, supporting AI from the network edge to data centres. Research from the 650 Group states that almost 1 in 5 ethernet switch ports that data centres acquire will be associated with AI/ML and accelerated computing by 2027. Strategic placement and sufficient Points of Presence will be crucial to manage the breakneck information flow in a cost-efficient and sustainable way. Giants like Microsoft seem to have already taken this into consideration, with plans to invest £2.5 billion in building “next-generation AI data centres” in the UK.
However, just creating data centres is not the solution to facilitate the effective functioning of AI across the UK. The network infrastructure needs to be comprehensive enough to transport ample data to and from these data centres. This not only involves a wide reach but also the capacity to support the heavy data traffic required for AI applications. As AI evolves, the demand and investment in data centres in the North of England, particularly in areas close to hubs like Liverpool and Manchester, are projected to grow.
Evolving along with all these rapid transformations, large data centres need to transition to faster, more scalable infrastructures. To keep up with increasing users, devices, and applications, high-capacity connectivity becomes crucial. While the OEM ecosystem has done an impressive job in pushing the boundaries of what is feasible, it is clear that the industry needs to further incorporate 800G and even 1.6TB solutions.
In conclusion, notwithstanding the numerous potentials of AI for telecom networks, it’s evident that the two have a much more interconnected relationship. If the nation’s static infrastructure is unfit for managing the escalating volume of data, its AI objectives may stumble. Telecom providers, businesses investing in AI, and even the government need to recognize and support the crucial role these networks will play in shaping the UK’s AI future.
