LEE MYALL AI INFRASTRUCTURE
We must strike a balance between encouraging this growth while also ensuring that our connectivity networks are up to scratch.
near future. Realising this decentralised approach to processing across the edge of the network will unlock the ultra-low latency that is essential to the effective use of AI applications by businesses and individuals. The full effect of these edge data centres won’t be realised unless the fibre connecting them back to the core network is equally high capacity. Investment in the UK’s edge network extensions should be distributed as evenly as possible. Not just to cater for current and future growth zones and the valuable hyperscaler investments, but to ensure that the UK isn’t leaving any communities underserved from a social, levelling up perspective. Otherwise, we could face a future where swathes of the UK are unable to leverage AI due to insufficient local network capacity and connectivity - a serious reputational dent for an aspiring AI superpower. But AI and data centres aren’t the only factors driving network demand. As a key landing point for transatlantic subsea cables (60 at last count) connecting Europe to the US, the network also needs to maintain fast and reliable connections from subsea cables to data centres and then out to where users are. As investment in subsea cable systems continues to increase due to the AI and cloud computing the strength of the UK’s network will become even more critical. AGEING INFRASTRUCTURE While a lot of the core networks we have in the UK do have capacity and deliver high performance, some of the UK’s network infrastructure is ageing, exceeding 20 years in service. Much of the fibre originally laid may no longer align with today’s digital demands. These ageing networks were not built to support the high-bandwidth, ultra-low latency requirements of AI applications, which could lead to bottlenecks as AI- driven workloads increase. Compounding this challenge is the lack of new core network routes in recent years. Comparatively, the US is
actively investing in fresh fibre corridors to improve redundancy, resilience, and capacity across the country. To remain competitive, the UK must follow suit by building modern, new high- performance fibre routes that directly connect key data centre locations across the country. Without targeted investment in long-haul fibre corridors and backbone network upgrades, the UK risks falling behind its transatlantic counterparts. A failure to modernise the nation’s digital infrastructure could limit its ability to fully capitalise on AI, to capture hyperscaler investment, and unlock a truly digital economy. A REGULATORY ENVIRONMENT FIT FOR PURPOSE? The UK government has demonstrated that it is prioritising the nation’s data centre and AI ambitions. Developments such as the reclassification of data centres as critical national infrastructure have been promising first steps, with the Deputy Prime Minister even going so far as to re-open previously refused planning submissions for sites last year. And, with the recent planning reforms, the government has also created fast- tracked planning processes to move data centre submissions on green belt land from local to central government purview. While we can’t expect the same level of government intervention on networks, some things can be done. The UK could benefit from a joined-up policy for integrating network builds with major infrastructure projects (roads, railways, etc.). The government’s support here could help bridge gaps and optimise national connectivity. A NEW TELECOM AI ACTION PLAN The investment that has been driven into the UK for AI and hyperscaler activities is immense, but so is the proposed data demand on our networks. With plenty of investment secured, it’s time to move the focus back to the foundations - building out
past our core network to bring all areas up to scratch to meet demand. With AI applications and hyperscaler data to account for, the focus needs to be on the edge. Building out edge computing infrastructure to ensure high-speed and high-bandwidth connectivity. Carrying out inference closer to the end-user will be essential to support the location- specific, low-latency services that are a prerequisite for running AI applications smoothly. Connecting this all with fresh, high-capacity fibre will do double duty, improving the overall performance of our networks while also making the UK particularly attractive for continued hyperscaler investment. The government also has a part to play. After stimulating this demand and setting new AI ambitions in its latest AI strategy, it needs to continue its support of data centre development with continued framework adjustments. Power generation could also benefit from government attention, with new, green energy sources key to driving the price tag of power down. We must strike a balance between encouraging this growth while also ensuring that our connectivity networks are up to scratch. We don’t want to over-promise and under-deliver.
Lee Myall, CEO, Neos Networks.
www.opticalconnectionsnews.com
17
ISSUE 41 | Q2 2025
Made with FlippingBook interactive PDF creator