AI is creating new demands on the infrastructure that sits behind our networks.
As more computing moves into data centres and closer to the edge, there will be a growing need for the power and connectivity to support it.
However, for Nokia, the rise of AI could mark a new “network super-cycle”, creating opportunities across the UK telecoms industry.
But making the most of that opportunity will depend on more than just building data centres. Power, connectivity and where critical infrastructure is located and managed will all play an increasingly important role.
Paul Alexander, Nokia’s UK Managing Director and VP for Sales for the UK and Ireland, argues that AI should be viewed not simply as a new technology cycle, but as a network super-cycle that will create new demands on telecoms and infrastructure.
“If you think about what AI is and how AI actually gets delivered to people, you’re talking about data centres containing large amounts of compute,” he tells Mobile Europe.
For Alexander, those data centres require three fundamentals, which includes the need of a physical site and the computing equipment, power to run it and a strong network connecting everything together.
“If you don’t have the network, you can’t transport the data, which is fundamentally what we’re talking about here,” he says.
For Nokia, that creates opportunities across several parts of the UK telecoms market.
Mobile operators are continuing to build out 5G and need to connect their growing numbers of base stations. At the same time, there is increasing interest in putting AI computing closer to the edge of the network.
However, the next opportunity could be in the networks connecting data centres, both within the UK and internationally, he reveals.
Building an AI grid
One idea the telecoms giant is increasingly talking about is the “AI grid”. Rather than putting huge amounts of computing power into a small number of enormous data centres, AI infrastructure could be spread across many smaller sites.
Alexander gives the example of a data centre requiring 100MW of power.
Instead of building everything in one location, the same amount of computing could potentially be spread across dozens of sites.
He claims this will “reduce the risk”, allowing potential opportunities for telecom operators as they already have buildings, sites and other infrastructure that could potentially be reused for smaller, edge data centres.
“They have a lot of that infrastructure set there, doing nothing that they could repurpose for those smaller edge type of data centres,” he explains
But spreading the computing around creates another requirement: connectivity.
“To connect all that together, you need, again, a high-capacity, low-latency optical network,” he says. That is an area where Nokia sees a clear opportunity.
A more connected ecosystem
Alexander also expects AI infrastructure to change the way companies work together.
“The difference in the AI infrastructure world is that it’s much more of an ecosystem than a traditional telecoms network,” he explains.
However, the UK already has a substantial data-centre industry, and Alexander believes the country is well placed to benefit from the growth of AI.
Despite having over 500 data centres dotted around the UK, the main issue for the country is low-cost power, he admits.
As a result, he points to the Nordic countries, where data centres can benefit from large amounts of renewable energy and relatively low-cost power.
“The issues we need to figure out is: how do we get more low-cost, renewable power?” he questions.
For him, solving that problem would make the UK an even more attractive location for data centres and AI infrastructure. He adds: “The growing importance of data centres means the lines between telecoms and critical national infrastructure are becoming less clear.”
The question of sovereignty
“Where the equipment is made, how it’s managed, where it’s operated from – all that kind of stuff. Sovereignty is becoming a critical issue,” he explains.
However, Nokia believes its European base gives it a role to play as governments and businesses think more carefully about where critical AI infrastructure is built and managed.
“Nokia is a European vendor. In fact, the only European vendor that is really able to provide the full complement of infrastructure that’s needed to underpin AI, he claims.
For Alexander, the future of AI is therefore not just about bigger data centres and more GPUs. It will depend on the infrastructure connecting all of them.
“It’s going to be in those networks that allow that AI grid concept to come to life,” he concludes.
IN RELATED NEWS
