2023 was the year of Artificial Intelligence (AI). 2024 will build on the incredible momentum of the likes of ChatGPT, Google Bard, Microsoft CoPilot, and others, delivering applications and services that apply AI to every industry imaginable.
A recent analysis piece from Schroders makes the point well: “The mass adoption of generative Artificial Intelligence (AI) …has sparked interest akin to the Californian Gold Rush.”
But – just like 1848 – there are already problems developing. One of these is the remarkable concentration of AI businesses. For example, approximately 80% of AI start-ups are based in London due the availability of talent, equity funding and a diverse ecosystem providing conditions ripe for business growth. Proximity to these resources gives budding AI enterprises a greater chance of success.
This is causing a huge issue for the infrastructure behind AI. The data centres, networks and facilities that underpin this revolution are straining.
AI workloads increase the demands for power and cooling in data centres substantially. We have already seen structural changes to facilities to take account of this.
For centres within dense cities, the problem is acute. Metropolitan locations have huge, aggregated demand, as established organisations and start-ups alike have applications of AI that they want to develop and bring to market.
The demands of AI applications, be it supercomputing, video analytics, smart grid management of thousands of sensors or the GPUs that are so critical to AI – force data processing that is close to source.
And these demands are increasingly rapidly across sectors. Retailers want to drive more engaging experiences that require real time access to data. And demands for real-time services in healthcare or autonomous vehicles are growing. But behind the scenes, low latency and fast data access are paramount to this delivery.
In short, AI has become the classic ‘unstoppable force.’ But the equally classic ‘immovable object’ is that all that data cannot all sit in London to be effective.
It is prohibitively expensive and disruptive to adapt facilities within densely populated, metropolitan city centres. Consider the delays that infrastructure projects inflict on a regular basis in places like London, Birmingham, Manchester, or Edinburgh.
And even if those facilities could be updated without extensive disruption, it would continue to force businesses to aggregate around cities. This would restrict and stifle the innovation within the regions where businesses are calling – very vocally – for better digital infrastructure to drive growth.
Businesses need to be empowered to embrace AI wherever they are. There are reasons for this expansion everywhere: The new supercomputing centres in the UK are in Edinburgh and Bristol. The Defence Science and Technology Laboratory (Dstl) is creating a new AI unit based at the National Innovation Centre for Data (NICD) in Newcastle. In Scotland, the AI Alliance was set up between The Data Lab and the Scottish Government to aid the delivery of Scotland’s AI Strategy and enhance collaboration.
The infrastructure to capitalise on this already exists. We know because we have built it. What is missing is the drive to risk going beyond London.
But beyond the capital is where we must venture. If we combined the six regions with the lowest regional gross value added (GVA) and simply got them up to the UK median, it will generate an additional £4.8bn for the economy. Combining the presence of start-ups and scale-ups across the UK, the power of AI and the possibilities afforded by regional edge computing, offers a chance to do this.