Objective
The objective of this piece of this Executive Viewpoint is to demonstrate the criticality of investment in edge based data centers to deliver economic growth across a region.
By positioning as an Executive Viewpoint, opinions can be embedded into the narrative to personalize the content and improving relatability with the audience. To ensure credibility, external data sources reinforcing the argument are utilised throughout the piece.
Like any good piece of content, this piece focuses on the real-world outcome as opposed to the technical specifications, to mirror the way audiences think.
Delivery
Once published on the company’s blog, the content can be amplified through:
Corporate channels to demonstrate thought leadership credentials and achieve scale engagement
Executive channels to humanize the brand and create a personal relationship between the audience and the brand
The content features thought provoking questions, real-world implications and data analysis to provide plenty of options for marketers to write engaging posts, drawing the audience from social media platforms onto the company’s domain.
Artificial Intelligence has the potential to revolutionize society, but only when it is distributed evenly for everyone to benefit.
Much has been made of the impact of Artificial Intelligence (AI). Whether it is automating mundane tasks, or deep data analysis, technologists suggest the technology could radically transform business.
Research from PwC suggests the adoption of AI could grow economic output by 15% by 2035. To put that in perspective, the UK’s economy grew by 1.1% in 2024. Alongside the forecasts, findings from XXX states 91% of UK firms have either adopted AI or are piloting, demonstrating business appetite.
But there are of course challenges.
The success of AI relies on the ability to access compute power in data centers. If this compute power remains out of reach, then AI simply does not work.
As it stands, we do not have enough data center capacity if AI scales as predicting. We have to build more data centers, but this must be done strategically as one scenario is that some businesses can access the compute power, while others cannot.
If we are not careful, we will create an AI digital divide.
So how do we avoid this two-tier economy?
What are Hyperscale and Edge data centers?
Hyperscale and edge data centers refer to the size of the facility itself.
Hyperscale data centers are massive, centralized facilities built for cloud-scale processing, while edge data centers are smaller, decentralized sites located near users to reduce latency and enable real-time computing.
Most AI applications will be hosted in a data center rather than having the data processing on-device. Therefore, if you don’t have enough data centers, you don’t have enough servers to host the increasing number of AI applications.
Many countries are investing to increase their data center capacity, the US and the UK have both committed multi-billion investments for example. Although ground is yet to be broken, we have to consider the different ways the money could be spent.
What type of facilities should be constructed?
You can’t ignore the edge
Many data center companies want to invest in hyperscale facilities.
You gain economies of scale through a single site for energy consumption or cooling, the logistics of construction are reduced with a smaller number of sites, while you can also benefit from a security perspective (both physical and virtual) by reducing exposure.
But, if you lean too far into the hyperscale model, you create winners and losers.
One of the benefits of AI is speed and power. It can analyse data in milliseconds and provide actions much faster than would be feasible for humans. It also does not require sleep so can be a 24/7 employee for certain tasks. This is very valuable today, but the AI hosted in a data center needs to be close enough to the endpoint.
This is where the hyperscale model becomes difficult.
If you require AI to react in less than 30 milliseconds, then you would have to be close to where the AI application is hosted. Data does have to obey the laws of physics, therefore, it can only travel so fast to the data center to be processed, then back to the endpoint to be actioned.
How many hyperscale data centers would be required to cover the entirety of a country and is this affordable?
If not, how do you choose who gets the best access to AI services?
Then are edge data centers the answer? Maybe not.
Edge data centers are optimized and built in locations to cut latency and improve real-time processing, but they lack the power and capacity to facilitate storage, training large AI models, or running SaaS platforms.
The consequence of getting this wrong
To understand the consequence of an AI-digital divide, we should look at previous technology innovations.
4G and 5G network infrastructure was deployed disproportionately at the beginning, favouring urbanized and high prosperity areas. This created what was known as the rural digital divide, which had a consequence on the economic prospect of those operating in those regions.
The Italian Institute for International Political studies suggests that for every 10% increase in connectivity coverage, local GDP increases by 1.5-1.6%.
This is one study, but there are countless others. Connectivity offered opportunity to innovate. Now we have AI, which potentially offers an even greater reward.
However, unless we democratically offer the opportunity, the gap between the “have’s” and “have nots” will grow even wider than it is today.
Finding balance in the middle
Realistically, the best possible outcome will be to find a balance.
You need hyperscaler data centers to benefit from economies of scale when training large language models to refine AI algorithms, while edge-based data centers are critical to ensure the presence of AI can be scaled across a wider region.
The risk we have today is that there is not a significant amount of evidence to suggest the world thinks in this way. The companies investing in these facilities will want to ensure the highest financial returns. This means locating in the places with the highest demand, and operating the most efficient sites possible.
This could mean the construction of a smaller number of hyperscale facilities, concentrated around the areas of existing prosperity.
Such an approach would make sense to an accountant, but does it make sense to the wider economy?
What about the people who live in the less attractive areas? Are we happy for them to get poorer with less opportunity or watch from afar as public services around the richer areas become enriched through the adoption of AI.
It would also be unlikely traditional industries are located in the areas which are attractive to the data center companies. Agriculture needs digital transformation to survive, but it might not get it.
We have to think about the big picture, and the big picture demands a more even distribution across a country. Edge compute facilities need to be a major prioritization for policy makers around the world to ensure we avoid an AI-digital divide.