I have been working in telecommunications for more than 25 years and have attended many different events over the years to network and broaden my knowledge of different topics in the industry. One event which I had not yet attended but one of my customers and friends had continually suggested that I should attend was Capacity Europe, as he believed this is where my customers are. So, this year as it fitted in with the other events I had to attend it was the first time that I went.
Capacity Europe’s focus is international wholesale, which is itself a niche part of the telecom industry providing the backbone related to connectivity including international termination, submarine cables, IP international interconnect, Internet exchange, international roaming, international messaging, and emerging project products in the international connectivity domain.
On the one hand, I felt familiar space but on the other hand I felt that I was in a new world dominated by traders hectically running around to find their next meeting, trying to navigate the labyrinth of floors and rooms of the Intercontinental hotel with the serious business is done behind closed doors (bedrooms converted into temporary offices). My understanding is that the ITW is very similar but on a global scale which takes place in Maryland from 5th-7th May 2025.
I had the opportunity to sit in a few presentations and wanted to share some insights on them.
AI
The first topic is AI and the keynote presentation given by Didem Ün Ateş from LotusAI Ltd who has extensive experience working in the era of AI for blue chip companies such as Microsoft and Schneider Electric.
Her presentation was regarding the use of generative AI and AI generally within companies with research suggesting that AI can achieve between 20% and 40% efficiency, which related to the Fortune 500 companies, translates to hundreds of millions of dollars in potential savings. However, she pointed out that this does not translate into an equivalent reduction in the workforce as there is an AI skills gap now, where the current workforce will need reskilling. The talent and skills race is now on and it is imperative that any company puts a programme in place to understand where and how they want to use AI within their organisations but just as importantly to scale up their workforce to be able to leverage this capability.
I particularly liked the following diagram from her presentation which comes from McKinsey. The chart compares different job types and indicates the amount of automation which can be achieved today without the use of AI versus the incremental impact AI could have on automating that job. Note that in none of the cases, the job is rationalised away.

Furthermore, she went on to say that due to its importance AI should be treated as equivalent to other enterprise-wide functions reporting directly to the board. Of course, AI can be beneficial but if not systematically and carefully adopted corporations might be quickly embroiled in lawsuits for misappropriation of data and huge costs without benefit. Therefore it is key for organisations to create an AI strategy before commencing.
Data Centres
The second topic relates to Data Centres (DC) who are core to being able to deliver services including AI. I would like to combine insights from two panel discussions but only focusing on the insatiable appetite for processing power and the need for data centres. The demand seems to be driven from the US, who have the expanse with energy costs being about a quarter of those in Europe. Having said that a data centre is of course an isle without connectivity from telecoms which also needs to be beefed up.
In certain parts of Europe, we have already hit the buffers related to new DC’s with five European countries having declared that they are closed, related to the building of any new data centres due to the power requirements of the DC’s and the limits of their power grids.
It is encouraging to hear that DC’s are now being asked to report on these SDG values down to an individual rack and with the ESG requirements coming into force those procuring data centre spaces are now asking for this type of information to achieve their corporate goals.
Although we will welcome the reduction of fossil fuels the by-product is thermal heat which renewables such as wind and solar do not have. Up to now, cities have been engineered to capitalise on this extra heat to warm houses and businesses. The answer can’t be to use fossil fuels to fill this gap. Therefore, if you are going to rate the efficiency and environmental impact of a data centre, it really should be considered in the round and not in isolation. There seem to be different ways to measure and count with scope for a lot of consulting work.
With many of the original DC’s being intercarrier connectivity hubs, many more than 50 years old, these vital arteries will need upgrading or replacing sooner rather than later. Despite often being in and around built-up areas.
Beyond sovereignty, and use of data legislation another consideration related to data centres location is the type of workload and latency required. Operators such as META who are servicing the public would of course, like to see edge networking as close as possible to the consumer. However, there are other workloads such as LLM which are not as time-sensitive and can be easily located away from inhabitants optimising access to renewable power for example.
Another thing that you do need to factor in is the geopolitical situation and whether the geography is seen as safe. For example, due to the war in Ukraine the Baltic states who tipped as the future digital hubs are no longer considered as attractive as long as the threat from Russia exists. If you want to hear the hot tips of where to locate in Europe, then contact me.
Automation
Last but no least, with automation being one of the main areas of interest I thought the comment by Tony Rossabi from OCOLO was worth sharing. He said that the growth in DC over the last ten years will be exceeded within the next three years. Most DCs do not have APIs— so we are a cutting-edge industry but need to move to more automation. This way we can move the service closer to the use and consumable.
To conclude, like in other parts of the telecoms ecosystem international wholesale is changing, driven by the demand primarily from the US to support AI workloads. There is a lot to be done and a lot of opportunity. The smart money is on those who are new entrants without legacy or legacy providers who can transform and move on quickly.