Service providers and data center specialists are preparing for a wave of AI, characterized by heavy and especially dense workloads. In Belgium, everyone is ready, albeit mainly preventively.
“In every presentation we give, we have to talk about AI,” says Robbert Lambrechts, Technical Presales Specialist at Lenovo. “It’s AI this and AI that, but at the same time, we see that there are two important types of customers in Belgium.”
Lambrechts shares his vision on the state of AI in Belgium as part of a roundtable discussion on data center infrastructure. Also joining are David Louis, Managing Director of Digital Realty in Belgium, Hendrik Devos, Technical Sales Leader at Kyndryl, Kenneth Deviaene, Senior Solution Sales for Combell, and Thomas Van Tricht, Key Account Manager Cloud & Service Providers at Schneider Electric.
Sniffing, Cycling, and Stepping on the Gas
“Belgium is a country of followers, not pioneers,” Louis remarks, while everyone nods in agreement. “In my opinion, AI will probably be consumed initially within a cloud-like model. People want to first explore the possibilities before making large investments.”
That’s what Lambrechts was also referring to. “There’s a stereotypical Belgian customer who rides his bicycle in the wheel of others and sees if he can keep up.” This customer is interested in AI but has no intention of taking a pioneering role.
“And then there’s also another type,” Lambrechts observes. That one jumps into his Ferrari and hits the gas pedal towards the autobahn. In other words, there are two trajectories. Many organizations are cautious, but there are also parties that are actually asking questions about clusters full of GPUs.
Lagging Behind Other Countries
This demand doesn’t come out of nowhere. Across the ocean, such clusters are actually being deployed. Van Tricht sees this very clearly. “We are an international player in the field of data center infrastructure. What we don’t really see in Belgium today is already happening in America.”
Deviaene experiences the two speeds firsthand from his position at Combell: both the sniffing around and the demand for action. “Customers are asking us questions. There are parties creating AI tools, and they need to run somewhere. However, that’s not so straightforward. For such workloads, you need to be able to offer very sophisticated infrastructure. In many cases, there’s no solution for that in Belgium yet.”
For such workloads, you need to be able to offer very sophisticated infrastructure
Kenneth Deviaene, Senior Solution Sales Combell
He notes that Belgium is falling behind in this respect. “We see that in other countries, there are often more sophisticated solutions in place.” He does understand the situation. “From a commercial point of view, we do see a demand in the market. From a technical point of view, AI infrastructure is expensive and complex. Technical teams also have to manage this on top of their existing workload.”
AI, With or Without ROI
Devos from Kyndryl makes another observation. “Actually, until recently, AI was a very disappointing technology,” he notes. “According to our own research, very few companies actually succeed in rolling out their use cases with a positive return on investment. Despite the evolution and the fact that everyone is interested in AI, this remains the exception.”
“We actually see that customers who do take the leap prefer to run AI on-premises,” Devos continues. “They fear peaks in usage and thus costs that are very difficult to control in the cloud. With their own AI infrastructure, on-premises or in a colocation data center, the limits are much clearer.”
We actually see that customers who do take the leap prefer to run AI on-premises.
Hendrik Devos, Technical Sales Leader Kyndryl
Devos refers again to their own research, making Louis beam: “We expect AI to increase data center capacity by fifteen percent year over year.”
Louis is ready in any case. Digital Realty has modern data centers around Brussels that can handle high-density AI workloads. Lenovo, for its part, has developed cooling solutions that allow powerful installations to run in traditional air-cooled data centers as well. “Depending on the customer’s needs, we can accommodate their type of workload,” Louis emphasizes.
Open Arms
The consensus around the table is that the high demand for AI workloads in Belgium is currently absent. However, there is interest: organizations are increasingly exploring concrete possibilities. Looking beyond borders, it’s clear that the scenarios for which IT infrastructure specialists are preparing are not hypothetical.
read also
Sea cooling in a closed circuit: the future of data centers?
“We don’t know exactly when the acceleration will come to us, but we know it can happen quickly and we all want to be ready for it,” says Van Tricht. “That’s the role we want to play. When customers want to run AI workloads, we’re prepared and can assist them.” When the demand for AI infrastructure materializes, the entire round table is ready to welcome customers with open arms.
This is the first editorial article in a series of three on the theme of AI in practice. Click on our theme page to see all articles from the round table, the video, and our partners.