The IT sector is grappling with a creeping and potentially disruptive problem of knowledge loss. For a topic like European digital sovereignty, alignment with modern frameworks is necessary. This is almost impossible without having in-house, specialized software skills. That’s why the Belgian company Klarrio is calling for investment in broadly educated engineers now to prevent companies from falling behind.
Many organizations are considering moving away from proprietary American cloud services. Think of financial institutions and utilities. A European alternative can be attractive. To implement this effectively, many companies quickly conclude that they must use open source. This requires engineers who are also skilled in maintenance and further development. If the highest possible level of digital independence is pursued, outsourcing typically isn’t an option. Ideally, you need to have the software skills readily available internally.
However, according to Kurt Jonckheer (CEO), Bruno De Bus (CTO), and Lieven Gesquiere (Senior Director R&D) at Klarrio, an acute shortage of knowledge and skills is looming. And this potentially major problem has multiple causes.
Education is deteriorating, the need for a ‘full-stack’ thinker is increasing
The foundation of the impending shortage is already being laid in the classroom. While the complexity of the field has increased exponentially over the past decade, degree programs have switched from building a solid foundational background in computer science to a more granular focus on designated specialty area—most notably data management, cyber security, and now AI. Consequently, new graduates increasingly lack the breadth of knowledge required to smoothly transition to the workforce.
Gesquiere sees daily how lack of a strong educational foundation clashes with reality. “For certain critical components within modern architectures, people are still needed who can reason from the bottom to the top of the software stack,” Gesquiere explains. “That should not be a rarity; it’s an absolute basic requirement for building and maintaining reliable systems.”
According to Gesquiere, incoming engineers increasingly lack essential fundamental knowledge. “Curricula have grown in breadth but often lack depth. Networking has been relegated to an elective in some IT master’s programs, whereas for engineers moving to the cloud, networking should be the absolute foundation.” Additionally, he points out the gap between theory and modern practice. “Cloud-native frameworks such as Apache Kafka, Flink, and Spark, or modern programming languages such as Rust and Go, are barely covered in many programs. There are simply too few lecturers with hands-on operational experience in these technologies. Students must acquire the practical knowledge themselves, which creates a structural knowledge gap from the very beginning.”
AI does not replace experienced teams
This knowledge gap leads to IT teams full of specialists with a very focused but limited perspective. De Bus warns of the disruptive consequences of this on the work floor. Within complex software environments, having a few individual experts is simply not sufficient, he says.
“The whole team must collectively have a common mental model of the entire platform in order to build upon it,” De Bus emphasizes. “If you only have experts for specific, isolated components, it doesn’t really help. If that person leaves, the rest of the organization is suddenly left with software that absolutely no one understands anymore.”
Without this broad, shared understanding, a team loses the ability to make good decisions regarding maintenance or modifications. De Bus draws a parallel to fundamental IT knowledge: “An understanding of many programming functions in cloud networking used to be passed on as a matter of course. Now we see teams deploying their own or AI-generated code into production without assessing the broader consequences for the infrastructure. Recently, this has caused major parties’ data centers to go completely down.”
The IT assassin
It is precisely this superficial knowledge, combined with a lack of overview and blind faith in automation tools that leads to software crashes. AI cannot simply compensate for the backlog. “AI may be able to detect security issues or generate code, but it introduces new, complex problems just as easily if the output is not evaluated very critically,” adds Jonckheer. “That is precisely why we need the broadly trained expertise that we are currently in danger of losing.”
If no one within the organization understands what the generated system does and how it works under the hood, then software crashes will inevitably occur. According to Klarrio, this uncontrolled code will lead to massive, system-critical crashes in the foreseeable future.
Also read our earlier article about how AI threatens to introduce software crashes.
Draining your own breeding pond
Despite the clear need for in-depth human knowledge, Jonckheer identifies a paradoxical reaction in the market. Driven by AI hype, many companies have stopped hiring junior developers, assuming algorithms can take over basic work. But the consequences of doing so extend far beyond the savings on payroll.
“We are creating a gigantic societal problem for the future with this,” warns Jonckheer. “Why would a 17-year-old today still choose a grueling, demanding technical degree of five to six years when the message everywhere says junior positions are being rendered redundant by AI?”
Jonckheer also emphasizes that the influx of juniors today lays the foundation for what will be the seniors of tomorrow. Anyone who cuts off this pipeline will face an unsolvable generational problem in a few years. “For the flow of knowledge, you need an unbroken chain—from young people in secondary school, through higher education, to experienced professionals who are given the time to gain operational experience,” Gesquierre emphasizes. “If you break that chain, you end up with a market full of unaffordable experts, and no one who can bridge the gap to practice anymore.”
Without knowledge, there is no independence
As the desire for European digital sovereignty grows, the lack of knowledge takes on an urgent, geopolitical dimension. Both the public and private sectors want to become less dependent on American hyperscalers, a movement Klarrio applauds, but which also places enormous demands on the European labor market.
“More and more of our customers, certainly in critical sectors such as financial institutions and the utilities industry, are considering switching to a plan B—a purely European cloud,” says Jonckheer. This is partly driven by extreme commercial pressure and a hunger for data from the major cloud providers. He cites the recent example of Atlassian, which intends to use the tools Jira and Confluence as the standard to train AI models, unless customers pay thousands of euros for an ‘opt-out’. “If all confidential business content ends up in such generic models, you lose control over your intellectual property. We must stop that kind of development,” Jonckheer contends.
However, the transition to the sovereign cloud is complex. De Bus highlights the technological side of this shift: “A sovereign, open-source environment does not offer the out-of-the-box functionalities you get with an American cloud service. The complexity shifts. We are talking about bare metal servers, container orchestration via OpenShift or SUSE Rancher, complex Kubernetes environments, and increasingly stringent compliance legislation. These are not processes that you manage with the push of a button.”
The transition requires engineers who not only write code, but who understand, can build, maintain, and take full responsibility for open-source solutions with 24/7 availability. “If Europe really wants to become less dependent, we must realize that new platforms must be built and secured by ourselves,” states Gesquiere. “It requires an army of highly skilled technical people with capabilities that must be protected and nurtured within the EU.”
Independence begins internally
The crux of the story is that digital independence goes hand in hand with internal knowledge. Companies that continue to outsource their core software infrastructure and view it as an afterthought rather than a core competence ultimately lose the ability to differentiate themselves.
This naturally applies to Klarrio as well, and the management team is addressing potential problems before they ever have a chance to materialize. When the company’s growth doubled post-COVID, their regular recruitment channels proved to be exhausted. Instead of passively waiting for a failing labor market, Klarrio launched its own academy in 2022. Over a period of nine months, the company internally trained and paid external contractors who lacked any recent or extensive data engineering background. There was, however, a requirement that participants had previous experience with data, ensuring that a certain level of knowledge was already established.
This approach has since evolved into continuous, company-wide programs. Gesquiere explains how theory and practice come together in this: “Last year, we rolled out a company-wide AI curriculum, in which teams had to set up a completely new, fictional company, partly with and partly without AI. This allows us to objectively measure efficiency gains as well as to immediately define strict rules. For example, we do not use US-hosted AI models for customer data.”
Insights from this project are now flowing into an ongoing Advanced Development Program, in which employees are structurally given the time and guidance they need by team leaders and architects to experiment with new technologies. In this scenario, AI functions are a learning tool, not a replacement for critical learning processes. The program is currently in its initial stages, with a more aggressive company-wide rollout planned for later this year.
Offering programs to existing employees can certainly help achieve the right skill level, but it’s also important to bring on people who demonstrate the talent and ability to develop a high level of skill from the very beginning. According to De Bus, the right mindset is primarily the decisive factor when attracting employees. “We are not necessarily looking for the perfect CV. We are looking for people who are curious. People who want to continuously learn new things and dare to go deep.”
Jonckheer added: “Age plays no role whatsoever in this. We see twenty-somethings who are already completely set in their ways of thinking and fifty-somethings who passionately fathom new frameworks,”
Knowledge as the beating heart of digital
The adoption of AI and the drive for European independence do not make software engineers redundant. On the contrary, they make the need for in-depth, broad-based tech professionals greater than ever. If companies and governments continue to expect complex digital infrastructure to keep running effortlessly while education narrows and the influx of talent is blocked, a clash is inevitable.
The good news is that the tide can be turned, provided organizations are willing to bear the responsibility for knowledge building themselves. After all, knowledge isn’t a professional service you can outsource with impunity. It is the heart of your digital raison d’être. Anyone who ignores this will sooner or later find themselves completely stuck and faced with a potentially devastating software infarction.