"Technical Debt": Insufficient Company Investments in IT

During the ongoing World Economic Forum in Davos, the head of the consulting firm Publicis Sapient, Nigel Vaz, pointed out that companies are investing too little in the expansion of IT systems. This leads to the buildup of “technical debt.” He told dpa: “Often, 80 percent of the funds go into the largest components of the infrastructure, only 20 percent into innovation.”

Banks, in particular, are reportedly affected. They still operate central mainframes. However, companies with outdated ERP systems for managing business processes also face this problem. This technical debt leads to increasingly difficult access to data and hinders the integration of more modern technologies. According to Vaz, even Artificial Intelligence does not fundamentally solve the problem: “If you don't modernize the core systems, you're just trying to pretty up the situation with AI.”

The outdated systems cause complex handling, especially in banks, healthcare, and retail. Many programs are written in COBOL, for instance, and are millions of lines long. New developers find the code difficult to understand. According to dpa, Publicis Sapient uses the platform “Slingshot,” which transforms old program code into modern technologies like Java or React. Using AI, a project could be realized in less than three years that was previously estimated to take ten years, Vaz further explained. However, fragmented data sets and different regulatory frameworks complicate the use of artificial intelligence.

Challenges and Costs

Vaz criticizes that the shortage of skilled workers and costs are the central challenges. Few young developers still learn COBOL, which is why “companies have to retain experienced teams at great expense.” However, the savings from modernization could partially finance the investments. Nevertheless, companies are under pressure due to tight budgets and high innovation needs, and the German banking and retail sectors in particular have a great need for modernization.

AI tools can also help with migrating old code. IBM, for example, has watsonx Code Assistant for Z in its arsenal, which translates COBOL to Java using AI. COBOL is not entirely dead, either. Current compilers for the language are still being released, such as the open-source compiler gcobol in 2022. COBOL programs are more likely to be modernized than replaced. However, the fact that COBOL represents a ticking time bomb was already clear in 1999. At that time, it was about the two-digit year numbers that were often used on mainframes to save storage space.

(dmk)