A SnapLogic survey released today, The Code to Unlock Gen AI, found that 63 percent of surveyed companies reported ‘moderate’ to ‘severe’ negative effects of technical debt – which are the costs associated with maintaining and updating outdated, legacy systems. These negative effects include the creation of data silos, diverted resources and reduced effectiveness.
According to Jeremiah Stone, CTO of SnapLogic, “Legacy systems are characterized by programming languages with low available expertise and architecture patterns which are no longer considered state of the art, and which increase the burden of ramp up time for engineers who must support the systems due to the need to learn the specific non-generally in use technology of the legacy system.”
Put another way, legacy systems and technology debt are the gifts that keep on giving. “Fifteen years ago, C++ and COBOL based applications were legacy and were being migrated to Java. Now Java is legacy, so those are getting migrated,” said Hansa Iyengar, Senior Principal Analyst, Enterprise IT with Omdia. “There will always be a cycle of 15 to 20 years where one set of technologies are no longer relevant.”
As for why tech debt accrues, Iyengar says it’s a combination of factors. Take, for example, the aging of the workforce – people who worked on COBOL retired 15 years ago; those who worked on Java will likely retire in the next 10 or so years. Additionally, documentation has gone missing, and nobody knows what will break if a system is removed. And there’s an element of job protection involved, too – nobody wants to be ‘that guy’ who broke order entry.
But not updating also carries risk, since those older systems may be more vulnerable to attack particularly as workarounds are introduced (e.g., via APIs) that allow modern applications to access the data in those legacy platforms. (For more detail, check out this episode of the Omdia Digital Leaders Podcast, hosted by Iyengar.)
“Companies that are less than 20 years old probably won’t have this exact problem because they are already on newer platforms and applications,” Iyengar said. “But, technical debt accrues every time new code is introduced, so there is a lot of it. The older the systems, the greater the debt would be. Most organizations pretend tech debt is under control when it is actually not. We call it the underwater portion of an iceberg.”
Iyengar commented further that these older companies that have been in existence for 60 to 70 years may carry a titanic amount of legacy and tech debt. Because of that, they may be less aggressive when modernizing their legacy systems because they just don't know what will break.
Updating legacy systems is also expensive. SnapLogic’s survey found that in 2024 IT decision makers asked for average budgets of $2.7 million USD to upgrade legacy systems. The report also found that:
“Quantifying legacy modernization is tricky as the initiatives either get combined with transformation programs or are absorbed into ‘run the business’ costs,” said Iyengar. “Our IT Enterprise Insights survey (which covers ~6k IT decision makers annually) consistently shows that around 65% of IT budgets go toward ‘Run the business’ while ~17% are ‘Transformation’ type work. Around 28-30% of IT budgets are spent on maintaining/upgrading legacy tech.”
One key reason to upgrade legacy tech and older software languages is they are no longer able to work at the speed that is required right now. SnapLogic cites another in its report: 57% of organizations surveyed plan to update up to 50% of the legacy tech within their organizations to use Gen AI technology.
According to SnapLogic’s Stone, Gen AI technology (as defined in the survey) covers everything from open-source and commercial LLMs to applications that serve as Gen AI support features for tools like Grammarly. The report found that the top intended use cases include customer support (chatbots and virtual assistants), customer databases (data aggregation and analysis) and applications that can be used by the IT department.
“The more we see enterprises try to scale things like Gen AI, for example, they need to have the solid foundations in place otherwise it will never work the way they expect it to work,” Iyengar said. “For that they need to modernize, and technical debt is one of the biggest problems they need to solve.”
SnapLogic’s research was conducted in June 2024 by Censuswide on behalf of SnapLogic. 750 IT decision makers, working in organizations with 250+ employees, were surveyed across the US, UK and Germany.
For more on Omdia’s research, check out the survey Iyengar mentioned: IT Enterprise Insights Spotlight Service, Global, 2024, as well as two reports published by Iyengar's team which provide more insight into technical debt and systems integration and application services (both require an Omdia subscription). Iyengar said further that the topic of legacy systems and tech debt will be a bigger part of her team's research in calendar 2025.
