Why legacy systems are holding you back, and how to fix it

Most organizations still spend the majority of their IT budgets maintaining systems built for a different era. Legacy infrastructure is expensive to run, hard to evolve and increasingly difficult to justify. Budget that should fund next-generation capabilities gets consumed keeping the old ones alive. The question is no longer whether to modernize, but how to do it well. 

The organizations getting ahead are already moving to platforms built to handle large data volumes and the demands of AI. At DXC we’re already implemented frameworks, like LabX and AdvisoryX, to help clients make that move, from benchmarking their starting position to executing modernization with precision. 

The Urgency to Cut Technical Debt 

Maintaining legacy systems can consume up to 80% of an organization’s IT budget. That means burning spend on aging infrastructure, unoptimized applications and extended support agreements, while adding complexity to any attempt at adopting new technology. This means that every opportunity to claw back efficiency matters. 

While modernizing has clear benefits, the path forward is not always obvious. Technical debt is significant enough that 99% of executives list it on their corporate risk register, yet there is no common standard for measuring its impact. Without consistent benchmarks, organizations have no reliable way to judge whether their modernization investments are paying off. 

This narrow view of technical debt – reduced to old code or end-of-life hardware – is part of the problem. That’s why we developed a tech debt analyser that gives enterprises a scored assessment and recommended next steps across a broader range of categories. 

A rigorous assessment starts with alignment to existing standards bodies, which broadens what counts as technical debt and surfaces opportunities that would otherwise go unnoticed – and a route forward that is incremental, but has major positive implications in the long term. Modern systems bring lower maintenance costs, fewer outages and reduced cybersecurity exposure. Most importantly, they free up budget to invest in what comes next. 

By reducing spend on legacy systems, enterprises can pursue a broader approach: consolidate to innovate. What’s recovered from smarter IT investment and shrinking service contracts can be reinvested in innovation. The newer systems also handle the performance and data demands of AI more effectively, with the scalability that modern workloads require. 

Find the Problem AI Can Solve 

With technical debt addressed and budget recovered, the natural next question is: where does AI fit? Efficiency and scalability matter, but those gains alone will not keep a business competitive. The innovation agenda must guide the modernization process, not follow it. 

The answer has to start with the business problem, not the technology. Many enterprises are rushing toward AI without the data foundations or technical architecture to deploy it well, and the gap between intent and execution is significant. According to the now famous MIT study, 95% of AI projects either fail or fall short of their goals. One reason is poor integration with existing systems, which modernization should address. The other is that AI does not fit real workflows. 

Enterprise systems exist to help organizations gain insight, work better or scale faster. AI should be no different: a means to an end, not the destination. Yet many organizations begin AI deployment without a clearly defined business outcome. 

The right approach is to work backwards: define the problem, outline the desired outcome, then determine which technology – machine learning, GenAI or deep learning – fits best. The technology choice should always follow the problem, not lead it. 

Take a large IT services firm where cybersecurity excellence was non-negotiable. In a threat environment increasingly shaped by sophisticated AI – including growing use of tools like Anthropic’s Claude in enterprise security workflows – the pressure keeps rising. A global cybersecurity skills gap running into the millions was a concrete, urgent risk. The response was a targeted investment in AI agents to augment security operations centers. The result: a 67% reduction in investigation time, 80% fewer alerts requiring human review, and a 62% drop in incidents. 

The AI worked because it addressed a specific, understood problem. That’s the only sound basis for AI investment. 

Make Distributed Teams Work 

Remote and hybrid work is here to stay. Even employees with office access spend much of their time collaborating remotely. But distributed teams consistently collaborate less effectively and generate more IT friction than co-located ones. 

The answer is not a return to the office, but smarter design of physical and digital working environments. AI can genuinely help here, not by replicating office dynamics, but by removing the friction that slows distributed teams down. 

One organization with 70,000 engineers across 70 countries was struggling with exactly this challenge. They implemented a knowledge management system, powered by AI, that could understand engineering context and connect people to the right expertise at the right moment. The result was a 60% reduction in incidents. 

The lesson holds for any AI application: start with the human problem. The technology choice will follow. 

The Strategic Opportunity 

Every organization carries technical debt. The difference is whether you treat it as a risk register entry or an active lever. A disciplined approach, scored against benchmarks and tied to specific business outcomes, drives better IT investment, faster AI adoption and more productive ways of working. That’s not just efficiency. It’s how you build for the long term.