Over the past few years, investments in AI initiatives have surged. Yet in conversations with IT leaders, Heads of Data, and Operations executives, a different reality is emerging: Many AI initiatives stall after early enthusiasm — not because of a lack of ideas, but because the foundation isn’t there. Legacy IT landscapes cannot keep up with the demands of modern AI, while pressure to innovate continues to rise. IT modernization is no longer optional. It is the prerequisite for turning AI into measurable business impact. The real question is: How do you move from rigid legacy systems to a scalable, AI-ready IT and data landscape?
With this article, we’re kicking off our blog series “IT modernization: Why most AI initiatives fail without a solid data and application architecture.”
Results from our modernization projects:
Many IT decision-makers describe the same dilemma: Up to 80% of IT budgets are tied up in maintaining and operating existing systems (Gartner 2025). These systems have often evolved over decades — technologically outdated, fragmented, and siloed. As a result, financial resources are tied up, and skilled teams are occupied with keeping systems running instead of building new capabilities. This has direct consequences. Data-driven services are difficult to implement, scalability remains limited, and time-to-market for digital products increases.
In addition to these structural challenges, several external factors are accelerating the need for action:
Delaying IT modernization is therefore not a viable option. Legacy systems eventually reach a point where they can neither scale nor be maintained efficiently. At the same time, cost pressure continues to increase.
Take the 5-minute self-assessment: evaluate five dimensions and receive a prioritized set of recommendations.
Another striking observation is that 50% of all AI projects fail after the proof of concept. The root causes are rarely algorithmic, they are structural: poor data quality, fragmented data infrastructure, missing governance. Without clear ownership, transparency, and access rules, data becomes unusable for AI.
Key questions remain unanswered:
Without a unified data strategy and a modern platform that integrates data from production (OT), ERP systems (IT), and other sources, AI remains a theoretical construct. True AI readiness does not come from adopting new tools. It is built by establishing a consistent data architecture that serves as a reliable foundation for machine learning models and data-driven applications.
More importantly, AI readiness is not about getting a single use case into production. It is about building a foundation where the second, third, and tenth use case can be delivered faster, more cost-efficiently, and reliably brought into operation.
A robust IT and data foundation ultimately determines how quickly new product variants or business applications can be rolled out across sites, which operational decisions are made based on data rather than intuition, and whether manual handoffs between MES, ERP, and the shopfloor can be eliminated. All of this requires one thing: consistent IT modernization.
IT modernization is often misunderstood as a purely technical exercise. In reality, it is a strategic transformation that enables new business models.
This includes modernizing applications, for example by moving from monolithic architectures to microservices, building cloud-native infrastructures, and establishing a centralized data platform that serves as a foundation for analytics and machine learning.
Across Device Insight projects, this translates into tangible outcomes:
One project we are particularly proud of: a global renewable energy manufacturer replaced its VMware/Cumulocity-based platform with a cloud-native Azure solution in collaboration with Device Insight. The result: up to 70% lower OPEX, scalable processing of high-frequency data streams, and a platform that brings new AI use cases into production in weeks, not months.
For many organizations, the idea of modernizing an entire IT landscape can feel overwhelming. In practice, however, successful transformations rarely follow a big-bang approach. Instead, they start with the systems that matter most. Focusing on critical applications allows companies to create immediate value while building a scalable foundation step by step. This approach also reduces risk and makes progress measurable.
Explore how IT, operations, and data teams approach modernization in practice.
Including case studies, migration scenarios, and a proven delivery model –
based on 200+ industrial projects in our Executive Guide (PDF).
IT modernization should not be seen as an additional burden, but as a business necessity. Every AI initiative ultimately depends on the quality of the underlying IT and data architecture. Companies that postpone modernization risk rising costs, limited scalability, and failed AI initiatives. In contrast, those that invest early in modern architectures create the conditions for long-term competitiveness.
The logical starting point is a structured Modernization Assessment of the current IT landscape. Within a few weeks, this provides clarity on existing systems, identifies technical debt, and defines a realistic roadmap for modernization. From there, the transition to an AI-ready architecture becomes a manageable, step-by-step process.
Don’t wait until legacy systems slow you down. Start the conversation now.
At Device Insight, a KUKA Digital company, we understand the realities of complex industrial environments – from legacy systems that cannot simply be switched off to shopfloor integrations, OT protocols, and production logic. For nearly 25 years, we have been helping companies connect machines, unlock industrial data, and turn it into real-world impact. We are your partner for Industrial Intelligence.
Do you have questions about your modernization roadmap or want to explore a first assessment? Book a 20-minute session with us – we’ll show you where your biggest levers are and what a realistic first step looks like.
About the author:
Ingo Forstner is Senior Portfolio Manager at Device Insight, responsible for the strategic go-to-market direction. His focus: helping industrial companies evolve legacy IT landscapes into scalable foundations for AI-driven business models. Previously, he held positions at Linde Group and MAN Diesel & Turbo.
With the new Broadcom licensing model, a move to Azure often pays off within 12–24 months — driven by a shift from CAPEX-heavy infrastructure to OPEX-based models
The integration of IT and OT data into a modern architecture (e.g., a data lakehouse) that enables fast and reliable access for ML models.
By going beyond virtualization and adopting containerization. This creates the flexibility and data availability required for AI workloads.
Outdated legacy stacks lead to innovation bottlenecks, increased security risks (e.g., NIS2), and growing skill shortages as experienced specialists retire.
A structured 3–6 week modernization assessment analyzes your IT landscape, identifies technical debt, and defines a clear roadmap.
If your legacy systems cannot provide structured machine data, critical operational decisions are made without the full picture.
Modernization creates the data foundation for predictive maintenance, enables the rollout of new product variants across sites, and eliminates manual handoffs between MES, ERP, and the shopfloor.
Because each new initiative starts from scratch: a new data pipeline, new governance discussions, a new deployment path. A modernized platform reverses this. The second use case builds on the first. AI readiness is repeatability – not a single successful pilot.
