Elnion

When organizations discuss reducing IT costs, the conversation often begins with assumptions. Mainframes are expensive. Cloud platforms are cheaper. Legacy systems consume too many resources. Hardware upgrades are the only path to improved performance. In reality, the most effective cost-reduction initiatives rarely begin with infrastructure. They begin with efficiency. For decades, I have worked with organizations running mission-critical applications on IBM Z systems.

When an opportunity to discuss COBOL in a public forum arises, it proves to be an illuminating experience. The 2026 GSE Nordic conference in Helsinki was no exception. Incomparable, or contemptible?

By design, the most important technology in the world is often the least visible. No splashy user interfaces. No viral launch announcements. No breathless headlines about disruption. And yet, every single day you relied on systems running on the IBM Z mainframe. You didn’t notice. You weren’t supposed to. When you withdrew cash, swiped a credit card, checked your insurance coverage, booked a flight, or even received a paycheck, there’s a good chance a mainframe was involved. Not as a relic.

Every few years the IT industry rediscovers something that experienced practitioners already know: moving data is difficult. You would think that by now organizations would approach data migration projects with the respect they deserve. After all, the industry has been migrating data since the earliest days of computing… from hierarchical databases to relational systems, from on-premises platforms to distributed environments, and now from legacy infrastructure to the cloud.

Enterprises today are investing heavily in analytics platforms, data lakes, and artificial intelligence initiatives. Cloud scalability, modern tooling, and machine learning frameworks are often portrayed as silver bullets that will finally unlock the value buried in corporate data. Yet despite these investments, many organizations struggle to realize meaningful business outcomes from analytics and AI. Dashboards lag, models underperform, and decision-makers quietly lose trust in the data.

For the global economy, the firewall is the digital equivalent of a castle wall, the fundamental barrier separating order from chaos, legitimate business from sophisticated criminal enterprise. When that wall is breached, the ramifications are immediate and devastating.

For decades now, the “death of the mainframe” has been confidently predicted by vendors, analysts, and industry pundits. Yet, here we are: the mainframe continues to serve as the backbone of mission-critical computing for banks, insurance companies, healthcare providers, retailers, and governments across the globe. It is not only still here—it is thriving. And yet, the drumbeat of “mainframe modernization” never seems to stop.

Artificial intelligence is no longer a speculative concept whispered about in technology circles; it sits on the desks of architects, woven into software tools and creeping into project workflows. The architecture profession, long defined by its combination of creativity, precision and regulation, finds itself at a crossroads as AI tools rapidly reshape how buildings are imagined, assessed and delivered.

For decades, the central gravitational point of enterprise computing has been the data centre. From the earliest mainframes encircled by teams of operators to sprawling cloud platforms accessed through high speed networks, the story of information technology has often been one of scale, consolidation and centralisation. Yet the rise of edge computing is beginning to flip that logic.

When most people talk about artificial intelligence (AI), the conversation tends to gravitate toward algorithms, neural networks, and large language models. But beneath all of that is a less glamorous yet absolutely essential foundation: metadata. Without rich, accurate, and well-governed metadata, AI cannot truly deliver on its promise.