What Hannover Messe 2025 revealed about AI

As the fifth and final day of Hannover Messe 2025 drew to a close this week, the roughly 130,000 attendees were left with a great deal to reflect on. Perhaps unsurprisingly, artificial intelligence (AI) was placed under the spotlight, with everything from AI co-workers to autonomous robots demonstrating what the future of industry holds. But with robots welding, chatbots assisting, and algorithms optimising, one question at Hannover Messe remained: how can all this data get to where it’s needed quickly enough?

Ivo Ivanov, CEO of DE-CIX further explores.

From the Cloud to the factory floor, the industrial machine is becoming an interconnected ecosystem of automation, digitisation, and AI-powered efficiency. Germany, the home of Hannover Messe, is ranked third in the world for ‘AI readiness’ according to research by the IMF, and more than 58% of businesses around the world are committed to stepping up their AI investments in 2025. Make no mistake – this is the year of AI. However, because algorithms are now being incorporated into every single step of the industrial process, from robots transporting components on production lines to ‘cobots’ welding and drilling – connectivity has become the gatekeeper of innovation.

Before we get to the connectivity question, however, let’s first take a look at some of the AI technologies showcased here at Hannover Messe 2025.

Picking up steam

Steam is an inevitable by-product of industrial innovation. Sometimes it’s used for drying or heating, and other times it’s regarded as a waste product to be minimised. Either way, without it, the chemical industry would grind to a halt. That's because water in its gaseous state is a basic ingredient in most processes, whether it's for operating reactors, separating substances, or distilling them. The problem is that steam is energy-intensive, and once it has been generated, it can only be stored for a short period of time. As strange as it may sound, AI may be the answer to this conundrum.

Process X, a new ‘data room’ initiative spearheaded by a small coalition of industry groups, attended Hannover Messe to demonstrate how this may be possible. Their focus was on ‘predictive steam production’ – with the help of AI algorithms and machine learning, steam demand can be predicted on the basis of historical data, production plans, and real-time data from the plant, making it possible to reliably calculate steam requirements so only what is needed is produced. This could be a major step forward for sustainability and dramatically lower energy and production costs – but it will, like almost every use of AI modelling and inference – depend on connectivity.

From ‘Cognibots’ to chatbots

One big showcase at this year’s event was Delta’s new ‘Cognibot’ – a new collaborative robot with advanced cognitive capabilities including voice recognition, 3D machine vision, and multi-modal AI interaction. Onlookers saw first-hand the impact this level of machine could have in digital twin and production settings. What was not visible, however, were the algorithms that process, analyse, and evaluate data virtually hidden in large data centres or small edge units that allow innovations like the Cognibot to recognise patterns, calculate probabilities, and improve and perfect themselves in a self-learning and self-organising way.

Robotic innovations may have had the wow-factor, but chatbots and assistants more than held their own. Beckhoff Automation unveiled its next-generation TwinCat CoAgent – an intelligent AI assistant which supports logic development, I/O configuration, and human-machine interface creation. It can even draw up advanced graphical user interfaces from a simple hand-drawn sketch. Across the hall, Festo showcased how it integrates design and manufacturing data into its own virtual assistant, offering contextual access to thousands of documents that can be discussed in a human-like manner. SAP also presented Joule, a new Generative AI assistant designed to understand business context and use natural language processing (NLP) to provide proactive, conversational insights.

Robots and chatbots like the ones above may not occupy the same physical space – or even take up any physical space at all – but one thing they do have in common is the need for high-performance, low-latency connectivity. Without this, their usefulness either diminishes or they become so unreliable that any investment would be a high-stakes gamble.

Connecting the dots

This year’s Hannover Messe offered some breath-taking demonstrations for all to see, but – like dark matter, the mysterious invisible energy holding the universe together – the real story was more about what couldn’t be seen. Connectivity is the enabling force that connects robots, chatbots, controls, storage systems, conveyer belts, and factory floors with the data superhighway that powers them. The time it takes for data to get from one network point to another has become the maker or breaker of such technologies.

According to Meta, AI models spend roughly 33% of their time waiting idly on networks. In the context of this next generation of industry, that means new innovations are already doomed to waste a third of their potential before they’re even deployed. In a setting where time so often is money, that’s not optimal, and maybe not even sustainable from an ROI perspective. This is where network peering comes in. Digital infrastructure operators like DE-CIX are constantly seeking ways to minimise throughput times on network and bring data, applications, and users closer together. If the paths along which information travels between the Cloud, Edge, and shop floor can be physically shortened, data – and AI – will be able to do its job faster. Direct peering at an Internet Exchange or an AI Exchange makes this possible, reducing the number of network ‘hops’ data needs to make in order to get from A to B. The result is high-speed, low-latency connectivity that is free from congestion and has resilience and redundancy measures baked-in.

The clarion call from Hannover Messe was that its ‘full steam ahead’ for AI, but the AI train can only move so fast on legacy tracks. Before the industry fully embraces the next-generation of cognitive robots and intelligent virtual assistants, we need to turn our attention to the connectivity infrastructure that will not only make them possible – but make them the best they can possibly be.