Industrial AI: Showdown in Hanover – A Trap for SMEs?

KI & Automatisierung · 13. April 2026 · Omer

The Hannover Messe celebrates Industrial AI with €800 billion. But is this summit an opportunity or a dangerous distraction for SMEs? What you need to do now.

Do you know that sound? That gentle, almost silent hum of a six-axis robot, polished to a high sheen, performing a ballet choreography on a trade fair stage. Next to it, a DAX executive talks about disruption and paradigm shifts. Applause. Impressive, no question. But last week I was in a medium-sized machining company in Sauerland. There, the predominant sound was the loud rattling of a 20-year-old CNC milling machine and the foreman's cursing because the damn order data from the ERP system didn't match the drawing again. And it is precisely here, in this chasm between trade fair show and workshop hell, that the battle for Germany's industrial future will truly be decided – not on a stage in Hanover.

Don't get me wrong. When the concentrated power of German and European industry gathers in Hanover in April 2026 under the banner of "AI in Industry," it's an important signal. Siemens CEO Busch, Christian Klein from SAP, Telekom's Höttges, flanked by people like Julie Sweet from Accenture – that's the A-list. When they formulate an "industrial policy appeal" and talk about scaling Artificial Intelligence with Chancellor Merz and half a dozen ministers, it carries weight. The announced "Made for Germany" initiative, with over 800 billion Euros in pledged investments by 2028, sounds like the long-awaited fanfare. One would like to applaud. But I cannot. Because in my experience, such summits quickly become echo chambers for the big players, while SMEs – the true backbone of our economy – are left out in the rain, wondering when some of that warm money will finally reach their watering can.

Why most people approach Industrial AI backwards

The thing is: the hype around Industrial AI leads to a fatal misconception that I see almost daily. CEOs and production managers believe AI is a technology you buy. Like a new machine. You order an "AI package for Predictive Maintenance" from one of the big providers, the IT service provider installs it, and – boom! – production suddenly runs as if optimized by magic. That is, with all due respect, nonsense. What happens instead is the birth of another "pilot project" that quietly burns out in the "proof-of-concept" purgatory after six months. Why? Because the fundamentals are not right.

Industrial AI is not a product. It is the result of processes. More precisely: of clean data and standardized processes. And honestly: who among the 50- to 500-person companies that make up the bulk of our value creation already has that? I see Excel lists sent by email. I see proprietary controls on machines whose data can only be extracted with a crowbar. I see three different software systems for warehousing, orders, and quality assurance that don't communicate with each other. Planting AI on this digital wasteland is like putting a Formula 1 engine in an old tractor – it looks great on paper, but in the end, only the engine smokes. Or in this case: the invested budget.

The Fata Morgana of 800 Billion Euros

And then come these dizzying figures. Over 800 billion Euros. One imagines how this money flows into the research departments of SMEs, into the further training of skilled workers, into the development of modern data infrastructures across the country. The reality is likely to be different. A large part of this sum will flow into large-scale projects of corporations that already have the means and the strategy. Siemens, SAP, and Deutsche Telekom (all of whom are conveniently on stage in Hanover) will secure the lion's share for the development of – hold on – industrial data spaces, AI infrastructure, and software platforms. From their perspective, this is completely legitimate. But the machine builder from East Westphalia must understand: he is not the recipient of the money here, but the future customer of these platforms. He is not co-financing the party; he will ultimately pay the bill. And there's no getting around that.

The uncomfortable truth: We are satisfied, but no longer fast

Let's look at the facts soberly. Yes, Germany is the European champion in automation. According to IFR figures, we have 449 robots per 10,000 employees. That's top in Europe (Western European average: 267) and far above the worldwide average of 132. We are good at automating processes that we have once defined with high efficiency. This is the legacy of our engineering prowess. But here's the catch: our annual growth in robot density since 2019 has been a leisurely 5%. Sounds okay? Asia, in the same period, is moving at a dizzying pace. They are not yet as saturated, but they are learning and installing faster. Much faster.

We are resting on our hardware excellence while value creation shifts to software, data, and algorithms. And that is precisely the playing field of Industrial AI. It's no longer just about making a process faster with a robot. It's about using data analysis to predict which process will be needed next, which product variant the market will want in six months, or which component will fail in three weeks. We are world champions at reacting. AI makes us proactive. But only if we create the conditions for it. And these are not in the machinery, but in the data highways of our companies. And in many places, these are still gravel roads.

Klaus, the problem is that boardrooms are still thinking in terms of machines, steel, and iron. But the next revolution will be paid for in silicon and code. Many of my SME customers are afraid to put their sacred production data on a cloud platform – whether from SAP, Siemens, or Microsoft. If we don't overcome this cultural hurdle, they can talk as much as they want in Hanover. Then AI will remain a topic for corporations.

— Dr. Lena Weisgerber, Head of Digitalization Strategy, Fraunhofer Institute for Production Technology and Automation IPA (with whom I spoke last week)

But... isn't the summit in Hanover still an opportunity?

Now I have to be fair. Of course, it's not all just show and hot air. The fact THAT these people come together is remarkable in itself. When a Roland Busch (Siemens) and a Christian Klein (SAP) jointly demand a framework from politicians, it has a different impact than if the VDMA did it alone. The opportunity lies in finally setting the course for what would really help SMEs: genuine, open standards. Interoperability. Legal certainty in handling data (keyword: EU Data Act).

If the meeting results in a common, binding roadmap for the development of federated data ecosystems like Catena-X or Manufacturing-X – i.e., networks where SMEs retain sovereignty over their data but can securely share it with partners – then much would be gained. If a Digital Minister Wildberger not only talks about 5G masts but also provides guarantees for affordable and secure edge cloud infrastructure for the provinces, then it gets exciting. And if an Economy Minister Reiche sets up concrete funding programs that target the scaling of proven AI applications in SMEs (e.g., through voucher models for external consulting) rather than pilot projects, then all the effort will have been worthwhile. Whether that will actually happen, I dare to doubt. Hope dies last.

What I see in practice: Between ingenious tinkering and the digital Stone Age

The contrast couldn't be greater. Two months ago, I was at an automotive supplier near Ingolstadt. A hidden champion, 250 employees. They built a quality control system for weld seams using computer vision that is absolutely insane. Self-made. An engineer spent months digging into Python libraries and an inexpensive camera and created a system that reduced scrap rates by 40%. That's Industrial AI in action. They didn't wait for Siemens. They just did it.

Three weeks later: A visit to a traditional machine builder in the Swabian Alb. World market leader in a tiny niche. Proud as punch of the precision of their machines. When I asked about data collection, the production manager shrugged: "The foreman writes down the quantities and downtimes on a piece of paper at the end of the shift. Then the temp types it into the system." This is not an isolated case; this is the rule! This company has a data treasure that would be worth gold – vibration data, temperature curves, cycle times. But it lies dormant. They think AI is science fiction, yet they are sitting on an oil well and don't even have a drill.

The Sales Problem: Selling AI Solutions to the Wrong People

And the providers? They diligently contribute to the confusion. They develop generic AI solutions and then try to distribute them across the market with a watering can. Salespeople who themselves barely understand what the algorithm does try to sell production managers a "revolutionary AI platform." The result is pure frustration on both sides. Honestly: most manufacturers don't need an all-encompassing platform. They need a solution for ONE specific, painful problem. Maybe it's the setup time on machine 7. Maybe it's the unreliable demand forecast for raw material B. So the mistake happens right at the beginning: in defining the ideal customer profile. Instead of looking for the pain point, a technical solution is presented. This is the sure way to a price war and project failure.

Do you really understand your customer? The ICP Playbook Before you develop or sell an AI solution: define crystal clear who it is for. Our ICP Playbook helps you sharpen your ideal customer profile and align your go-to-market strategy with real pain points – not buzzwords.

What needs to happen now: A 4-point plan for SMEs

Okay, enough analysis. What does this mean specifically for you as a CEO or sales manager of a manufacturing SME? Wait until the money rains down from Hanover? Certainly not. Here's what I would do in your place – immediately.

1. 1. Start a data inventory, not an AI initiative: Forget the word 'AI' for six months. Assign one of your most capable process engineers (not the IT manager!) to conduct a ruthless inventory of your data landscape. Where is data generated? How is it stored? Where are the breaks in the chain? Which machine spits out data that no one uses? The result will be painful, but it is the only honest basis for everything else.
2. 2. Solve ONE problem, not the world formula: Pick the one process that costs you the most money, causes the most trouble, or requires the most manual work. And focus only on THIS ONE case. Find a solution for it. Maybe it's not even AI yet, but just a simple automation or better sensor technology. Success with this small project creates acceptance and know-how for the next, bigger step.
3. 3. Train 'data translators': You need people who understand both the language of the machine and the language of IT. People who can explain to a data scientist what a 'chatter mark' on a turned part means and why the vibration sensor goes crazy at 3,000 RPM. Send your best foremen and technicians to data analysis courses, not just the engineers from the office. Invest in people, not just software.
4. 4. Be a brutally demanding customer: When a provider comes with an AI solution, ask three questions: a) Which of my specific problems does it solve? b) What does the business case look like – what exactly will I save and when? c) Show me a reference customer who is exactly my size and industry and where it demonstrably works. Don't let them fob you off with demos on clean test data. Demand a proof of value, not a proof of concept.

Ready for Data? The Amplifa Industrial Data Readiness Check Is your data just digital junk or already the raw material for the future? Our practical check helps you assess the maturity of your data infrastructure and identify the most important construction sites on the way to value-adding AI use.

The Most Pressing Questions on Industrial AI

Does Industrial AI even benefit me if I only have 100 employees?

Yes, absolutely. Perhaps even more than a corporation. The advantage of smaller companies is their agility. You don't have to convince ten hierarchical levels. If you identify a clear problem – for example, the high scrap rate for a specific product – you can implement a targeted solution much faster. It's not about overhauling the entire company, but about making your existing processes more profitable with smart, data-driven tools. Savings of 5-10% in material or energy through better control are often crucial for survival, especially for SMEs.

What is the biggest mistake when introducing AI in manufacturing?

The biggest mistake is starting with the technology and not with the problem. It's the classic "We have a solution, where's the problem to go with it?" dilemma. The IT department is excited about a new platform, buys it, and only then considers what to do with it. That's backward. The impulse must come from production, from sales, from maintenance. Start with the pain, not with the pill.

More than just cold calling: AI Sales & Lead Generation Use AI not only in production but also in sales. With Amplifa's tools, automatically identify companies that currently need your solutions. Convert market data into qualified leads and appointments.

Hannover Messe 2026: Go or stay at the factory?

That's the crucial question. My advice: Send someone, but not the CEO, to applaud the DAX CEOs. Send your most curious engineer and your best salesperson. With a clear mission: Ignore the big stages. Go to the small booths in the back halls. Look for specialists who offer concrete solutions for your specific problem. Talk to startups. Collect ideas, not glossy brochures. And in the meantime, the boss stays home and starts the data inventory mentioned above. That's the best division of labor.

The showdown in Hanover will therefore take place, with or without us. It will generate headlines and convey the good feeling that Germany is at the forefront of AI. But the real work – the dirty, thankless, but crucial work – takes place in the factory halls between Itzehoe and Garmisch. It is the work on data, on processes, and on culture. And I bet that the companies we will celebrate as winners of this transformation in five years will not be those who applauded loudest in Hanover, but those who quietly and diligently did their homework in the meantime.

So the question is not whether Industrial AI is coming. It's already here. The question is whether you wait for a corporation to sell you an expensive standard solution, or whether you start now to create the conditions for AI to work FOR YOU. So, what are you waiting for?