I was scrolling through the news the other day and something hit me like a ton of bricks. We’re all hyped about AI, right? It’s the biggest tech shift since the internet, promising to supercharge everything from medicine to manufacturing. But while some parts of the world are hitting the accelerator, it feels like Europe is pumping the brakes. Hard.
And I’m not the only one who thinks so. The CEOs of two of Europe’s biggest industrial and software giants, Siemens and SAP, just went public with a bombshell statement. They’re essentially telling the EU:
your AI rules are broken, and they’re kneecapping our ability to innovate.
This isn’t just some random complaint. When Roland Busch (Siemens) and Christian Klein (SAP) speak, it’s a big deal. These aren’t social media companies. They build the software that runs our factories, manages our supply chains, and powers our economies. They’re on the front lines, and they’re saying Europe is getting it all wrong.
⚙️ The Big Problem: The EU AI Act
So what’s all the fuss about? It centers on a piece of legislation called the EU AI Act. On the surface, it sounds great. The goal is to make sure AI is safe, transparent, and doesn’t stomp all over our rights. To do this, the law sorts AI systems into different risk categories, from minimal to unacceptable.
Think of it like this:
- Unacceptable Risk: AI that’s just plain creepy or dangerous, like social scoring systems. These get banned.
- High Risk: AI used in critical areas like medical devices, self-driving cars, or hiring. These face super strict requirements for testing, documentation, and oversight.
- Limited Risk: Things like chatbots, which just have to tell you you’re talking to an AI.
- Minimal Risk: Most other AI applications, like spam filters or video games.
Sounds logical, right? The problem, according to Busch and Klein, is in the execution. They argue the law is so broad and the compliance requirements for “High Risk” systems are so crushing that it creates a massive wall of bureaucracy. Instead of fostering innovation, it’s creating hesitation. Companies are getting so bogged down in navigating overlapping and confusing rules that the actual work of building cool, useful AI gets pushed to the back burner.
It’s like wanting to build a race car, but the rulebook is 5,000 pages long and requires you to fill out a form in triplicate every time you want to change a tire. You spend more time on paperwork than on engineering.
☠️ The “Toxic” Law You’ve Probably Never Heard Of
But the AI Act isn’t even the worst part. Siemens CEO Roland Busch dropped a pretty intense word to describe another piece of EU regulation, the Data Act. He called it “toxic” for business. Whoa.
So what is this Data Act? In simple terms, it’s a law that dictates how data generated by devices and services can be used and shared. The idea is to give consumers and businesses more control over their data. Again, the intention is good.
But here’s the reality for a company like Siemens: they build smart trains, connected factories, and intelligent power grids. These things generate an unfathomable amount of data every second. This data is the lifeblood of modern industrial AI. You need it to:
- Predict failures: Analyze data from thousands of wind turbines to predict when a specific part will fail, so you can fix it before it breaks.
- Optimize processes: Use data from an entire factory floor to find tiny inefficiencies that, when fixed, save millions in energy and materials.
- Build better products: See how customers are actually using your machines in the real world to design the next generation.
The Data Act, in its current form, makes accessing and pooling this data incredibly complex. It creates legal hurdles and uncertainty around who owns the data and what they can do with it. For AI developers, this uncertainty is, well, toxic. It kills projects before they even start.
💡 The Real Bottleneck: A Data Treasure We Can’t Unlock
This brings us to the most important point these CEOs made, and it’s a total game-changer for how we should think about this.
For years, the story has been that Europe is falling behind in AI because it lacks the massive computing infrastructure and data centers of the US and China. The solution? Spend billions on building more data centers and buying more GPUs!
Klein and Busch are calling BS on this. They argue that a shortage of computing power is NOT the main problem.
The real problem is that Europe is sitting on a digital goldmine, a “treasure trove of data,” as Busch put it, but we’ve locked it up and thrown away the key.
Think about it. Europe has decades of high-quality data from world-leading industries:
Manufacturing: Precision engineering data from Germany.
Healthcare: Pharmaceutical and medical records from across the continent.
Logistics: Data from some of the world’s busiest ports and supply chains.
Finance: Decades of structured financial data.
This isn’t cat photos or TikTok videos. This is dense, valuable, industrial-grade data that could be used to train incredibly powerful and practical AI models. But because of restrictive regulations like the Data Act, it’s all siloed away, inaccessible.
It’s like owning the world’s most amazing library, but every book is locked in a separate vault, and you’re not allowed to read them. Investing in a fancier building (more data centers) is completely pointless until you get the keys to the vaults (access to the data).
✍️ A New Playbook for European AI
So, if the current plan is flawed, what’s the alternative? Based on their arguments, here’s a practical roadmap for how Europe could flip the script and become an AI powerhouse.
- 📌 Step 1: Fix the Data Rules First. Before spending another euro on hardware, reform the Data Act. Create a genuine “Single Market for Data” in the EU. This means creating clear, simple rules that allow companies to legally and ethically access, anonymize, and pool data for research and development. Protect privacy, absolutely, but don’t make it impossible to innovate.
- 📌 Step 2: Make Regulation Smarter, Not Just Stricter. The AI Act needs a rethink. Instead of broad, one-size-fits-all categories, regulation should be more focused on the specific use case. The rules for an AI that diagnoses cancer should be incredibly strict. The rules for an AI that optimizes a factory’s cooling system? They can be much more flexible. This approach encourages progress while managing real-world risks.
- 📌 Step 3: Lean Into Europe’s Strengths: Industrial AI. Europe doesn’t have to beat Silicon Valley at its own game of consumer AI. Instead, it should aim to dominate “Industrial AI.” Focus on creating the world’s best AI for manufacturing, energy, logistics, and B2B services. This is Europe’s home turf, and it’s where that “treasure trove” of data provides a unique, unbeatable advantage.
- 📌 Step 4: Unleash the Investment. Once the data is flowing and the regulations are smart, then it’s time to pour money into infrastructure. The investment will be ten times more effective because there will be a clear pipeline of projects and data ready to be used.
✨ Why This Isn’t Just a Corporate Problem
It’s easy to dismiss this as just a couple of big companies wanting fewer rules to make more money. But this goes so much deeper and affects every single one of us.
If Europe gets this right, it means:
- Better Healthcare: AI that can spot diseases earlier and more accurately.
- A Greener Planet: AI-optimized energy grids that reduce waste and integrate more renewables.
- Stronger Economies: More efficient industries that create high-quality, future-proof jobs.
- Technological Sovereignty: Europe becomes a creator of essential technology, not just a consumer of tech built elsewhere.
If Europe gets this wrong, it risks becoming a technological colony of the US and China, dependent on their platforms and their rules. The future will be built on AI, and the question is whether we want to be architects or just tenants.
What Busch and Klein are calling for isn’t a free-for-all. It’s a plea for a smarter, more pragmatic approach. The debate shouldn’t be whether to regulate AI, but how to do it in a way that protects citizens and unleashes the incredible potential sitting right under our noses. It’s time to unlock the treasure chest.
The EU’s AI Act operates on a risk-based framework, classifying AI applications into categories from minimal to unacceptable risk. High-risk systems face strict requirements regarding security, transparency, and data quality.
Roland Busch of Siemens described the EU’s Data Act, which governs the use of consumer and corporate data, as “toxic” for digital business models. Both he and SAP’s Christian Klein argue that complex and sometimes contradictory regulations make it difficult to access and use the data necessary for innovation.
Unlike other tech leaders who signed an open letter calling for a delay, Busch refrained, stating that a simple delay is not enough. He advocates for a fundamental change to the law to truly foster European competitiveness.
The AI Act is being implemented in stages. Obligations for general-purpose AI models will take effect in August 2025, with most rules for high-risk systems following by August 2026. Non-compliance can lead to significant financial penalties.