Siemens and Germany’s Chancellor Are Demanding a Carve-Out for Industrial AI. What That Means for Every SME Using AI in Europe.

The Billion-Euro Ultimatum from Hannover Messe

At one of the world’s largest industrial trade fairs, the Hannover Messe, the CEO of Germany’s most valuable company and the country’s chancellor made their position on the EU AI Act unambiguous. Siemens CEO Roland Busch warned that most of the company’s €1 billion industrial AI investment would go to the US and China unless the EU changed course. Chancellor Friedrich Merz said Germany would push to “exempt industrial AI from the current regulatory straitjacket.”^[1]

This is not a fringe complaint from a lone company. It is the most high-profile, publicly stated business challenge to the AI Act since it came into force. Busch’s statement to Bloomberg was direct: “It’s complete nonsense to treat industrial and machine data the same way as personal data. I can’t explain to my shareholders why I’m investing money in an environment where I’m being held back.”^[2] Merz called the current EU framework “too tight” for industrial AI and said Germany would work to “extricate industrial AI from the current, overly restrictive straitjacket of the EU’s regulatory framework.”^[2]

The timing matters. The EU AI Act’s high-risk obligations come into force on August 2, 2026 — approximately 105 days from now. Siemens is not alone in raising concerns. But it is now the loudest and most specific voice making the argument that the regulatory burden is large enough to redirect investment decisions at the billion-euro scale.

What Siemens Is Actually Objecting To

Busch’s core argument is specific and worth understanding precisely. He is not saying industrial AI should be unregulated. He is saying industrial AI is already regulated by sector-specific rules — in manufacturing, critical infrastructure, aerospace, automotive, and similar domains — and that the EU AI Act is adding a second layer of high-risk obligations on top of rules that already apply.^[1]

The EU AI Act’s Annex III lists high-risk AI systems. Many industrial AI applications — quality control in manufacturing, predictive maintenance, supply chain optimisation — fall within those categories when they are used in the specified sectors. The argument from Siemens and its supporters is that sector regulators in those domains already have the technical expertise and enforcement relationships to oversee AI in those contexts, and that layering the AI Act on top creates duplication without meaningful additional protection.

The Data Act angle is part of this. Busch’s complaint to Bloomberg covered both the AI Act and the Data Act, arguing that industrial machine data is being treated as if it were personal consumer data, creating compliance costs for data uses that are already governed by existing frameworks.^[3]

The Political Dimension: What Merz’s Position Changes

This is not just a CEO speaking at a trade fair. Merz is the Chancellor of Germany — the EU’s largest economy and the country with the most to lose from a regulatory environment that pushes industrial AI investment toward the US and China.

Merz’s statement at Hannover Messe explicitly commits Germany to pushing for legislative change: “I will push to ease the regulatory burden in the EU on AI and, where possible, to exempt industrial AI from the current regulatory straitjacket that is too tight for AI within the European Union.”^[2] This is a political position taken by a head of government, not a corporate PR statement.

The EU AI Act’s Omnibus legislative negotiations are currently in trilogues between the Commission, Parliament, and Council. Germany’s position in those negotiations just got significantly harder for the AI Act’s supporters. Chancellor Merz’s public call for an industrial AI carve-out gives Germany’s negotiators a clear mandate to push for that exemption in the final Omnibus text.

What This Means for SMEs

There are two ways to read this story, and SMEs should be clear about which one applies to them.

If you are an industrial AI user or developer — manufacturing execution systems, predictive maintenance, factory-floor AI monitoring, supply chain AI — this story is directly relevant. Siemens’s argument is that the compliance burden for industrial AI is excessive relative to the actual risk and existing sector regulation. If that argument gains traction in the Omnibus negotiations, the specific obligations landing on your AI systems on August 2 could be materially reduced or delayed.

If you are using AI in non-industrial contexts — hiring, HR, customer service, credit scoring, content moderation, education — the Siemens story is less directly applicable to your compliance obligations, but it matters for the broader regulatory environment. The outcome of the debate over whether the AI Act is too broad will shape how the regulation is enforced across all sectors, not just industrial AI.

The Omnibus legislative process is the vehicle most likely to produce changes before August 2. The Parliament’s IMCO and LIBE committees have already voted to back fixed 2027 and 2028 deadlines, replacing the Commission’s flexible trigger approach. Nothing has passed into law yet, but the political momentum for some form of delay or carve-out is growing. Siemens’s intervention at Hannover Messe adds high-profile business pressure to that political momentum.

The Counter-Argument Worth Considering

It is worth noting what the AI Act’s supporters say in response to this kind of challenge. The regulation’s defenders argue that sector-specific rules do not cover AI-specific risks: a manufacturing quality control system that makes consequential decisions about product acceptance needs AI-specific oversight that sectoral rules for industrial equipment do not provide. The AI Act’s requirements for transparency, logging, human oversight, and bias testing are designed precisely because existing sector frameworks were not built for algorithmic decision-making at scale.

The debate is not settled. But it is real, and it is happening at the highest levels of European industrial policy right now — not in academic circles or legal journals, but at the world’s largest industrial trade fair, with a chancellor and a Fortune 500 CEO making specific threats about capital allocation.

What You Should Do Right Now

Three things to watch and act on:

• Monitor the Omnibus trilogues. The Digital Omnibus negotiations are the live legislative process that could change the August 2 deadline and add sector carve-outs. Political agreement is expected around April 28, 2026. If you are running industrial AI, the outcome of those negotiations directly affects your compliance timeline and scope.
• Do not delay compliance planning because of the debate. The law on August 2 is still the law unless and until the Omnibus is enacted. Businesses that pause compliance work in anticipation of a delay that does not materialise will face a compressed timeline with no wiggle room. Continue preparing your documentation and logging systems, but watch the legislative calendar closely.
• If you are using AI in an industrial context, assess your sector-specific obligations separately from your AI Act obligations. The AI Act is layered on top of existing sector rules, not a replacement for them. Understanding which sector regulator has oversight of your AI system and what requirements already apply will give you a clearer picture of whether the “double regulation” argument applies to your specific situation or whether the AI Act adds genuinely new obligations worth taking seriously.

The Bottom Line

Siemens’s €1 billion question to Brussels is now also Germany’s question. Chancellor Merz’s explicit commitment to push for an industrial AI carve-out in the EU’s AI Act framework is the most significant political intervention in the regulation since it was passed. The Omnibus negotiations are the venue where that question gets answered.

For SMEs: the debate matters for your compliance timeline, but it does not change the law yet. Watch the legislative calendar, continue your compliance work, and pay particular attention to whether the final Omnibus text includes specific carve-outs for industrial AI applications. If it does, your compliance obligations under the AI Act may be narrower than the August 2 deadline currently implies.

This article is for informational purposes only and does not constitute legal advice.