BRUSSELS, Aug. 2 — The European Union’s Artificial Intelligence Act now has a price tag, and the European Commission has put the number in writing: compliance will cost affected businesses between 10,000 and 400,000 euros a year, depending on company size and the complexity of the AI systems involved. That range is the story before the story. A 40-fold spread does not describe a uniform regulatory burden; it describes a regime that hits different parts of the market in radically different ways.
The mechanics here are easy to trace. High-risk AI systems face the strictest obligations, which translate into extensive documentation, formal risk assessments, and human oversight.
None of that is a one-time setup cost. It is an annual operating expense — headcount, legal review, audit trails, compliance infrastructure — built into the balance sheet of every deployment.
The Compliance Bill
So the first question for a European business is simple: which side of that range am I on? The Commission’s own estimate makes the answer largely a function of system classification. Low-risk, low-complexity tools sit near the bottom of the range.
High-risk systems sit near the top. On closer reading, the real exposure is not the raw figure but the fact that compliance cost scales with ambition — the more your AI does, the more the Act requires you to prove about it.
That is a structural brake on the most advanced use cases, precisely where the largest costs concentrate. There is, of course, the official argument for all of this. Brussels frames the Act as the necessary condition for trust, and trust as the precondition for growth.
The EU’s AI market was valued at approximately 10 billion euros in 2023 and is expected to grow to over 30 billion euros by 2027. The claim is that a clear regulatory floor is what turns that projection from hype into something durable.
The Trust Dividend
There is some evidence behind that claim. A 2023 study by the Centre for the Governance of AI at the University of Oxford suggested that clear regulatory frameworks could increase AI adoption by up to 20 percent in sectors such as healthcare, finance, and manufacturing. The catch is that the study’s premise is clear frameworks.
Industry groups are already questioning whether the Act’s stringent requirements qualify as clarity or merely as burden. Their worry is the familiar one in EU tech policy: rules written to discipline large operators tend to land hardest on small ones.
A firm at the low end of the Commission’s cost range can treat compliance as an overhead line. A startup deploying a high-risk system faces the top of that range — documentation burdens, human oversight duties, ongoing risk assessments — without the revenue base to absorb them. The Act may not have been designed as a barrier to entry, but the cost structure it creates functions like one.
Follow the money in the short term and the winners are easy to identify: consultants, auditors, compliance officers, and every firm that sells the machinery of regulatory proof. The genuine question is whether that spending stays defensive — a tax on doing business in Europe — or converts into the trust dividend the Commission is banking on.
The market projection offers a timeline for judgment. If the EU’s AI sector is to reach the expected over 30 billion euros by 2027, the adoption gains the Oxford study envisions in healthcare, finance, and manufacturing will have to show up by then — and they will have to show up among smaller firms as well as large ones.
That is the metric worth watching over the coming quarters: not whether the Act survives contact with industry, but whether the growth it promises arrives fast enough to justify the cost it imposes now.
