The Unforced Error: How European Management Is Rationing Its Way Into Irrelevance

Economics is not a mysterious subject. When one input is expensive, you substitute toward the cheaper one. Europe has famously expensive labor. Software — almost all of it imported from the United States — is priced at a small fraction of an hour of that labor. Why, then, does European management ration the software?

A parable in $30 increments

A friend at a large continental firm told me the following story. To obtain a Microsoft Copilot license — list price around thirty dollars a month — the requesting employee had to build a business case demonstrating five thousand dollars of expected annual value. The document had to be reviewed by IT, then by finance, then by a governance committee. The process took months. The tool, once approved, was subject to quarterly re-justification.

I have heard variants of this story from enough peers that I no longer treat it as anecdote. A colleague in Amsterdam told me his firm demands a documented ROI for any SaaS subscription over ten euros per user per month. Another, at a manufacturing company in the Rhine valley, described a sixteen-week approval cycle for a code-completion tool. A third told me his firm banned ChatGPT and mandated a homegrown alternative that, in his words, “works like a search engine from 2005.” The specific firm changes; the pattern does not.

Meanwhile, the same firms pay engineers, analysts, and consultants six-figure salaries loaded with payroll taxes and mandatory benefits that push the fully-loaded cost of an hour of skilled labor well past a hundred euros. And they are spending that hundred euros an hour to protect a thirty-euro monthly license.

Consider the arithmetic. The requesting employee spends, conservatively, four hours writing the business case — four hundred euros. Their manager spends an hour reviewing it — another hundred. The IT ticket triage consumes half an hour of the ticket owner’s time and another half hour of a specialist’s — call it another hundred. Finance reviews the ROI projection: half an hour of an analyst’s time, plus fifteen minutes from a finance director. The governance committee meets to consider it, four people at the table for twenty minutes each, plus preparation time. A vice-president of IT signs off. That single signature is the cheapest thirty seconds of the process, and its opportunity cost still exceeds the monthly license fee. Somewhere between eight hundred and twelve hundred euros of labor time have now been consumed to decide whether to spend three hundred and sixty euros a year on a tool. The process to protect against overspending has itself vastly overspent, and it has done so whether or not the license is ultimately granted. The bureaucracy is the expense. The license is the rounding error.

The wage/tool inversion

Here is what makes this strange. The entire history of technological adoption in Europe is a history of expensive labor pulling in cheaper machines. The mechanical loom, the assembly line, the industrial robot — each spread because paying a worker cost more than buying the tool that could do part of the worker’s job. Daron Acemoglu has written extensively on this: rich, high-wage economies mechanize faster than poor ones, because the incentive to substitute capital for labor is stronger where labor is expensive.

By that logic, Europe should be the world leader in AI adoption. Wages are high. Payroll taxes make them higher. Employment protection makes firing costly and hiring slow. The economic incentive to replace the marginal hour of expensive human effort with the marginal minute of cheap machine effort is enormous.

And yet the data run the other way. In early 2026, 43% of American workers reported using generative AI at work. In Germany and France, the figure was around 30%; in Italy, 26%. Bick, Blandin, and Deming show in a recent Brookings paper that employer encouragement statistically accounts for nearly all of this gap. It is not that European workers cannot use these tools, or do not know about them. It is that their employers actively discourage the use.

The pattern extends beyond AI. American firms spend 5.7% of revenue on IT; European firms, 4.2%. At the frontier, roughly 34% of American firms report using AI in production versus 20% in the EU. European ICT capital investment fell from 80% of American levels in 2000 to 57% in 2011 and has never recovered. Labor productivity per hour in the eurozone grew half as fast as in the US between 1995 and 2019.

This is the wage/tool inversion. High labor costs should pull in more tools. In Europe, they pull in fewer. The incentive says one thing; the management does the other. That is not economics failing. That is management failing to read its own economics.

Enter Draghi

The most authoritative articulation of this problem is not a blog post — it is a nearly-four-hundred-page report commissioned by the European Commission and delivered by Mario Draghi in September 2024. Its diagnosis is unflinching. Roughly 70% of the per-capita GDP gap between the EU and the US is attributable to lower productivity, and that productivity gap is, in Draghi’s words, “largely explained by the tech sector.” Seventy percent of foundational AI models developed since 2017 came out of the United States. Three American hyperscalers control 65% of the European cloud market. Europe, having failed to grow its own tech giants, is now a customer of American ones — and increasingly a reluctant one.

Draghi called for annual investment of eight hundred billion euros, roughly four to five percent of European GDP, to close the gap. As of the one-year retrospective in September 2025, roughly eleven percent of his recommendations had been implemented.

Eleven percent is not an accident. It is a revealed preference. European institutions read Draghi’s report, agreed with its diagnosis, and then declined to act on it. The document sits on a shelf, occasionally invoked, rarely followed.

Verticals of doom

Zoom in from the continental scale to the corporate one and the same pattern repeats. In the firms I have observed, IT is not a service function that reports to the business it serves. It is a vertical of its own, reporting up through its own chain to its own executive, with its own budget targets that are almost always denominated in cost reduction. The person deciding whether you can have a Copilot license is not measured on the productivity you would gain. They are measured on whether they came in under their software budget for the quarter.

This is the structural problem. When the enabling function and the enabled function report to different executives, and only the enabling function’s costs are visible on the balance sheet while the enabled function’s foregone productivity is invisible, the incentives are asymmetric. Every euro spent on a license shows up in a report. Every euro not earned because a license was denied does not.

A rational cost-cutting IT department, evaluated on its own metrics, will always ration. There is no counterparty inside the organization whose job is to say: this is worth spending. The business unit that would benefit lacks the budget authority. The CFO sees only the aggregated IT line. The CEO does not descend to the level of thirty-euro decisions. Everyone is doing their job, and the collective outcome is a firm systematically starved of the tools that would multiply its expensive labor.

This is not, I want to emphasize, a critique of any individual IT professional. Most of them are talented people responding to the metrics they are given. The failure is one level up: the executives who structured the incentives. And it is the executives who will bear the consequences when a leaner, better-tooled competitor eats their lunch.

Underneath the incentive structure sits a deeper cultural preference. European management, in my experience, is far more afraid of taking the wrong action than of falling behind through inaction. A wrong action is visible — a bad procurement decision, a data breach, a tool that flops in pilot — and it can be traced back to a name. Inaction is invisible. Nobody is blamed for the productivity gain that never happened, because it never happened. The risk-averse manager, weighing the two, chooses inaction almost every time. And what looks locally like prudence aggregates, across thousands of such decisions, into continental decline.

The American mirror

I do not want to romanticize the American approach. “Move fast and break things” produced its share of avoidable disasters — data breaches, botched deployments, security incidents whose costs eventually exceeded any productivity gain. The failure mode of American IT is over-provisioning: every employee gets every tool, and half of them never log in. This has its own price.

But between the two failure modes, one is recoverable and the other is not. A firm that over-provisioned can trim later. A firm that under-provisioned for a decade has watched its competitors internalize new workflows, hire the people who know how to use those tools, and build products at a cadence its own workforce can no longer match. The productivity gap compounds. By the time the executive committee decides to act, the market has moved on.

This is the sense in which the European under-investment is an unforced error. The regulatory environment did not cause it — the same GDPR-bound firm can adopt Copilot or refuse to; both are legal. The macroeconomic environment did not cause it — European wages have been high for decades and the wage/tool logic works exactly the same as it does elsewhere. The choice to ration is a management choice, made firm by firm, quarter by quarter. It can be reversed the same way, but only if someone chooses to reverse it.

The Europe that is not losing

A caveat, because “Europe” is not one place. The Nordics do not fit this pattern. Denmark, Finland, Sweden, and the Netherlands post AI adoption and IT investment rates comparable to or above American benchmarks. Their firms are among the most digitized in the world. Estonia, for reasons of its own history, runs a public sector more technologically fluent than most private American firms.

The story is really about the large continental economies — France, Germany, Italy, Spain — and about the specifically European style of large firm: hierarchical, cost-conscious, and organized around functional verticals that report to specialized executives rather than to the operating business. It is the management culture of the postwar continental corporation, in an economy that no longer rewards it.

Wolfgang Münchau made this point sharply in a February 2026 essay titled, unambiguously, “Europe Will Lose the AI Century.” His example was Mercedes-Benz relaunching a petrol S-Class on the same day Tesla announced its exit from consumer cars to focus on AI. Two industrial civilizations, two allocations of executive attention. One of them is going to be right.

The exit

The mechanism by which unforced errors get corrected is rarely a change of heart from the executives who made them. It is talent flight. The engineers, analysts, and product people who are paying closest attention to the tooling gap are also the ones most likely to notice that they could have a different career elsewhere.

If you are one of those people, reading this at your desk in a firm where the last software approval took three months and the answer was still probably no, the question is not abstract. Your peers at better-run firms — often American, sometimes Nordic, occasionally an EU startup that never internalized the corporate reflex — are shipping things you cannot ship, learning things you cannot learn, and building career capital that compounds while yours flattens. Five years of that gap, and the two of you are no longer the same candidate.

The firm may well survive. It will muddle through, or capitulate and outsource its technology work to an external provider that operates on a different mindset. But the individual worker who stayed will not have the same portfolio, the same network, or the same operational fluency as the one who left. The firm’s mediocre survival is not the worker’s survival. The two are different quantities, and confusing them is another version of the same accounting failure that produced the ration in the first place.

I do not mean this as career advice — I am not qualified to give any. But the situation described in this essay is not just a story about economies and executives. It is also a story about the day-to-day working lives of thousands of talented people, and those people have a decision to make. Whether they know it or not, they are part of the mechanism. Every talented person who leaves for a better-tooled competitor is one more increment of the pressure that will, eventually, force the change.

What comes next

I do not think this is hopeless. Management culture is downstream of executive incentives, and executive incentives can change. Boards can demand it. Shareholders can demand it. The competitive market can, eventually, force it — often through the talent flight described above. Firms that resist long enough will be acquired, restructured, or bankrupted, and their replacements will be built by people who never internalized the ration-first reflex.

The correction is not a matter of if but of when, and at what cost. Much of that cost will be borne by the workers of firms that waited too long — those who did not exit early enough to protect their own trajectory. The engineer who wanted to try Copilot in 2024 and was told no might spend the late 2020s watching her firm shrink, then losing her job to a merger, then being retrained on tools she asked to use half a decade earlier. Or she might leave in 2026, join a competitor, and skip the entire episode. The choice is not one she asked to have to make, but it is one she has.

The mistake was not the technology. The mistake was the ration. And the tragedy of unforced errors is that they were, at any point, avoidable.