In the latest issue, I debunked the narrative of “AI-driven layoffs” in the U.S. market. Oracle is cutting 30,000 jobs not because of artificial intelligence, but due to $108 billion in debt. Amazon announced 30,000 fewer jobs, yet filed a financial statement with the SEC showing 51,000 more employees. The New York Federal Reserve confirms: only 1% of service companies have actually laid off employees due to AI. 12% have hired fewer people.

Jobs disappear and reappear under a different name. SAP restructured 10,000 positions, yet its workforce grew by 2,371—because in the meantime, it hired in research and development, cloud computing, and artificial intelligence. The people who held those 10,000 jobs are not the same ones filling the new ones. The restructuring is real. Blaming AI for the layoffs is not.

Today I’m bringing those numbers to Europe. And the response is more uncomfortable than I thought.

Three things to know before continuing:

SAP announced an AI-related restructuring in January 2024: initially 8,000 jobs and €2 billion, later expanded to around 10,000 jobs and €3.2 billion. Headlines in every publication: SAP cuts thousands of jobs for artificial intelligence.

The 20-F filing with the SEC: workforce from 107,602 (December 2023) to 109,973 (December 2024). An increase of 2,371 employees. Growth was concentrated — to quote from the filing — ‘primarily in the areas of research and development and cloud.’ The jobs cut were operational and administrative. I The jobs cut and those created do not have the same profile.

Vodafone: 11,000 cuts announced, average workforce rose from 85,887 to 87,205. Up 1.5%.

Where staff numbers have actually fallen — at Nokia and Ericsson — the driving force isn’t artificial intelligence. It’s the 5G cycle, falling demand and shrinking margins. AI is just the label they’re slapping on it.

The best-documented case is Klarna. The F-1 filing with the SEC in March 2025: 5,527 employees at the end of 2022, 4,352 at the end of 2023, 3,422 at the end of 2024. By the end of 2025, approximately 2,907. Almost half the workforce gone in three years — without a single redundancy.

CEO Sebastian Siemiatkowski said it openly: “We simply told our employees that we would stop hiring. Natural turnover in a company like ours is 15–20% a year.”

The figure is confirmed: the average turnover rate in the European tech sector in 2025 is 17.4%. If you stop hiring and let people leave of their own accord, you lose a sixth of your workforce every year.

The accounting advantage is decisive: a hiring freeze does not trigger IAS 37 provisions, does not require consultation with trade unions, and does not generate disclosure obligations. No visible costs. No newspaper headlines.

Two structural differences give rise to opposing dynamics that lead to the same outcome.

In America, the at-will contract allows for rapid dismissal and rehiring. The cycle is clear: acquisition, restructuring, financial results.

In Europe, the OECD Employment Protection Index gives Italy a score of 2.9 out of 6, France 2.7 and Germany 2.3. The United States: 1.3. Dismissal costs more, takes longer and requires formal procedures. The OECD describes the result in its 2020 report: “Strict regulation tends to reduce dismissals — but it also tends to reduce hiring.”

The consequence is counterintuitive: aggregate employment is unaffected. But labour mobility is not. Those who have a job keep it. Those who don’t struggle to get one. Cahuc and Palladino (2024) put it bluntly: employment protection creates “a segmentation that disproportionately affects those entering the labour market for the first time.”

If you are an entrepreneur, this means that the Klarna model — don’t hire, wait — is the rational strategy under European labour law. If you are a parent with a child graduating in engineering, this means that the labour market awaiting them is structurally different from that of five years ago.

According to Eurostat data from 2025, 55% of large European companies use AI, compared with 17% of small ones. OECD data: 40% of those with over 250 employees, 11.9% of those with between 10 and 49. It is the widest gap among all digital technologies.

Large enterprises are quietly restructuring through workforce turnover, reaping efficiency gains and boosting margins. SMEs do not capture those benefits — but they inherit the effects on the labour market.

In the United States, the cost of restructuring appears on the balance sheet. In Europe, it is externalised: it falls on those who are not hired and does not appear anywhere.

The aggregate data paints a reassuring picture. The ECB survey from March 2026 confirms that companies using AI are 4% more likely to hire. The EIB/BIS report, based on 12,000 businesses, confirms a 4% increase in productivity without any negative impact on employment.

But the aggregate data hides what is growing and what is disappearing. AI is not causing the mass redundancies that the newspapers report — we have seen that. But it is deciding which jobs are created and which are not. And here the figures change completely.

Ravio’s data on the European technology market in 2025:

AI-related roles are growing at all levels — including entry-level positions. And the roles that are disappearing are not just junior ones: administrative and coordination roles are being cut at every level of seniority. But the net balance is lopsided. The overall decline in hiring in the European tech sector is 7%. At the P1/P2 level — entry-level positions — it is 73%.

The roles most affected at entry level: marketing (hiring rate down 75.6%), human resources (down 72.3%), and junior engineering (down 72.2%). 50% of remuneration managers surveyed by Ravio cited AI as the reason: “They can be automated.”

Meanwhile, those with AI skills receive a 23% pay premium compared to equivalent profiles without those skills (Oxford, 10 million job adverts in the UK). According to PwC, salaries in sectors exposed to AI are growing at twice the rate. The jobs that AI creates are worth more. But in terms of volume, this does not compensate for the jobs it eliminates at the grassroots level.

In the UK, every graduate role receives 140 applications — the highest figure since 1991. Jobs in the British tech sector have fallen by 46% in a year. In the Netherlands, entry-level developer roles have dropped by almost 40%.

The jobs that AI creates are worth more. The jobs that AI eliminates were the stepping stone to moving up.

Everything I have described so far concerns large corporations. SAP, Vodafone, Klarna, Nokia — companies with thousands of employees and direct access to capital markets. Contraction by omission is their mechanism. The -73% figure is theirs.

But SMEs — 99% of Europe’s business fabric, two-thirds of employment — are paying the consequences from a distance.

The pool of candidates with three to five years’ experience that an SME normally draws from is formed within large companies, at the entry-level. If those positions are no longer being created, in three to five years that pool will be empty. And even today, when an SME tries to recruit, it faces a problem that large companies can ignore: AI has levelled everyone’s output. CVs, portfolios and work samples are converging towards a uniform, indistinguishable quality — what I called the B+ trap in my recent research.

You cannot tell who has actually learnt the ropes from who simply has a better chatbot. Large companies filter candidates through multi-stage processes and dedicated recruitment budgets. SMEs do not.

Europe did not choose to destroy jobs. It chose to protect those who already have them. Job protection — designed to prevent people from losing their jobs — has had a side effect: a market in which it makes sense not to hire, where the cost of restructuring falls on those who do not yet have a contract, and where no commonly cited data captures the full extent of this. The result is a figure: a 73% drop in first-time hires over the course of a year, across Europe.

The decline is not in redundancies. It is in the jobs that are never created. In recent graduates who send out 140 applications and do not realise that the problem is not their lack of suitability — it is that the position simply no longer exists.

It does not concern only them. It concerns the European economy’s ability to produce the next generation of managers, entrepreneurs and innovators.

This article is the second in a series on how artificial intelligence is reshaping the operational architecture of businesses. The first instalment — Anatomy of an AI-driven redundancy — analyses the US market. For a more in-depth operational analysis — architecture, costs, implementation roadmap — see the white paper: AI for European SMEs: The 2026 Playbook.

Fabio Lauria

CEO & Founder, ELECTE

Every week, we explore AI without the hype — with data, analysis and an independent perspective.

Sources:

· SAP SE — 20-F 2024, SEC filing (Feb. 2025)

· Vodafone Group — FY2025 Annual Report; Light Reading

· Ericsson, Nokia, ASML — 2024 Annual Reports

· Klarna — F-1 filing with the SEC, 14 March 2025; statements by Siemiatkowski (Bloomberg, May 2025)

· Ravio — Tech Job Market Report 2025; Compensation Trends 2026

· Institute of Student Employers (ISE) — 2024–25 Graduate Survey · Stephany et al. (2026), University of Oxford, SkillScale project

· PwC — 2025 Global AI Jobs Barometer

· OECD — Employment Protection Index v.4; Employment Outlook 2020; AI adoption by SMEs (2025)

· Cahuc and Palladino (2024) — IZA Discussion Paper No. 16747 ·

· ECB — SAFE Survey, 4 March 2026

· EIB — Working Paper 2026/02 (Aldasoro et al.)

· Eurostat — AI adoption in EU enterprises, Dec. 2025

· Carnegie Endowment — ‘How Europe Can Survive the AI Labour Transition’, Feb. 2026

· Sifted — ‘European tech companies slash entry-level jobs as AI demand surges’, June 2025