AI is repricing work faster than most companies are ready for. People who work well with AI are taking work from people who don't. Teams are getting smaller, hiring is slowing, and the price of a lot of knowledge work is starting to fall.
Europe can't stop this and shouldn't try. Markets move to whoever delivers more for less. What we can decide is whether that works for us or against us.
The container lesson
This has happened before. In the 1960s, the shipping container cut the number of people needed to load a ship. British ports covered by the National Dock Labour Scheme protected the old way of working. Felixstowe, too small to be included in the scheme, opened Britain's first container terminal in 1967 and became the country's largest container port. London's docks could not handle the new ships and had all closed by 1980. Ships went to Rotterdam and Antwerp instead. Felixstowe; London Docklands.
The protected jobs disappeared anyway, and the trade moved to other ports. Containerisation then multiplied trade: one estimate found containerisation increased trade between developed countries by around 380% within 10 to 15 years. El-Sahli, University of Nottingham. Fewer people were needed per ship, trade grew far larger, and the ports that adapted first took the gains. I expect the same from AI.
It has already started
The clearest public example is Deel. Its CEO, Alex Bouaziz, says the company added more than $140 million in annual recurring revenue in 90 days without increasing headcount, by using Akai, its in-house AI agent platform, to automate roughly 600 full-time employees' worth of work in finance, HR, accounts payable and compliance. Revenue per employee went from $130,000 to $215,000. Akai started as an internal tool and is now a product Deel sells. Alex Bouaziz on X.
Deel published its numbers. Most companies don't. I know many founders building the same kind of systems internally to stay competitive. Everyone has access to the same models, so if they don't do it, a competitor will, and in a global market you fall behind fast. Their results show up as flat headcount and rising revenue.
Companies that can't build these systems themselves are paying to have them built. Consultancies are already placing forward-deployed engineers inside companies at around €2,000 a day to build AI systems.
The research points the same way. In an experiment with 776 Procter & Gamble professionals, individuals using AI matched two-person teams without it. The Cybernetic Teammate. Stanford found that among workers aged 22 to 25, employment in the occupations most exposed to AI was roughly 19% lower than it would have been at the growth rate of less-exposed occupations, mostly because of fewer hires. Stanford update. Neither study proves AI caused the change on its own, but both point the same way.
Why prices are about to fall
Start with marketing. Many agencies charge a percentage of their client's ad spend. Meta's Advantage+ and Google's Performance Max now automate much of the targeting, budgeting and placement that agencies used to do. It is hard to justify a fee that grows with spend when the platform does more of the work. Meta; Google.
I already see agency models under pressure. Creative was supposed to be the safe part, and it is getting cheaper too. Google already offers image and video generation inside its ad tools, and in my view routine static ads for ecommerce are close to solved. Google's creative tools. Much of the work European agencies used to sell is moving to American platforms.
Kevin Warsh put it plainly in November 2025: "AI will be a significant disinflationary force, increasing productivity and bolstering American competitiveness." Warsh's essay. He was writing about American competitiveness. European companies that don't adapt will feel the same shift as pressure.
One company lowers its costs. A competitor passes the savings to customers. Customers start questioning prices that used to seem normal. You can become more productive and still get paid less for what you sell. Klarna's revenue per employee approached $1.4 million in the first quarter of 2026, four times its 2022 level, yet its second quarter brought $1.042 billion in revenue and just $27 million in operating income. Klarna Q1; Klarna Q2. In a competitive market, much of a productivity gain goes to customers rather than into profit.
Lower prices can still grow the economy. When something gets cheaper, some buyers pay less for what they already bought, and others start buying for the first time. A small brand can now advertise well without a big marketing budget. Services that were out of reach for small businesses become affordable. New companies can serve customers nobody could serve profitably before.
Websites are the clearest example. A website used to mean hiring a developer or an agency. Today any kid can generate a personal website with AI and put it live on Hostinger in an afternoon. Hostinger, a Lithuanian company, grew revenue 51% in 2025 to €275.4 million, its fourth year in a row above 50%, and its customer base grew from about 1.5 million in 2022 to 4.6 million. Hostinger 2025 results. Cheaper websites meant far more of them, and a European company captured the growth. The container did the same for trade. Across the rest of the economy, the outcome depends on who builds the new supply, and I want it built in Europe.
Why companies are slow
Most companies are slow to adapt because of how they are organised. A manager whose standing depends on a large budget has a reason to protect it. A team paid by billable hours has a conflict the moment it finds a faster way to work. An employee who expects every improvement to become more workload has little reason to share it. These are incentive problems, and leadership has to fix them.
Electricity went the same way. Electric motors did not raise factory productivity until the early 1920s, four decades after the first power stations opened. Owners first swapped the steam engine for a dynamo and kept the old layout. The gains came only when factories were rebuilt around the new technology. Paul David, The Dynamo and the Computer. McKinsey's August survey shows the same thing today: 80% of respondents reported better personal productivity with AI, but only 37% saw better company operating profit. The high performers were the ones who had redesigned their workflows. McKinsey, August 2026. The gains come from redesigning the work.
I expect founders to lead this. A founder with a large stake captures the upside and has their wealth tied to the company's survival, so painful restructuring can still be the obvious call. A hired executive with little equity carries the disruption and career risk now, for a small share of the eventual profit. Keeping things as they are is the rational choice for them. Boards that want reinvention have to reward it.
What this means for employees
I don't expect most employees to end up worse off. Some people will lose their roles, and for them the transition is hard. But the labour market turns over all the time. People change employers, industries and professions every year. AI may speed that up, and it also makes the move easier for anyone willing to learn.
We see this in Germany. As of July 2026, 88% of our German graduates had been hired, counting everyone whose programme ended at least six months earlier and whose employment status we could confirm. Turing College. Employers are hiring people who can build with AI.
The people who lose out are the ones left without a route into new work. Employers and governments can fix that, and if they do, faster turnover becomes an opportunity for most people.
Europe's weak spot
Mario Draghi's 2024 competitiveness report found that no EU company worth more than €100 billion had been built from scratch in the last fifty years, and close to 30% of the unicorns founded in Europe between 2008 and 2021 moved their headquarters abroad. The Draghi report. We lean on old incumbents, and our best young companies often leave to scale. If incumbents adapt slowly, founder-led companies have to carry the change, and they need to be able to grow here.
Two paths for Europe
The first path is to use the market. People whose routine work disappears get a route into more valuable work. Companies redesign delivery and pass part of the savings to customers, which grows demand. Founders build services that were uneconomic before. We get more companies, more transactions and a larger tax base.
The second is to fight it. The displacement happens anyway, because customers buy from whoever delivers the better result at the better price, including suppliers outside Europe. Displaced workers without a route forward create political pressure. The response is to protect incumbents, subsidise roles that no longer pay for themselves, restrict foreign competition and expand the state's role. Together, these steps lead to less trade, less investment and our best people leaving for markets where the work is being built. The London docks tried this. Rotterdam took the trade.
Moving first matters because the advantage compounds. The companies that adapt first set the prices everyone else must match. They see the new problems first and build businesses around them. Skilled people also move to where the most interesting work is.
It is already happening here
I see the first path already in the Turing College community. Edvard Sivickij studied AI Engineering with us and co-founded condense.chat, which cuts the running costs of coding agents. condense.chat. Martynas Krupskis built Fieldy, a wearable AI note-taker that had shipped 9,000 units by its 2025 Product Hunt launch. Fieldy. Many more alumni freelance and earn independently on the same skills.
Chaseit is the one I think about most. It is a Lithuanian startup building AI voice agents for loan servicing and collections, running more than 20,000 automated calls a day in March. FF News. Its customers include Eleving Group, a fintech listed in Frankfurt and Riga. Eleving Group. Those are calls people used to handle, so this is displacement. That work would be automated either way. A European lender buying the capability from a Vilnius startup keeps the value, the engineering jobs and the know-how in Europe.
The junior problem
If routine junior work disappears, where do people get the experience to handle more consequential work? Companies cannot close the entry routes and still expect a supply of experienced judgment. The answer is practical learning on real problems, designed with care. One study found that people using AI learned less on average, unless they asked for explanations and worked to understand the material. Skill-formation study. Education has to build both speed and understanding.
Control the controllables
We do not control the direction of the technology, what customers will pay, or what competitors elsewhere decide to build. We do control how fast we learn, how we redesign our companies and where we put public money.
Employers should invest in deep technical skills and give managers the authority and incentives to rebuild how work gets done. Boards should reward reinvention. Governments should widen access to practical learning, especially for displaced workers and smaller businesses, and judge it by what learners can build afterwards. Paying for routes into better work now costs less than protecting jobs that no longer pay for themselves later. And each of us should treat working with AI as a skill that needs constant practice. Last year's tools are a poor guide to what a competitor can deliver today.
Capitalism will not slow down for Europe, and fighting it has never worked. Customers will judge the result, the price and the alternatives. Europe can build the better results or buy them from others. I want us to build them.