On Tuesday, Statistics Finland published its September labor force survey. There were 39,000 more unemployed people than a year earlier and 11,000 fewer employed. The trend unemployment rate, in the agency's own word, looks stuck around 10.5%. Finland has had the highest unemployment rate in the EU since November 2025.
But at the same time, the economy is finally growing! In the second quarter, the flash estimate put GDP 2.5% above a year earlier, while employment fell 1.1% and hours worked fell 1.4%. Finland is producing more with fewer people working fewer hours.
I always say that Finland is a country where trends come fast, and where they quickly become passé. Supermarket sushi, millennial gray, remote-first work... Now the country is answering in real time a question that most of the AI-and-jobs debate still treats as hypothetical. Can an economy grow without employment growing? Yes. Finland is doing it right now. The harder question is what that does to a state whose finances assume the opposite.
The boring explanation
Let's start with the explanation that would make this whole post unnecessary, it's a rational and solid one.
Coming out of a recession (has there been anything but recessions since 2008, though?), firms do not hire first. They squeeze more out of the people they kept, run idle capacity back up, and wait until they trust the recovery before adding headcount. Output leads, employment lags. The US went through exactly this in the early 2000s, when productivity stayed high and GDP rebounded well before employment did. Economists have a name for the lag and decades of evidence behind it.
Finland's recovery also has an obvious lumpy component. Second-quarter exports were lifted by the delivery of a multi-billion cruise ship by Meyer, and a single ship leaving Turku says little about the labor market of 2028. You would think that given two or three quarters, hiring would catch up. However, that does not explain the shape of what is happening underneath.
What the lag can't explain
The productivity jump is not a rebound to trend, because there was no trend to rebound to. According to the Bank of Finland, labor productivity grew 14.6% in total between 2000 and 2008, and only 1.1% in total between 2009 and 2023. Fifteen years of flat output per worker, and then, on the flash numbers, output per hour worked rising by close to four percent in a year while the workforce shrinks. That doesn't look like catching up, it's something different.
The work that does exist is progressively leaving the employment contract. In July there were 26,000 fewer wage earners and 25,000 more entrepreneurs than a year earlier, which is why the headline employment total barely moved. In the second quarter, solo entrepreneurs rose by 5,000 while employer-entrepreneurs fell by 9,000. A recovery headed toward hiring would show the opposite: more firms with staff, not more people working alone. Liisa Hanén of Mikro- ja yksinyrittäjät put it well: there may not be less work, but more and more of it is done without an employment relationship. And it is a known fact that the productivity of self-employed entrepreneurs and small firms consistently lags that of large firms.
Hiring demand has not paused. It has collapsed. Open vacancies in 2025 were 34,100, down for three years running and half the 2022 level.
The money arriving to drive the next phase of growth is overwhelmingly capital, not people. Planned and under-way data-center projects in Finland now exceed €67 billion. Google's commitment alone is at least €13 billion over 2027 and 2028, roughly a fifth of everything invested in Finland in a typical year. Its existing site in Hamina, operating since 2009, employs about 400 people. Google says the new investment will support tens of thousands of jobs during construction; an ETLA economist called the company's numbers optimistic, pointing out that a running data center is automated to the point of needing very few people. Construction jobs end when construction ends, even if the servers stay.
None of this is uniquely Finnish. Goldman Sachs economists argued last year that most US output gains will now come from productivity, increasingly AI-driven, while aging and lower immigration hold down labor supply. Oxford Economics expects AI's labor-saving effect to raise corporate profits as a share of the economy, with workers taking a smaller slice. Finland is simply further along, with fewer buffers, and with a state that is unusually exposed to the result.
The payslip state
Here is the exposure. Labor taxes account for 50.9% of all Finnish tax revenue. Income tax and social contributions on wages are not 'one pillar' of the Nordic model: they are the foundation the other pillars stand on, along with the pensions, unemployment insurance and parental benefits that are earned through employment.
The model works beautifully when growth arrives as jobs. Every new employee is a new taxpayer, a new contributor to the earnings-related insurance funds, and one fewer claimant. It works much less well when growth arrives as a ship exported to a foreign cruise line, a hall of servers run by a few hundred technicians, and a growing number of people invoicing through a toiminimi.
We can already see what that looks like in macro terms through the national accounts. In 2024, Finland delivered a nominal GDP increase while tax revenue fell. The economy grew; the part of the economy the state taxes most did not.
The longer arc points the same way. Wages made up about 65% of Finnish national income in the late 1970s and 1980s, fell to about 56% by the financial crisis, and have stayed well under 60% since. Growth has been migrating away from the payslip for thirty years. AI and capital-heavy investment are accelerating a drift that was already there.
Yang saw the shape of it
In 2018, Andrew Yang published The War on Normal People, one of my favorite books. Most readers outside the United States barely remember it as the manifesto behind a quirky (and failed) presidential campaign. Yang's argument was that automation would decouple economic output from the employment of ordinary people, and that a society which distributes income and dignity almost entirely through jobs would therefore break.
His proposals were radical and no-nonsense economic policy: A Freedom Dividend of $1,000 a month for every adult aged 18 to 64, regardless of work status. A value-added tax at roughly half the European level to pay for it, so that the companies capturing automation's gains would pay at the point of sale. Digital Social Credits to reward the care, volunteering and community work that markets do not price. And an umbrella philosophy he called human-centered capitalism, but in reality they're closer to what you would call 'democratic socialism' in a capitalist, Nordic-style system. I guess the word 'socialism' performed horrendously with American focus groups.
Read today from Helsinki, the worrying part is that Finland already has most of Yang's toolkit, and the problem is arriving anyway.
Finland ran the world's first nationwide, statutory, randomized basic income experiment. Between 2017 and 2018, 2,000 unemployed people received €560 a month, tax-free, regardless of other income or whether they looked for work. The employment effect was small, slightly positive, and muddied by the activation model introduced in 2018. Recipients reported better finances and less mental strain. The experiment did not prove that basic income creates jobs. It did show that unconditional money does not make people stop working, which was the main objection.
Finland also has a VAT far beyond anything Yang proposed. The general rate has been 25.5% since September 2024, the second highest in the EU. Yang wanted half the European level. Finland is near the top of it.
So why doesn't Yang's funding mechanism save the Finnish model? Because it was designed for a different economy. Yang's VAT works in the United States because the value created by automation is sold, mostly, to American consumers. Amazon's warehouses ship to American doorsteps. The tax catches automation's output on its way to the people who buy it.
Finland is a small export economy. About two-thirds of the Finnish VAT base is household consumption, which sits downstream of wages. When the payroll stops growing, consumption follows it, and the VAT base erodes fast. Meanwhile exports are zero-rated. The cruise ship leaves Turku without paying Finnish VAT. Much of the computing that will run in Kajaani and Muhos will be sold across borders, where Finnish VAT does not reach. The growth Finland is now generating is precisely the kind a consumption tax is built not to see.
Yang identified the right problem and, for a continental consumer economy, a plausible fix. For a small Nordic exporter, his VAT is a tax on the shrinking part.
Own the socket
Growth in Finland's near future happens at the grid connection. (I'm an energy engineer, so I guess when you're a hammer everything looks like a nail...). In the twelve months to June, Fingrid received connection inquiries for about 68 gigawatts of new consumption, in a system that peaks at around 16. Taxing that queue would erode the very advantage everyone is queuing for, but ignoring it sends the value to shareholders elsewhere.
Yang foresaw a third option. He called his payment a dividend on purpose, likening it to a company paying its shareholders, and his model was Alaska's Permanent Fund, which turns oil revenue into an annual dividend for residents. Alaska doesn't tax its oil harder, it owns a share of the proceeds. Numerous Native American nations have similar systems in place whereby they share profits from tribally-operated businesses with their members.
Finland already has an excellent shareholder. Its earnings-related pension funds held about €285 billion at the end of 2025, more than a year of GDP, and they are the part of the payslip state most exposed to a shrinking payroll. Give them a standing co-investment facility, alongside the state's holding company, to take minority stakes at market price in the data centers and electrified plants built on Finnish power generation. Developers get patient local capital, nobody's electricity bill moves, and the returns flow back into the system payroll used to fund. When work moves from the payslip to the server hall, the pension contract follows it as an owner.
Pension funds answer to pensioners and can't be ordered to invest, and Finnish retirements in Finnish server halls double the exposure to one industry's cycle. EU rules require non-discriminatory grid access, so co-ownership must be an offer, never a toll. But a state that wants a share of growth without pricing away the reason it came has to start thinking like a shareholder. That is Yang's dividend, Nordic edition.
Where growth comes from
Put the pieces together and in the near future, growth will come from three places: cheap clean electricity converted into exportable capital-intensive output, AI-driven productivity inside the firms that already exist, and a widening layer of firms of one.
If income security is attached to an employer, and employers are the part of the economy that is shrinking, then security has to be attached to something else. Finland's earnings-related unemployment insurance, pensions and sick pay were built for wage earners, and the 25,000 new entrepreneurs in this summer's data sit either outside them, or inside a heavily controversial and in-need-of-reform YEL system.
The question Finland faces is not whether growth without employment is possible, but whether the state can move its tax base and its safety net out of the payslip as fast as growth already has. The country with the EU's highest unemployment, and some of Europe's cheapest clean power, may be the first to have to answer it.
Disclosure: I work at Ververica, which is owned by Alibaba Group. Alibaba is among the companies that have announced data-center plans in Finland. I have no involvement in those plans, and the views here are my own.