A labour market that flipped in twelve months
A year ago, unemployment in Iceland was the kind of number nobody bothered to put in a board deck. Then it nearly doubled. Statistics Iceland counted 15,400 unemployed people in the second quarter of 2026, 6.5% of the labour force, against about 8,400 (3.5%) a year earlier. There are 9,000 fewer people in work, in a labour force of just 237,700.
The monthly numbers haven't brought any relief since: seasonally adjusted unemployment was 6.8% in August. Back in February, Íslandsbanki was forecasting a 4.5% average for the whole of 2026. The spring figures blew straight past it.
| Statistics Iceland indicator | Q2 2025 | Q2 2026 |
| Unemployed, ages 16–74 | about 8,400 | 15,400 |
| Unemployment rate | 3.5% | 6.5% |
| Youth unemployment, ages 16–24 | 7.6% | 17.3% |
| Employment rate | about 80% | 75.8% |
There wasn't one single shock. Íslandsbanki pointed to hits in tourism and metals manufacturing in the second half of 2025. Construction cooled too, with nearly 1,200 people in construction and civil engineering unemployed at the end of June. Iceland Review adds a broader slowdown and the pressure high inflation puts on wage agreements.
Here's the odd part. A Gallup survey of executives, run early this year and published by SI, the Federation of Icelandic Industries, in September, found that 32% of tech and IP leaders were short of staff. Across business as a whole the figure was 21%. So Iceland has nearly twice as many people out of work as it did a year ago, and its tech sector still can't hire.
Related reading: How Nordic Employers Are Beating the 2026 Tech Talent Shortage · Continuous Learning Funding in Finland 2026: JOTPA After the Allowance Cut · Norway Energy Workforce Reskilling in 2026: Follow the Grid, Not the Turbines
The 9,000-specialist gap, sized honestly
Start with what the sector is worth. SI puts tech and IP export revenue at ISK 369 billion in 2025, up from ISK 46 billion in 2008. That's 18.6% of all Icelandic exports (it was around 7% in 2008) and roughly a quarter of all export growth since then. About 18,000 people work in it, across software, games, pharma, biotech and a long tail of smaller high-tech firms.
Then the headline number. SI's analysis, published in 2024 and based on a survey of its IP-sector members, says that if companies deliver on their growth plans, the sector will need up to 9,000 more specialists over five years, around 1,800 a year. The Gallup figures point the same way. 74% of tech and IP leaders expect to add staff within five years (47% across all business), and 35% think staff shortages will hold back growth (25% overall). Only about 40% believe the education system will meet their skills needs.
Now the honest part. Nine thousand is a ceiling built from growth plans, and growth plans are optimistic by design. It also dates from 2024, before the downturn, and it's half the sector's current headcount. If you're a vendor, don't multiply 9,000 by your course fee and call it a market. Our read: the sector wants to grow a lot, can't find enough university-educated specialists to do it, and will pay for anything that moves people into productive roles faster than a three-year degree. Even at half of SI's estimate, that's around 900 people a year in a country whose entire labour force is under 240,000.
Why 6.5% unemployment doesn't staff a data centre
Because the people losing jobs and the jobs going unfilled barely overlap. The Gallup shortage sits mainly among university-educated specialists. The new unemployment shows up among young people, in construction, tourism and heavy industry, and disproportionately among foreign nationals. Directorate of Labour data reported by mbl.is shows that foreign citizens, roughly a quarter of the workforce, made up 56.7% of unemployment-benefit recipients between January and July 2026. On average that was 5,210 foreign citizens against 3,980 Icelandic citizens, and they drew ISK 13.2 billion of the ISK 23.5 billion paid out.
So here's our view. Iceland can't import its way to 9,000 specialists, and it can't turn laid-off tourism and construction workers into machine-learning engineers either. Anyone selling you a twelve-week bootcamp that promises the second is selling you a brochure.
The realistic middle is adjacent roles. Data-centre technicians. QA and test engineers for game and software studios. Cloud operations. Support engineering and IT service desks that run in English. These jobs need real training, but they reward habits people bring from shift-based, safety-heavy or customer-facing work, and they free up senior engineers' time (arguably the scarcest thing in a small tech team).
There's a second group buyers tend to overlook: foreign jobseekers whose qualifications from home were never formally recognised in Iceland. Picture someone with an electrical engineering diploma from Gdańsk who has spent three winters on a hotel front desk. The action plan promises to harmonise how prior learning is recognised (raunfærnimat, in Icelandic). If it delivers, people like that become your fastest reskilling wins.
What the 2026–2027 action plan actually funds
In April 2026 the Prime Minister's Office published Atvinnustefna Íslands, Aðgerðaáætlun 2026–2027, the first action plan under the government's growth plan to 2035. For training buyers and sellers, these are the lines that matter:
- ISK 200 million a year to expand micro-credentials (örnám) and continuing education at university level, including money to develop micro-credentials together with employers.
- A law giving public universities a clear right to charge fees for micro-credentials, plus incentives in school funding models to offer more of them.
- A new partnership between government and the social partners on professional development and lifelong learning.
- A separate ISK 200 million a year aimed at more STEM graduates at university level.
- A new 30,000 m² Tækniskóli in Hafnarfjörður, and ISK 500 million to fund more students in vocational and technical programmes.
- ISK 1.1 billion for AI compute infrastructure and the first phases of an Icelandic language model.
- One channel for foreign specialists' work and residence permits at Útlendingastofnun, the Directorate of Immigration. The law change behind it has already landed: Alþingi passed it on 18 June and it took effect on 8 July, moving all work permits over from the Directorate of Labour. The plan's other pieces, an updated regulation for specialists and rules for prioritising applications, carry an autumn 2026 date.
- A longer run for the 25% income-tax reduction foreign experts get in their first three years, with legislation planned for autumn 2026.
- A clearer tax and regulatory framework for data centres, also scheduled for autumn 2026.
Put the micro-credential line in perspective. ISK 200 million is roughly €1.4 million a year. That's seed money, not a market. What it does is tell you which way public universities are being pushed, and it gives them a reason to go looking for industry partners and content.
Micro-credentials: the quiet opening for training providers
Most foreign vendors will read the fee clause as a footnote. We think it's the most commercially interesting line in the plan.
Public universities are already learning to charge. In June, Alþingi let them bill tuition to students from outside the EEA, a right only private universities had before. Micro-credentials come next: a public university's right to charge for a short, credit-bearing course isn't clearly set out in law, and the plan commits to legislating it. Once that law passes, the University of Iceland, the University of Akureyri and the other public institutions can sell örnám with a price tag attached. That changes three things for you:
- Public universities become buyers. They'll need course design, assessment, platforms and instructors with industry credibility faster than they can hire them.
- They also become competitors. Private providers, including privately run Reykjavík University, will be pitching against priced short courses from public peers.
- Employers get a credential they can co-fund and defend. A micro-credential with a university's name on it is an easier sell to a finance director than a vendor certificate.
If you're a foreign EdTech company, partner rather than enter cold. Iceland is small, a local university's name carries weight, and the plan explicitly funds micro-credentials built with industry. A three-way deal (university, employer consortium, content partner) looks to us like the shape most likely to win a slice of that money. Price it as a ladder, not a course: a first module that makes someone job-ready as a technician, a second that moves them into junior cloud operations, an optional third for the ones who want to keep climbing.
For enterprise buyers, the contract matters more than the syllabus. Ask for outcome clauses tied to placement into named roles inside your company, not completion rates. Ask who owns the content if the university partnership ends. And put paid training time in the budget from day one, because the course fee is often the smaller number.
Is importing talent still the fastest route?
For some roles, yes. You won't grow a senior graphics programmer or a biotech regulatory lead from a micro-credential. For those, imports stay, and policy is moving your way. Approved foreign experts already pay income tax on only 75% of their pay for their first three years (applications go through Rannís to a special committee), and the action plan promises legislation this autumn to make the relief last longer. Permit handling changed in July too: the Directorate of Immigration took over work permits from the Directorate of Labour, so one agency now handles both.
But don't overrate the tax relief. It matters at the offer stage, and most hiring problems start well before that. Intra-company transfers are still awkward, too: SI points out that the permitted stay is 12 months in Iceland against an OECD average of 32 months, and calls it a regulatory barrier. If your plan relies on bringing people from a sister office for two years to train a local team, you'll need a different structure.
How the four main routes compare. Lead times are our rough estimates, not official processing times.
| Route | Lead time (our estimate) | 2026 policy change | Cost signal | Best suited roles |
| Hire foreign specialists | Several months, much of it permit time | Útlendingastofnun has handled work and residence permits together since 8 July 2026; extension of the 25% three-year tax reduction planned for autumn 2026 | International salary plus relocation, partly offset by the tax relief | Senior ML, security, graphics and biotech specialists |
| Reskill the unemployed through micro-credentials | One to two semesters | ISK 200m a year for örnám; fee-charging at public universities to be legislated; prior-learning recognition to be harmonised | Course fees plus paid training time at local wages | Data-centre technicians, QA, cloud operations, support engineering |
| Graduate and STEM pipeline | Three years or more | ISK 200m a year for STEM graduates; new 30,000 m² Tækniskóli; ISK 500m for more vocational students | Internships and sponsorships now, payback later | Junior developers, technicians, lab staff |
| Intra-company transfers | Depends on your existing staff | None in the April action plan; stay capped at 12 months vs a 32-month OECD average | Expat packages for a short stay | Knowledge transfer, launch teams for a new site |
Húsavík and Vatnsmýrin: two hiring problems, one playbook
In Reykjavík's Vatnsmýrin district, CCP Games became Fenris Creations in May after a USD 120 million management buyout from Pearl Abyss. The EVE Online studio said there'd be no layoffs, announced a research partnership with Google DeepMind and took Google on as a minority investor. It reported more than USD 70 million in 2025 revenue and has studios in Reykjavík, London and Shanghai. That's a nice hiring problem to have. It can recruit almost anywhere, and its friction is mostly paperwork: permits, plus that 12-month cap if it wants people from London or Shanghai in Reykjavík for longer.
On the north coast the picture looks nothing like that. At the Bakki industrial site outside Húsavík, a town of about 2,400 people, PCC BakkiSilicon paused silicon production in July 2025 and let around 80 people go, then another 30 that September. Now Akur wants to put Eyrie DC in the same industrial area: a 100 MW data centre costing around ISK 120 billion (about USD 1 billion), with 400 to 500 construction jobs and 50 to 70 permanent ones. Construction could start in 2027 and operations in 2028. Akur is in talks with Landsvirkjun for power and with a large US AI company as a customer, and there's no final investment decision yet.
Fifty to seventy permanent jobs sounds small. In a town that size it isn't. You won't staff a round-the-clock operations floor in Húsavík by flying people in, and we doubt Iceland has spare experienced data-centre technicians waiting for the call. atNorth said in 2025 that about a third of its Iceland-based staff work in Akureyri, where it broke ground on a new service building for its regional team. That's regional data-centre hiring when it works: local people, trained locally.
The playbook is the same in both places. Only the weights change.
- Map adjacent roles before you post a single ad. At Bakki, former smelter electricians and process operators are closer to data-centre facilities work than any graduate scheme. At a studio, it's QA, build and release, and player support moving towards engineering.
- Build the ladder with a university that will put its name on it. In the north, the University of Akureyri is the obvious partner.
- Import for the top rungs only, and make the imported seniors teach.
- Write the training contract around placement, not attendance.
If you sell training into Iceland, the next twelve months matter more than the next five years. The fee law, the social-partner partnership and the first rounds of that ISK 200 million all get designed inside the 2026–2027 plan period, and whoever helps design them writes the templates everyone else follows. If you're an employer holding your training budget until the tax-relief extension clears, you're waiting on the wrong lever. The tax relief makes a hire cheaper. It doesn't create one.