The Talent Equation: the August OUTLOOK report.
The evidence on artificial intelligence and entry-level work is descriptive rather than causal, but it now runs consistently in one direction. For employers in Switzerland and Europe, the question has moved from whether AI displaces jobs to where the entry point into a company will sit in 2030.
Insights yazıları İngilizce yayımlanır.
A revised version of the Stanford Digital Economy Lab’s study of early-career employment, published in August 2026, extends its evidence to payroll records through June 2026. Its central observation, from the authors Erik Brynjolfsson, Bharat Chandar and Ruyu Chen, is that employment of workers aged 22 to 25 in AI-exposed occupations “now stands 19% below where it would be had it kept pace with that of their less-exposed peers”. The same data support a second finding: there is no evidence of widespread, economy-wide job displacement.
Both come from the same tables. The aggregate reassurance and the entry-level warning are not in tension: they describe different parts of one labour market, and it is the second that decides who will be qualified to run a company when today’s senior staff retire.
The gap in early-career employment widened again, and it runs through hiring
The paper draws on ADP payroll data covering millions of United States workers. Employment of 22-to-25-year-olds in the two most AI-exposed occupational quintiles fell by about 11% between November 2022 and June 2026, while the same age group in the three least-exposed quintiles grew by about 10%. Experienced workers show no comparable gap.
Three mechanical details matter more than the figure. The divergence operates primarily through reduced hiring rather than increased separations. It is concentrated where AI substitutes for human tasks; where use is complementary, employment is flat or rising. And it appears in employment rather than in base compensation.
The number has moved as data accumulated: regression estimates gave 13% on July 2025 data and 16% on September 2025 data; the descriptive measure now headlined stood at 15% on July 2025 data and 19% through June 2026.
The authors are explicit about the limits: they present the results as “early, descriptive indicators — canaries in the coal mine — rather than causal estimates”, and note that the pattern is more pronounced in the ADP sample than in national benchmarks. It is also American evidence. What transfers to Europe is the mechanism rather than the percentage: a labour market adjusting at the point of entry rather than the point of exit.
Europe’s adoption divide runs by company size, and the barrier firms name is expertise
Eurostat’s survey of ICT usage in enterprises, with 2025 results extracted in December 2025, sorts adoption by size: 55.0% of large EU enterprises used AI technologies in 2025, against 30.4% of medium-sized and 17.0% of small ones. Among enterprises that considered AI and decided against it, 70.9% cited a lack of relevant expertise, ahead of uncertainty about legal consequences (52.5%).
The constraint firms report is people, not equipment — and that market is thin. Eurostat counts more than 10 million ICT specialists in EU employment in 2025, 5.0% of the total, after growth of 59.4% since 2015 against 9.8% for employment as a whole; the European Union’s Digital Decade objective is 20 million by 2030. In 2023, 57.5% of enterprises that tried to recruit an ICT specialist had difficulty filling the vacancy — and 51.9% of large enterprises tried, against 6.2% of small ones. Smaller firms mostly stay out of that market, which makes the skills they build internally the limit on what they can adopt.
The Swiss labour market is loosening at the edges while remaining tight at the centre
On 18 June 2026 the Federal Government’s Expert Group forecast Swiss GDP growth of 0.9% in 2026, well below the historical average, and 1.6% in 2027, with unemployment averaging 3.1% this year and 3.0% next and full-time-equivalent employment rising 0.5% then 0.7%. The same day the Swiss National Bank left its policy rate unchanged at 0%, observing that unemployment had risen somewhat since its previous assessment.
In its labour-market report for July 2026, published on 6 August, SECO recorded 139,276 registered unemployed — a rate of 3.0%, or 3.1% seasonally adjusted, and 7.8% more than a year earlier. Long-term unemployment rose 29.9% year on year to 23,327. Among 15-to-24-year-olds, 12,294 were registered unemployed, 6.9% more than a year earlier, a youth rate of 2.8%. The same month, 45,156 vacancies were notified to the regional employment centres, 13.4% more than a year earlier.
Vacancies and unemployment rising together describe a matching problem rather than a collapse in demand. Eurostat, using the international definition rather than Swiss registration, put EU unemployment at 6.0% in June 2026 and under-25 unemployment at 15.5%; the two are not comparable, but the Swiss market remains tight by any European standard, and the cyclical case for cutting junior intake is correspondingly weak.
Two thirds of young Swiss people enter work through a training contract, which changes the exposure
The State Secretariat for Education, Research and Innovation states that two thirds of young people in Switzerland acquire their occupational foundation through vocational training. Entry-level employment here is not an annual graduate-hiring decision but a multi-year contract inside a company, with a curriculum and a federal certificate at the end.
That structure slows adjustment: apprenticeship places cannot be cut as fast as a hiring freeze takes effect, so a Swiss echo of the American pattern would appear later and more quietly — as places not offered rather than as departures.
It also concentrates the exposure. The Stanford analysis finds employment declines for young workers in occupations involving codified knowledge, while occupations involving tacit knowledge show faster growth for experienced workers. Codified knowledge — the documented procedure, the standard letter, the first-pass calculation — is very largely what apprentices produce while they learn. When that output becomes inexpensive, the training place has to be justified by what remains: customer contact, judgement in non-standard cases, verification of machine output, and the tacit knowledge that the same evidence shows holding its value.
The skills horizon to 2030 is an employer expectation, not a technology forecast
The World Economic Forum’s Future of Jobs Report 2025, published in January 2025 and based on more than 1,000 employers representing over 14 million workers in 55 economies, reports that workers can expect 39% of their skill sets to be transformed or outdated between 2025 and 2030. That measure of skill instability has fallen from 44% in the 2023 edition and 57% in 2020, which the report suggests may reflect more workers having completed training — 50%, against 41% two years earlier.
The training task is expressed as a proportion: if the world’s workforce were 100 people, 59 would need training by 2030, and 11 would be unlikely to receive it. Skills gaps are named as the biggest barrier to business transformation by 63% of employers; 85% plan to prioritise upskilling and 40% expect to reduce staff where AI automates tasks. These are stated intentions rather than measurements.
What a management team can decide this quarter
Four decisions follow from the evidence rather than from prediction.
- Separate automation from augmentation, task by task. Declines cluster where AI substitutes for work, not where it complements it. List the tasks that made a junior employee productive in the first year, and mark those a system now does.
- Keep the intake and change its content. The observed adjustment runs through hiring, not dismissal. A company that quietly stops offering training places in 2026 has decided the composition of its senior staff in 2035 without recording it as a decision.
- Budget expertise before licences. Missing expertise was cited by 70.9% of EU firms that considered AI and declined it; a named internal owner and protected training hours address that, another subscription does not.
- Measure the pipeline directly. Reduced hiring is invisible in headcount reports, which show a stable total while the age profile shifts. Track the share of staff under 25, training places offered against places filled, and the time a new entrant needs to work unsupervised.
Three public series carry the next information at no cost: SECO’s monthly labour-market report; the Expert Group’s next forecast, scheduled for 17 September 2026; and Eurostat’s annual enterprise ICT survey.
The pipeline that produced today’s experienced staff was largely automatic: young people entered, did routine work, and learned the trade while doing it. That mechanism is being altered at its first stage. The firms with a strong senior bench in 2035 are likely to be those that treat the entry level as a design problem now.
The full August OUTLOOK report, The Talent Equation, is available on request at info@sdav.ch.