AI: growth without employment?
The US economy looks to be on the right track, with GDP growth revised up to 3.8% on an annualised basis for the second quarter, due, in particular, to a stronger than expected contribution from consumption. Granted, annualised growth for the first half of 2025 is only 1.6% – a level far below that to which we have been accustomed in the US in recent years but nothing out of the ordinary by the standards of the past.
However, in tandem with this resilient growth, the employment market is at a standstill. Job creation has been barely positive in recent months and clearly negative if we look at the private sectors that are sensitive to the economic cycle1. The unemployment rate has risen only moderately, mainly because of the many people leaving the workforce without joining the ranks of the unemployed. And whilst there is currently no rush to lay people off, corporate workforces are nonetheless under pressure due to a lack of new hires and the non-replacement of leavers.
The boom in Artificial Intelligence provides some answers to these apparent inconsistencies. Firstly, with regards to the resilience of US growth, investment in IT equipment represented 70% of the total investment volume made in the US in the first half of 2025 and accounted for close to half of real GDP growth. Significantly, this component alone contributed almost as much as private consumption, the major driver of the US economy. In other words, if we exclude IT investment, primarily in AI, the US economy would have been close to stagnation over the last two quarters.
This would be consistent with the sluggish employment market, all the more so given that AI is directly involved here. According to a recent study by Stanford University2, the adoption of AI has acted as a major brake on the hiring of young graduates in sectors and jobs that are most exposed to replacement by AI. In these sectors, since the end of 2022, the employment of young workers between the ages of 22 and 25 has fallen by 13% in comparison to those sectors that have the lowest exposure to AI. This is particularly striking among software developers and in customer support, whilst jobs that are not under threat from AI, such as health aides, have seen no such decline.
This study corroborates statements from some companies confirming that they prefer to train a Generative AI3 model than to train up a junior. In particular, it highlights data on the employment of young workers to be found in the latest official US employment reports. Whilst the national unemployment rate has risen modestly from 4.1% at the end of 2024 to 4.3% in August 2025, unemployment among young workers has jumped from 8.4% to 10% over the same period. Of course, uncertainty linked to the economic environment and, in particular, to US trade policy, is having an undeniable impact on companies’ willingness to hire. But it is significant that the sharpest decline is to be found in job opportunities for workers at the start of their careers in sectors with the greatest exposure to AI.
This atypical situation could be portrayed in a positive light – robust growth without the creation of new jobs implies a strong rise in productivity. Yet it is valid to question the sustainability of this explosion in IT investment – the main driver of recent growth – particularly as, sooner or later, the issue of the profitability of such colossal investment will be raised. Furthermore, the concentration of investment into a single sector will not necessarily result in transmission to the economy as a whole, thus hiding underlying weaknesses beneath the surface of resilient growth. Such areas of weakness include employment, with a knock-on effect for consumption. Once the wave of AI investment begins to ebb, how vulnerable will the US economy look?
Final version of 26 September 2025, Enguerrand Artaz, Strategist, La Financière de l’Échiquier (LFDE)
[1] With the exception of the Educational Services, Health Care, and Social Assistance sector which has been countercyclical for several decades.
[2] Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence, Brynjolfsson, Chandar & Chen, 26 August 2025