AI is not destroying jobs where it is most adopted, but it is concentrating them.

Artificial intelligence has entered the job market surrounded by two opposing narratives. One announces mass layoffs; the other promises productivity, growth, and new jobs. A new study by Ramp and Revelio Labs adds a layer of real data to this discussion: companies in the United States that invest most in AI have not reduced their workforce after adopting it but have actually increased employment by around 10% over the following two years.

The nuance matters. The data alone does not demonstrate that AI automatically creates jobs, nor does it negate that certain occupations and tasks are under pressure. It is an observational study, and the authors themselves caution that companies with higher levels of AI adoption were already more technical, paid higher wages, and experienced faster growth before integrating AI. But it does require us to change the question: perhaps it’s not enough to ask how much AI will destroy jobs. We need to look at where jobs are being created, which companies are doing so, and what kind of adoption is taking place.

The study: 21,559 companies and actual AI spending

The paper, titled A New Look at AI’s Impact on Jobs: Firm-Level AI Spending and Workforce Adjustment, combines actual payments to AI vendors recorded by Ramp with monthly workforce data from Revelio Labs for 21,559 U.S. companies. This combination allows us not only to see whether a company is ‘potentially’ exposed to AI based on the tasks it performs but also whether it is actually paying for AI tools and how its workforce changes afterward.

Sustained adoption is defined as at least three consecutive months with a expenditure of $100 or more on AI vendors. The researchers then categorize companies by intensity: low intensity, with an average of $2.78 spent per employee per month on AI; and high intensity, with $33.67 per employee. This distinction is relevant because positive results appear almost exclusively in the high-intensity group.

The immediate takeaway is that AI adoption is not neutral. Companies that adopt earlier and more intensively are different—they tend to be younger, more tech-savvy, more productive, or better financed, and have greater capacity to turn tools into actual processes. This selection bias limits the conclusions but also explains why a simple chatbot license does not produce the same effect as a serious adoption strategy.

More total employment and also more entry-level jobs

The main result is that high-intensity AI companies increased their total workforce by 10.2% over the 24 months following adoption, whereas low-intensity adopters showed no significant change. The increase in entry-level positions was even greater: 12%. This contradicts one of the most repeated hypotheses in public debate—that AI would primarily eliminate lower-level jobs first.

The increases are not limited to engineering. Sales, administration, customer service, finance, and scientific profiles also grow. Operations is the only category where the study does not find a clear increase among intensive adopters.

This pattern suggests that companies truly using AI do not only automate tasks but can also accelerate product development, increase sales, serve more clients, launch new lines, or reduce internal costs, making further hiring justified. While this does not eliminate the risk of task-specific substitution, it shifts the debate from individual jobs to a company’s growth capacity.

Spain, Europe, and the US: similar adoption, still unequal effects

The international comparison provides an interesting snapshot. Spain, the European Union, and the United States are now roughly at 20% of companies declaring AI use, although the methodologies differ slightly. In Spain, INE estimates that 21.1% of companies with 10+ employees used AI in Q1 2025, up 8.7 points from a year earlier. Eurostat reports 19.95% in 2025 for companies with at least 10 employees. The U.S. Census Bureau estimates that AI use was between 17% and 20% from December 2025 to May 2026, with 19.8% as of May 3, 2026.

The convergence around 20% might be misleading. Not all adoptions are equal. In Europe, Eurostat shows AI use at 55.03% among large companies but drops to 17% among small ones. In the U.S., the Census Bureau reports 37% of companies with 250+ employees and less than 20% among those with four or fewer employees.

Spain aligns well with this gap. INE reports 21.1% of companies with 10+ employees using AI, but only 13.4% of those with fewer than 10. Sectorally, services lead at 25.7%, followed by industry at 17.5%, and construction at 11.4%.

Regional disparities also matter. According to INE, Madrid reaches 30.1% of companies adopting AI, Catalonia 25.6%, and Castilla-La Mancha 11.1%. AI thus not only separates large and small firms but potentially amplifies regional differences—those regions with more tech services, corporate headquarters, and digital talent lead in adoption, while less digitalized areas lag behind.

Jobs will grow where adoption capacity exists

Ramp and Revelio Labs’s findings should not be interpreted as a guarantee that AI will be positive for the job market overall. In fact, the study shows most gains concentrate in the Information sector—including many software, internet, media, and tech firms. In this group, high-intensity adopters increased their workforce by 13.4%. In other sectors, the results are weaker or not statistically significant.

This concentration aligns with European data. Eurostat highlights that the information and communication sector is the most AI-intensive in the EU, with 62.52% of firms using AI, followed by scientific and technical activities at 40.43%. In other sectors, adoption remains below 25%.

The real challenge for Spain and Europe is that if employment gains predominantly occur in tech-savvy, process-ready companies, the risk is not only job destruction but also the concentration of new employment and productivity boosts within a limited number of firms, sectors, and regions.

The Bank of Spain already noted in 2025 that nearly 20% of surveyed Spanish companies used AI, but most were still in experimentation phases. It also identified three barriers: lack of skilled personnel, implementation costs, and data availability. This explains why simply purchasing licenses isn’t enough. Useful AI requires clean data, clear processes, security, system integration, and teams capable of working differently.

Less apocalypse, more industrial policy

The U.S. study urges caution against headlines attributing layoffs solely to AI. Some companies may use AI as a convenient explanation for adjustments driven by costs, market shifts, financing, or strategic changes. But it would also be a mistake to assume automatic optimism. High AI usage correlates with growth in some companies but does not guarantee it across all.

For a Spanish or European company, the practical question isn’t “Do we have AI?” but “Do we have a real way to turn it into productivity?” The difference between low and high intensity in the study suggests that results are seen where there is sustained investment and process redesign.

This implies a shift in mindset for executives, workers, and policymakers. For managers, AI adoption should be viewed not as a software purchase but as a process redesign. For workers, the message isn’t to abandon exposed sectors but to seek companies capable of using AI to grow. For governments, the challenge is to prevent adoption from remaining confined to large corporations, tech firms, and regions with more digital infrastructure and talent.

The comparison between Spain, Europe, and the U.S. underscores a straightforward conclusion: adoption is around 20%, but the decisive phase is just beginning. The labor impact will depend not only on how many companies claim to use AI but also on how many can leverage it effectively, embed tools into real operations, and create new roles around this increased productivity.

AI can wipe out tasks but also generate demand, accelerate product development, and enhance the value of complementary work. The labor market won’t shift uniformly; it will divide between companies turning AI into growth engines and those stuck in pilot projects, headlines, and underutilized licenses.

Frequently Asked Questions

Does the study demonstrate that AI creates jobs?
Not definitively in causal terms. It shows a strong association between intensive AI adoption and workforce growth in comparable U.S. companies, but the authors warn that adopters were already different before using AI.

Is Spain lagging in AI adoption?
According to the latest data from INE and Eurostat, no. Spain is slightly above the EU average for companies with 10+ employees, though clear gaps exist by size, sector, and region.

Which companies are more likely to create jobs with AI?
According to the Ramp and Revelio Labs study, those that invest continuously, possess greater technical capacity, and operate in sectors like information, software, internet, media, and professional services.

More information at Noticias Inteligencia Artificial