You can’t see it or touch it, but it shakes markets and is an investment magnet. Artificial intelligence (AI) has become the object of desire of large technology companies, which allocate astronomical sums for its development in a climate of record profits. The other side of this fever are the cuts to the workforce, with automation in the background, which multinationals such as Amazon, Meta or UPS have announced and which, among other things, threaten to extend the impact of new technologies to another field: public coffers. Fewer people working means fewer taxpayers for the exchequer, so the question arises: if machines and algorithms replaced humans in their work, would they also have to shoulder the taxes they would stop paying?
The labor factor, through personal income tax and social contributions, is one of the pillars of the tax systems of virtually all countries, and the impact of automation on tax bases – or, in other words, the reduction it can generate in revenues – is not the first time it has caused concern. In 2019, Nobel Prize winner Edmund Phelps has proposed a tax on robots to help maintain social benefits. Shortly before, Bill Gates, founder of one of the largest technology companies in the world, Microsoft, equipped with its own artificial intelligence (Copilot), had done so: he suggested applying the same tax burden to robots that the worker they replace would bear.
“The trend toward automation and artificial intelligence could lead to a decrease in tax revenue. In the United States, for example, about 85% of federal revenue comes from earned income,” says Sanjay Patnaik, director of the Center for Regulatory and Market of the think tank American Brookings Institution. He suggests that governments address “the risks posed by artificial intelligence” by increasing taxation on capital rather than creating a specific tax on it, due to the difficulties in its design and the distortions it could generate. The repeated use of the conditional is due to the fact that the impact of generative artificial intelligence, the one capable of creating content on order, is still uncertain, both in positive terms (improvement in productivity and economic growth) and negative terms (destruction of jobs).
Nonetheless, the forecasts are crowded in both directions. Goldman Sachs, for example, estimates that artificial intelligence will increase global GDP by 7% over the next decade; The International Monetary Fund projects that it will contribute up to an additional eight-tenths per year to growth between now and 2030. On the other hand, the World Labor Organization estimates that one in four workers worldwide, concentrated in high-income countries, is in an occupation with some degree of exposure to AI, but at the same time predicts that the majority of jobs will be transformed rather than disappear.
“We know there will be an impact, but it is difficult to quantify it,” confirms Luz Rodríguez, professor of labor law and former Secretary of State for Employment. “The previous wave of automation hit jobs at the center of the production chain the most; generative AI is headed further upstream, towards higher-skilled jobs that require thinking skills,” he summarizes. “I’m not optimistic, but I’m positive: there are jobs that are being created that wouldn’t exist without new technologies, such as content moderators on networks or bitcoin miners.”
Daniel Waldenström, professor at the Stockholm Industrial Economics Research Institute, rejects the idea of a specific tax on artificial intelligence and in its favor argues that there has not been a significant increase in unemployment even in the United States, the cradle of new technologies and at the forefront of their implementation. Furthermore, he highlights the difficulty of circumscribing it: “What is automation, robots or artificial intelligence? A chip, a humanoid machine, an application or a computer program? We will never be able to define it precisely. We should continue to tax what already exists: income from work, consumption and capital gains.”
The International Monetary Fund (IMF) has also joined the debate. In a report released last summer, the agency’s economists reached a hybrid conclusion: They did not recommend specifically taxing artificial intelligence – it could slow productivity and distort the market – but urged states to remain vigilant in the face of possible disruptive scenarios. Among their proposals were raising taxes on capital – which has been falling as the tax burden on labor has tightened – creating a top-up tax on “excess” corporate profits, and overhauling tax incentives for innovation, patents, and other intangibles that, while boosting productivity, can also encourage the displacement of “human labor.”
Carl Frey, associate professor of artificial intelligence and work at Oxford University and author of the book How progress ends (Princeton University Press, 2025), takes a similar position: He does not support a tax on artificial intelligence, but he recognizes that the tax system has become unbalanced. “In many OECD economies we have seen an increase in income taxes and a decrease in capital taxes,” he recalls. A scheme that encourages companies to invest more in automation than in technologies that create jobs. “Addressing this imbalance is essential to supporting job-creating technologies of the future.”
The latest movements of large technology companies and the evolution of tax systems in recent years justify concern. Amazon, for example, announced a 38% increase in its profits and millionaire investments in artificial intelligence, notifying 14,000 layoffs worldwide, including 1,200 in Spain. Meanwhile, corporate tax rates have plummeted over the past decade in OECD countries, from 33% in 2000 to 25% today; The tax wedge for the worker – IRPF and contributions – decreased in the same period by only 1.3 percentage points, from 36.2% to 34.9%.
Susanne Bieller, general secretary of the International Robotics Federation, defends this principle by applying taxes ad hoc starts from “a problem that does not exist”, since automation and robots “create new jobs by increasing productivity”, and warns that taxing production tools rather than company profits “would have a negative impact” on competitiveness and employment. “We need incentives for (European) companies to use technologies such as robots and digitalisation to remain competitive on a global scale,” he says. He adds: “The world faces a labor shortage of around 40 million jobs per year (…). Robots cannot take over entire jobs, but they can take over certain tasks.”
Inequality
In addition to jobs, big tech companies’ skyrocketing spending on artificial intelligence and their escalating market shares are worrying, raising fears of a bubble. Analysts also warn that the energy consumption of these technologies is so high that their climate footprint could offset the benefits they promise for growth.
At best, new jobs created by AI can be “more productive, better paid and easily accessible” and offset job and revenue losses, Patnaik predicts. However, the latent – and very probable – risk that the process is not automatic would remain. New job creation may be delayed, less-skilled professionals may struggle to adapt, and a gap may arise between – and within – countries and productive sectors.
MIT economists Daron Acemoğlu and Simon Johnson had already warned about this in 2023. “Over the past four decades, automation has increased productivity and multiplied corporate profits, but it has not led to shared prosperity in industrial countries,” they warned in a paper for the IMF. “Technology and artificial intelligence produce social impacts that have to do with politics. We cannot allow technological determinism,” says Rodríguez. “The debate is necessary and we will go where we want to go.”
