Artificial intelligence promises to revolutionize productivity, work, and, possibly, our way of life. A few days ago, in these pages, Andreu Mas-Colell reflected on the future of work and Cal Newport wondered if we will end up going extinct. These are not minor debates.
But before AI transforms the world, it is already transforming the economy. And, for the moment, it is not doing so much because of the productivity gains it generates as it is because of the investments it is mobilizing. This is especially visible in the United States, where AI has become the main engine of growth thanks to colossal investments in programmers, chips, data centers, and energy. Meta, for example, is building a data center that will be able to consume as much electricity as all of Catalonia and will occupy more space than the entire Eixample of Barcelona.
A second mechanism is added to this momentum. Stock markets have risen sharply and a good part of this increase is linked to the expectations generated by AI. When investors see the value of their assets grow, they tend to consume more. It is the well-known wealth effect, which contributes to stimulating economic activity. Regarding employment, the aggregate balance in the United States is positive. Until now, the demand for engineers, technicians, electricians, and construction workers has more than compensated for the losses in some occupations.
The impact of AI is not limited to growth and employment. It is also leaving its mark on American inflation. In the long term, it could help to moderate prices thanks to productivity gains. However, nowadays, the demand for electricity and the needs for infrastructure are pushing costs upward. On the energy issue, we must take note. If the energy transition already requires significantly expanding generation capacity and electrical grids, the deployment of AI intensifies this need.
The consequences of AI are also being felt in long-term interest rates, which have risen. After all, these rates reflect the balance between the supply of savings and the demand for investment. And this demand has skyrocketed. This movement coincides with two other factors that are also pushing yields upwards: the gradual withdrawal of liquidity injected by central banks during the pandemic and the persistence of very high public deficits in several countries. The United States is the most obvious example, but we also find the United Kingdom, France, and Japan there. Spain does not stand out negatively for its deficit, but it continues to have a relatively high level of debt that should be reduced.
AI also poses an inevitable question: are we facing a new bubble? I do not know, and I think it is still too early to tell. The expectations associated with AI could end up being disappointing in two different ways. The first is that productivity gains are lower or arrive later than expected. The second is that, even if these gains materialize, AI companies may not be able to charge high enough prices for their services and make their investments profitable.
And current expectations are enormous. The ten main companies linked to AI, all American, have a combined valuation of approximately 30 trillion dollars. It is almost four times more than what they were worth when ChatGPT burst onto the scene in November 2022, and fifty times higher than the capitalization of Europe's most valuable company, the Dutch firm ASML. If, at some point, expectations were revised downwards, stock prices would fall and, at the same time, investments linked to AI would be reduced. Both effects would act in the same direction: less economic growth, especially in the United States.
It is also worth monitoring the risks to financial stability. Some observers argue that the concentration of investments in a small group of companies and with cross-investments between them increases these dangers. I tend to think the opposite. The fact that a very large part of the spending falls on tech giants with solid balance sheets and an extraordinary capacity to generate profits can reduce the usual risks associated with bubbles.
Another important question is: who assumes the risk in the rest of the chain? Who is taking on debt to build infrastructure, provide services, or take advantage of the AI boom without having such solid balance sheets or contracts that offer good enough revenue prospects? This is where it will probably be necessary to watch more closely. The question is not what will happen to the big tech companies, but who is betting behind them without having the same capacity to absorb possible errors.
