In 2025, in the US, 410 billion were invested in AI and the plan Big technology is to spend about 650,000 million in 2026 (thousand up, thousand down). The promise behind this madness is that AI is synonymous with wealth; With AI we are more productive, we do more and faster, and companies can spend less on staff. Economists have something to say.
Zero. It is the impact that investment in AI has had on the United States economy and It was said by none other than Jan Hatziusthe chief economist at Goldman Sachs, a global leader in investment banking. According to Hatzius, there is a lot of misinformation about the impact that AI investment is having on the US GDP and it comes from the error of assuming that because you invest a lot, you gain a lot. As they point out in FuturismGoldman Sachs’ rhetoric has been hardening; First they were subtle warnings about the dangers of investing so wildly in AI and now they directly say that in 2025 the contribution was insignificant or, in their words, “basically zero.”
Debate. During 2025, speeches were circulating that placed AI as responsible for a large part of the country’s GDP growth, some with figures as optimistic as 92%. The economist Hanna Rubinton conducted an analysis somewhat more restrained, but also optimistic, in which he stated that spending on AI had contributed 39% to economic growth during the first nine months of 2025. However, he recognizes that it includes spending on software and computers that was not necessarily linked to AI, so the figure could be inflated. Goldman Sachs has not been the only bank to pour cold water on AI enthusiasts, Economists from JP Morgan and Morgan Stanley agree in which the real figure is closer to zero.
Distorting the economy. That there is such a variety of analyzes has largely to do with the difficulty of knowing the true impact that AI is having on the economy. Already in November of last year, in Reuters They reported that the investment boom in AI was distorting the figures and making it very difficult to read the true situation of the country’s economy. On the one hand, the GDP registered a growth of 4%, but layoffs increased, perhaps partly due to AI. They called it the “bifurcated economy.”
The geographical problem. In his invention, Jan Hatzius pointed out a fact that often goes unnoticed: to build the infrastructure that drives AI, many GPUs, memories and components are needed that are imported from other countries. “A lot of the investment in AI that we are seeing in the US actually adds to the GDP of Taiwan or Korea, but not really that much to the GDP of the US,” he said.
The problem of productivity. It is the great promise of AI, that thanks to it we become more productive. However, the speed, quality or quantity of work produced are intangible aspects that do not always translate into an immediate economic return, much less one that has an impact on the global economy, but rather the improvements remain “trapped” within the companies.
What is coming. If the 2025 investment was already truly crazy, the forecasts for 2026 are even crazier. The combined capex of large technology companies amounts to around $650 billion, which would be the equivalent of spending $1.2 million per minute for an entire year. There are those who think that the AI bubble does not existbut of course the economic return on this tremendous investment is, at the very least, debatable.
Image | Unsplashedited
In Xataka | Investing in data centers for AI is insane, and it’s going to get worse. much worse
GIPHY App Key not set. Please check settings