Artificial intelligence has been growing rapidly for several years, but recently that growth has become explosive. At first glance, it seems to have no ceiling. However, signs are beginning to appear that, at the very least, call for caution.
AI wasn't born yesterday. For years, it advanced relatively discreetly, integrated into search engines, machine translations, and recommendation systems. Everything changed when OpenAI launched ChatGPT. From that moment on, public and financial perceptions of AI exploded. It seemed as if the future had arrived suddenly.
Expectations skyrocketed. People began talking about radical transformations in work, education, medicine, and industry… and in the very short term. To achieve this, it was necessary to invest enormous sums in specialized chips, data centers, and energy, with the goal of training increasingly larger and more powerful models.
In a very short time, dozens of new AI companies emerged. Many of these companies reached valuations in the billions of dollars without yet generating any real profits. The strategy was clear: lose money today to dominate the market tomorrow.
The most striking example is OpenAI, which was once valued at nearly $500 billion, based on the premise that it could be profitable in about five years. Meanwhile, its quarterly losses are in the hundreds of millions.
Other tech giants like Google, Microsoft, and Amazon have also invested heavily in AI, but with a key difference: they do so with profitable businesses behind them, capable of absorbing losses and financing the transition without jeopardizing their survival.
The problem arises when we look at many AI startups. Their business model depends almost entirely on constant capital injections, and a significant portion of that money comes from bank loans, not just venture capital.
This is where the discomfort begins. The rapid growth of AI has been largely financed by debt. Debt to buy chips, build data centers, and pay enormous energy bills. Debt backed by future expectations, not current profits.
Some analysts are already openly discussing a potential AI bubble. Not because the technology isn't real or useful, but because cash flows aren't keeping pace with the enthusiasm.
In this context, reports have begun circulating that major investment banks, such as Morgan Stanley and others, are starting to hedge against the risk associated with their loans to the AI sector. In banking terms, this means preparing for the scenario in which some of these loans aren't repaid on schedule… or aren't repaid at all.
Banks don't usually make these moves on a whim. If they start protecting themselves, it's because they see signs of increasing risk in the repayment capacity of some of their clients.
I'm no banking expert, but the story is well-known. And there's a phrase that sums it up: When the banks sneeze…
When large financial institutions stop making money, someone ends up footing the bill, and it's rarely those directly responsible. Adjustments, credit cuts, disguised bailouts, or crises of confidence usually end up affecting the real economy at the expense of the poor.
This doesn't mean that AI is going to collapse tomorrow. Artificial intelligence is here to stay and will continue to transform the world. But it is possible that we are entering a cooling-off phase, where expectations adjust to economic reality.
I don't have a crystal ball. But when banks start getting nervous, it's worth paying attention. Because technology can be revolutionary… but in the end, the numbers always have to balance.
EN 
ES