Fynsa
For those of us who lived through 2008, the cost of creating "too big to fail" systems, that is, systems so large and concentrated that, when they fail, they must be bailed out to avoid a major collapse, became clear. The banks became so important that the state had to intervene to prevent their collapse. Today, 17 years later, it seems that artificial intelligence is going down a similar path.
Over the last few months, the results of the big tech companies have confirmed that investment in AI is not stopping. Every day there is something new that surprises us and reinforces the feeling that this train is only moving forward.
OpenAI, Microsoft, Google, Amazon and Nvidia are deploying a large infrastructure of chips, servers and data centers, all based on projections of future growth.
The numbers are hard to quantify: we are talking about trillions of dollars in capital spending, which is growing at an excessive rate. However, in parallel to record investments in AI infrastructure, the technology sector has seen multiple rounds of layoffs.
A couple of weeks ago, Amazon announced the departure of nearly 14,000 people, and it is not an isolated case. Other technology players, such as Microsoft and Meta, also cut their staff by the thousands recently.
What is most disturbing is that many of these eliminated jobs correspond precisely to areas where artificial intelligence is already beginning to replace human tasks: customer service, support, marketing, writing, and even software development.
The efficiency promised by AI is being achieved, yes, but at what cost? Work is being automated faster than we thought, and the growth generated by this new economy does not necessarily translate into welfare.
In this context, a phrase said almost in passing set off alarm bells. Sarah Friar, CFO of OpenAI, recently mentioned the idea of a backstop: a state guarantee to support the expansion of artificial intelligence infrastructure.
It was this same concept that sustained the financial system during the 2008 crisis, when governments intervened to prevent the collapse of the banks.
Investment in AI has become self-referential: companies invest more to sustain the expectation that AI will generate future revenues, while actual returns are still modest, and energy and financial costs continue to rise.
In this scenario, the technology giants become systemic entities, closely interconnected, replicating the structure of the pre-crisis banking sector.
History has taught us that when an industry depends more on promises than results, the downfall is not a question of if, but when.
Artificial intelligence is one of the most powerful innovations of our time, and, personally, I am happy to be part of this cycle. For the same reason, we must make sure we don't end up needing another ransomware.
