AI bubble: Why Technology Giants Can Be Playing Billions in Trash

Yes, the AI bubble exists and is more inflated than many would like to admit. A recent article by The Atlantic, based on Goldman Sachs data and Model Evaluation & Threat Research experiments, reveals that software companies are bursting budgets with AI tools on an absurd scale, while real productivity gains are still doubtful. The question remains, how long will this card castle last?

AI bubble: What did Goldman Sachs find out about the expenses?

Goldman Sachs investment bank published an alarming analysis: software companies are spending more magnitude orders on artificial intelligence tools than initially predicted. The study shows that instead of a gradual increase, budgets exploded in multiples, often without a clear return plan on investment. The expectation was that the AI would bring immediate efficiency, but the numbers indicate that the account is coming out much more expensive than the benefit generated.

AI bubble: The experiment that put developers to the test

If the figures are already worrying, the practical results are even more challenging. Model Evaluation & Threat Research conducted an experiment in which software developers were divided into two groups: one with access to AI tools for coding and one without. Surprisingly, the group that used AI did not perform superiorly — in some metrics, even stayed behind. This raises serious questions about the real productivity of generative AI solutions, especially in complex tasks that require logical reasoning and debugging.

AI bubble: Signs that euphoria can end

The market is already starting to react. Large companies such as Google, Microsoft, and Meta invested billions in AI infrastructure, but as we show in Billionaire Costs in AI: Google, Microsoft, and Meta Invest, but Only One Was Punished by the Market, only one of them was penalized by investors. This suggests that market patience may be running out. If spending continues to grow without proportional returns — and the evidence from Goldman Sachs and the experiment indicates that this is happening — correction can be brutal.

AI bubble: Adjustment of expectations or imminent collapse?

Some argue that we are only at the beginning of a revolution and that current investments are needed to build the foundation of the future. But the article in The Atlantic highlights that if promises do not materialize in tangible revenue in a short term, the bubble may become unsustainable. Smaller companies, who took debt to finance the adoption of AI, may be the first to break.

AI bubble: Implications for executives and developers

For decision makers, the message is clear: it is not enough to adopt AI by fashionism. It is necessary to measure results, question suppliers and have objective feedback metrics. On the other hand, Brazilian developers, who often see AI as a threat or a savior, should view this data with healthy skepticism. The tool does not replace human judgment — at least not yet.

Frequently Asked Questions about the AI Bubble

What's the AI bubble?

It is a cycle of inflated investments and unrealistic expectations around artificial intelligence, with companies spending much more than the financial return justifies.

Will the investments in AI continue to rise?

Probably not at the same rate. If the practical results remain disappointing, funds can dry and the bubble can begin to deflate.

Should I stop using AI tools in my work?

No, but use it with caution. Prioritise tools that solve specific and measurable problems rather than adopting generic solutions that promise miracles.