I believe we are building a major economic bubble around artificial intelligence. Large technology companies are investing hundreds of billions in data centres, chips and computing capacity, convinced that future demand will justify these investments. However, one important question seems to be missing from their forecasts: the fact that artificial intelligence succeeds does not mean that every business will need enormous amounts of computing power. Most companies will need tools capable of organising documents, handling calls, automating administrative tasks or improving their marketing. They will not necessarily need the most powerful models or massive infrastructure. We are confusing the potential of a technology with the size of the market that can realistically sustain it.

There is another, more fundamental problem. Artificial intelligence can increase productivity, reduce costs and allow companies to produce more, but that does not mean families will have more money to spend. If production capacity grows much faster than people’s purchasing power, demand will not be sufficient to sustain all these investments. Capital is being committed today based on expectations of future revenues that have yet to be proven. Something similar happened with the internet bubble and the housing market before 2008. Technology can transform the world while simultaneously destroying those who invested too much money believing that its growth would be unlimited. The problem is not artificial intelligence. The problem is how much money we are putting at risk by assuming that technological success will automatically guarantee economic success.