Artificial return?

It's been a relatively uneasy start to the year. To the more contractionary bias of monetary policy in the U.S. are added the tariffs announced by the Trump administration on its main trading partners.

Even some of the most “obvious” theses of recent years are being challenged. If we knew one thing, it was that in AI the absolute leadership was in the U.S., led by the big tech giants. In addition, competition was limited by the extremely high demand for the computational resources required for these tasks.

With the arrival of DeepSeek—the project of a Chinese hedge fund that, with a small fraction of the resources, achieved performance equivalent to the most advanced models—the picture changed drastically. First, this model's code is open, that is, anyone can run it and build on it. In addition, the processing requirements are minuscule in comparative terms. Thus, the entire AI chain (from energy companies to the owners of the algorithms or chips like Nvidia) saw significant corrections in company values.

We don't know who will win the AI race, or whether it will keep driving the market. Sector diversification allowed us to capture the returns of communications and IT companies last year but also, in 2018 and 2019, to benefit from the rise of energy companies. Now more than ever, one must invest with discipline.

Original publication in La Segunda February 4, 2025.