Originally published in Svenska Dagbladet by Björn Jeffery, July 16, 2025

After investments of hundreds of billions, Meta has now found how to make money from AI. Don’t count on any superintelligence to get there — the answer lies considerably closer than that.

600 billion kronor. That is how much money Meta is expected to invest in chips, computing capacity and data centres in order to develop AI services this year. That figure likely does not include the more than 130 billion kronor they just invested in the AI company Scale AI.

To put it in perspective, the investment is a little under three times the entire market value of Ericsson. And this is, as noted, only for 2025.

Many are now asking what these gigantic bets will actually lead to, and whether there are better ways for Meta to allocate this money. Explanations have been somewhat thin on the ground. Until now.

It is about advertising. That may not sound particularly exciting given competitors like Google’s DeepMind, which mapped protein structures for the first time ever in 2021. But if you look at Meta’s revenue breakdown — despite its big investments in VR glasses and the like — the answer comes quite easily. 97 percent of Meta’s revenues in 2024 came from advertising. Improving them is therefore the quickest way to start earning a return on the large investments being made.

According to the Wall Street Journal, the idea is to move away from ready-made ads created by the advertiser, and instead hand that assignment over to AI. In practice, companies would no longer provide or create specific ads. Instead they list what they want to achieve and how much money they are willing to spend.

With that information, Meta’s AI system would create new ads presented to potential customers. But since the ads are generated at each impression, they can become very specific. Do you live somewhere sunny? A convertible is suggested. Are you in Sweden in winter — something with a heated steering wheel.

This development is a couple of steps away from what happens today. Including real-time data in ads is something advertising agencies have been able to do for a long time. What is radical about what Meta is proposing is that it would potentially bypass advertising and media agencies entirely. Their job includes advising on messaging, method and effectiveness measurement. If Meta gets what it wants, it can take over that entire assignment itself.

That companies like Volvo or Volkswagen would hand their advertising messages over to an AI system is, however, unlikely in the short term. The risks for brands of that calibre are still too high.

So it is likely the smaller advertisers who will be most attracted by this. That can go a long way. A 2022 study showed that small and medium-sized advertisers accounted for 61 percent of all ads on Meta’s advertising platform. If they can avoid the costs of various advisers, they can allocate their entire marketing budget to Meta instead.

Meta’s initiative points towards a considerably less grand and exciting future for AI development, at least in the near term. The debate has to a large extent been about what happens if — or when — we reach what is called AGI, artificial general intelligence.

This major step happens when AI systems are smarter and more capable than humans, and has been painted both as a doomsday scenario and as a salvation for the world. It is also precisely that polarisation which has made so many people anxious about where AI development is heading. It quickly becomes an existential question for humanity.

Meta’s advertising bet is rather the opposite of that. But it can nonetheless make a large impact when it comes to efficiencies. Already today, Meta uses AI as an integrated part of many services without it being noticeable. This can involve recommendations of content or moderation of unsuitable material. But gradually the human oversight decreases and the algorithmic increases. The efficiencies will reasonably lead to the number of jobs decreasing — or at least to new ones not being created.

Billions upon billions are being poured into AI development right now. Great things have been promised — cures for serious diseases and solutions to the climate crisis. But here and now, it seems we are getting a small improvement to things we already have, and something we might have preferred to do without: marginally better advertising.