“We fear we are not fast enough to adapt”

Few companies are putting as much money on the table as Meta in the AI ​​race: it failed miserably with the multiverse, so it changed its goal and directed all its efforts to artificial intelligence. In fact, to win it invested more than 14 billion last year alone of dollars in one operation. However, it is not where I expected. This situation is not to the liking of Mark Zuckerberg, who has been self-critical regarding the pace of advances in agentic AI.

What’s happening. The CEO of Meta acknowledged to his staff in an internal meeting last Thursday that the profound restructuring that the company has carried out is not going as well as they thought. The reason? AI agents have progressed as quickly as expected, according to a recording seen by Reuters.

Zuckerberg acknowledged that the company’s reorganization had not been as “clean” as he had hoped and that executives misjudged the timing of the changes, intended to fund investments in AI infrastructure and capitalize on efficiency gains from AI-assisted work. This restructuring plan began to take shape at the beginning of the year and at that time, Meta’s main managers had a concern: “Not acting quickly enough to adapt” to the advantages of agentic AI.

Why is it important. Because Meta is one of the most aggressive companies when it comes to betting on AI agents as a driver of corporate restructuring, with large layoffs and internal reassignments justified precisely by that transition. That its own CEO publicly admits that results are not coming clearly calls into question the industry’s widespread narrative about the quality and speed of the “agent revolution.” One fact: at this moment big tech companies are spending more than 700 billion dollars.

Context. The Meta thing is being a drain: at the beginning of the year it fired 10% of its staff: about 8,000 people said goodbye to the company. Likewise, also relocated 7,000 workers to AI teams. We knew what the objective of this movement was (not very “clean”, in their own words) and now we have discovered its motivation: the fear of managers of being left behind by the competition. The reason for his optimism at the time was an external tool with enormous potential: Anthropic’s Claude Code.

In detail. Of those more than $700 billion that big tech companies are investing in AI, Meta plans to invest up to $145 billion in AI infrastructure this year alone. And Zuckerberg predicts that the reward is just around the corner: according to ReutersMeta expects to realize significant benefits from these investments within three to six months. While this arrives, the human layer is dealing with friction: in the upper echelons, with frustration and uncertainty and a rarefied atmosphere in the workforce (some They say they work “in the gulag”)

Yes, but. First of all, this is a leak of a recording obtained by Reuters and Meta has not yet made any statements. On the other hand, Mark Zuckerberg recognizes the delay, but remains optimistic about the account it brings him: he maintains that it is a matter of time and continues to defend stratospheric investments. That is to say, despite this attempt at self-criticism towards this way of releasing ballast, there is no going back in spending or strategy.

In Xataka | Meta spent at least $14 billion to win the AI ​​race. It’s been a year and it’s still exactly where it was.

In Xataka | Meta moves to the other side of the AI ​​counter. You are going from buying computing to selling it

Cover | Goal, Unsplash