Even Meta — the holding company that controls Facebook, Instagram, and WhatsApp — has run headlong into its own limits. During an internal meeting held on Thursday, July 2, founder Mark Zuckerberg shared a far from reassuring assessment with his team: the development of artificial intelligence agents is not proceeding at the pace the company had expected. It is a rare admission of difficulty from one of the most aggressive players in the global AI race.

Across the tech industry, an epochal transition toward AI-powered automated tools is underway. It is a complex transformation — one that does not happen overnight, even for the giants of Silicon Valley. According to Reuters, which reviewed a recording of the meeting, Zuckerberg acknowledged that over the past four months AI agent development has not accelerated the way the company had anticipated. That is a statement that carries weight, given that Meta’s entire corporate reorganization over recent months was justified precisely by that promise.

Meta’s $145 billion bet on AI infrastructure

Meta plans to spend between $125 billion and $145 billion ( €116–134B) on artificial intelligence infrastructure this year — a figure already revised upward from an initial estimate of $115–135 billion, due to rising costs and expanding data center capacity. The number sits within an even broader context: the four major Big Tech companies — Amazon, Alphabet, Meta, and Microsoft — have collectively committed between $650 billion and $725 billion in capital expenditure for 2026. It represents the largest single-year infrastructure investment ever recorded in the history of the tech sector.

8,000 layoffs, 7,000 reassignments — and an admission that it was mishandled

The pursuit of greater agility and automation pushed Meta to make drastic choices. In early 2026 the company laid off 8,000 employees — equivalent to 10% of its administrative workforce — and reassigned another 7,000 to divisions focused on artificial intelligence. These changes collectively affected nearly one-fifth of Meta’s entire headcount. During the meeting, however, Zuckerberg admitted that the layoffs were not managed as cleanly as they should have been, and that the timing of the reorganization had been miscalculated. The decision had grown out of a fear, widespread among company leadership, of failing to adapt quickly enough to the pace of change sweeping the tech industry.

The human cost of this transformation has been documented in several investigative reports, which describe Meta’s new AI unit as a grueling work environment for engineers, subjected to mounting pressure to deliver results on ever-shorter timelines.

Three to six months to prove it works

Despite the difficulties, Zuckerberg refused to change course and told his team to expect concrete signals of improvement within the next three to six months. What remains to be seen is whether Meta’s team — under intensifying pressure — will be able to keep the pace set by its founder before investor patience runs out.

Editor’s note

This article was originally published in Italian on money.it by Alessandro Nuzzo on July 04, 2026 as «Mark Zuckerberg è preoccupato per le sue aziende. «Temiamo di non essere abbastanza veloci nell’adattarci»». It has been translated and adapted for an international audience by the Money.it International desk.