Mark Zuckerberg Unfazed by Bubble Talk: Meta Projects Up to $135 Billion in Spending This Year, Doubling Down on AI Investments

ITHome reported on January 30 that, according to the BBC on January 29 (local time), Meta CEO Mark Zuckerberg is preparing to further ramp up the company’s AI investment this year—even as a growing number of industry leaders warn that the current AI boom could be turning into a bubble.

During an earnings call with financial analysts alongside the release of Meta’s 2025 results, the company said it expects 2026 spending to reach as high as $135 billion, with the bulk directed toward building out AI infrastructure.

That would represent a near doubling from last year’s $72 billion spent on AI initiatives and related infrastructure. Over the past three years, Meta has poured an estimated $140 billion into the AI push, signaling a sustained effort to secure a leading position as the technology reshapes the market.

Zuckerberg also forecast that 2026 could be a pivotal year in which AI materially changes how work gets done across organizations.

In the final quarter of 2025, Meta’s expenses grew faster than revenue, putting pressure on profit margins. Even so, investors responded positively, with Meta shares rising about 6.5% in after-hours trading following the announcement.

While underscoring the long-term value of AI investment, Zuckerberg also hinted that additional workforce reductions may remain on the table. He noted that projects that once required large teams may increasingly be handled by a single exceptional person—reflecting how productivity dynamics could shift as AI capabilities mature.

Meta has already cut hundreds of roles, largely within Reality Labs, the division responsible for metaverse initiatives, hardware products, and portions of the company’s AI work.

Zuckerberg added that Meta is expanding the use of AI tools across the company to help employees—such as software engineers—deliver more output. As these tools make people “significantly more efficient,” he said, the gap between those who can effectively use AI and those who cannot may widen sharply.

“The future shape of work inside organizations is hard to predict,” he suggested, emphasizing that the moment AI agents begin operating at scale could prove especially consequential.