Mark Zuckerberg, CEO of Meta Platforms, Inc., during the MetaConnect event on Wednesday, September 17, 2025 in Menlo Park, California, USA.
David Paul Morris | Bloomberg | Getty Images
Meta CEO Mark Zuckerberg plans to increase his company’s spending on artificial intelligence in 2026. Wall Street seems fine with that strategy.
In Wednesday’s fourth-quarter earnings report, Meta posted higher sales and bottom line results, while also revealing that AI-related capital spending this year will range from $115 billion to $135 billion. That’s nearly double the amount Meta spent on capital expenditures when it revamped its AI division last year.
Investors had previously expressed concerns about Meta’s sharp increase in AI spending, but were reassured by the company’s latest results, which showed 24% year-over-year revenue growth, driven by online advertising. Meta shares, which lagged the market last year, soared as much as 10% in after-hours trading.
“Our future plans include continuing to make significant investments in our infrastructure to train our leading models and deliver personal superintelligence to billions of people and businesses around the world,” Zuckerberg told analysts on an earnings call.
Zuckerberg was referring to Meta’s ambitious construction of a data center aimed at anchoring both current and future AI projects.
Metafinance chief Susan Lee told analysts that the company remains “capacity constrained” and needs more computing power to further improve its core advertising business while also providing its AI team with the resources it needs to create more advanced models and products.
“Our teams have done a great job of hardening our infrastructure by 2025, but demand for computing resources across the company is growing faster than supply,” Li said.
Zuckerberg said 2026 will be a big year for AI, and Meta’s investment is aimed at supporting his mission to “build a personal superintelligence.”
The big question remains whether Meta will be able to generate revenue from its new AI products, but Zuckerberg hasn’t given a clear answer.
“So we’re going to be introducing new products by the end of the year,” Zuckerberg said on a conference call. “I think the important thing is that we’re not just launching one thing, we’re building many things.”
Perhaps Zuckerberg’s biggest move last year was his $14.3 billion investment in Scale AI, which brought founder and CEO Alexandr Wang and some of his top engineers and researchers into Meta. CNBC reports that Wang currently heads Meta’s TBD AI division and is testing a new frontier model codenamed “Avocado” that is intended to be a successor to the company’s Llama family of models.
“We expect our first model to be good, but more importantly, it will demonstrate the rapid trajectory we are on,” Zuckerberg said Wednesday. “And we look forward to steadily pushing the frontier throughout the year as we continue to release new models.”
Asked on the conference call why Meta needs to develop its own strong AI underlying model, Zuckerberg said it’s important because Meta is a “deep technology company.”
He said Meta can’t risk being “constrained by what other companies in the ecosystem are building or what they allow us to build,” adding that by controlling the model, it can help “shape the future of these products.”
Meanwhile, online advertising still makes up the vast majority of Meta’s revenue. As long as the business remains dominant in the mobile space, exceeds expectations, and continues to shell out billions of dollars in cash every quarter, Zuckerberg will likely have plenty of leeway to pursue his AI ambitions.
WATCH: Meta and Google go to court over child safety – what’s at stake here?
