Meta CEO Mark Zuckerberg is doubling down on artificial intelligence, signaling plans to accelerate investment as the company aims to position itself as the leading frontier AI lab in an increasingly heated race against OpenAI, Google, Microsoft, and others.
During its third-quarter earnings report on October 29, Meta raised its 2025 capital expenditure outlook to $70–72 billion, compared with the earlier estimate of $66–72 billion. The company also noted that spending in 2026 will substantially exceed 2025 levels.
Meta now expects total expenses to reach $116–118 billion, slightly higher than its previous forecast, representing a 22–24% year-over-year increase. A major contributor is soaring compensation tied to its aggressive AI recruitment drive, which includes billion-dollar pay packages intended to attract leading researchers from competing firms such as OpenAI, Google, Anthropic, and Apple.
Growing Concerns Among Investors
These massive investments have triggered renewed anxiety among investors about whether Meta will see meaningful returns on its escalating AI infrastructure and talent outlays. The concern was reflected in the stock, which dropped as much as 9% on October 29.
On the earnings call, however, Zuckerberg strongly defended the accelerated AI spending, stating that Meta must prepare for the possibility that “personal superintelligence for everyone” could arrive sooner than expected. Superintelligence refers to AI systems that outperform human capabilities across all domains.
“We think it’s important to front-load capacity so we’re ready for the most optimistic timelines,” he said. “If superintelligence emerges earlier, we’ll be in a prime position to benefit from a once-in-a-generation shift.”
Even if progress is slower, Zuckerberg argued, the extra compute could still be used to boost Meta’s core business, which “can profitably absorb far more compute than we’ve been able to supply.” In a worst-case scenario, Meta would simply pause new infrastructure expansion until usage caught up, he added.
“We’re already seeing strong returns in the core business,” he said. “That’s why we’re confident we should be investing heavily—and the last thing we want is to under-invest.”
Overhaul of Meta’s AI Strategy
The company has been overhauling its AI roadmap after several setbacks, including underwhelming results from its Llama 4 model, delays in rolling out more advanced systems, and the exit of key researchers.
In June, Meta spent $14.3 billion for a 49% stake in data-labeling firm Scale AI and appointed its CEO, Alexandr Wang, as Meta’s chief AI officer. It also reorganized its AI efforts under a new umbrella group, Meta Superintelligence Labs, unifying its foundation model teams, FAIR researchers, product groups, and a new next-generation model development unit.
Earlier this month, Meta hired Andrew Tulloch, co-founder of Thinking Machines Lab, a startup established by former OpenAI CTO Mira Murati.
“Meta Superintelligence Labs is off to a strong start. We’ve already assembled the highest density of top-tier talent in the industry,” Zuckerberg reported. “We’re fully focused on building our next wave of models and products.”
Despite its expansion, Meta cut approximately 600 employees from its AI organization last week to improve operational agility—though none of the recent high-profile AI hires were affected.
Meta’s Q3 Results
For the third quarter, Meta posted $51.2 billion in revenue, a 26% increase from a year earlier. Net income, however, dropped 83% to $2.7 billion, largely due to a one-time $15.93 billion tax charge tied to the implementation of former U.S. President Donald Trump’s One Big Beautiful Bill Act.
Across Facebook, Instagram, WhatsApp, and Threads, Meta saw 3.54 billion daily active users in October 2025—an 8% year-over-year rise.
