Meta delivered a powerful revenue performance in its third-quarter earnings report Wednesday, but investors had more than one number to digest. The Facebook and Instagram parent also unveiled a planned stock restructuring, raised the lower end of its data center spending forecast and offered a fresh glimpse into its increasingly expensive artificial intelligence ambitions.

Meta reported earnings after the closing bell, posting earnings per share of $3.83 and revenue of $64.60 billion, topping LSEG’s revenue estimate of $63.37 billion. Revenue climbed 26% from the same period a year earlier. It was not immediately clear whether the reported earnings per share were directly comparable with the $5.75 analysts had expected.

The results arrive as the company races to turn its AI investments into a competitive advantage — while Wall Street weighs whether the enormous spending will ultimately pay off.

Meta Stock Faces a New Test as AI Ambitions Expand

CEO Mark Zuckerberg framed the quarter as a major step toward the company’s vision of delivering personal superintelligence to everyone.

“We hit major milestones this quarter toward delivering personal superintelligence for everyone - including launching Muse and Muse Wonder, our first frontier model,” Zuckerberg said in the earnings release.

The early reception for Muse, Meta’s personal AI assistant introduced in September, helped propel the company’s shares through their strongest month on Wall Street since 2022, according to CNBC.

Meta is not the first technology company to offer digital assistants capable of organizing email or searching the internet for product deals. But its enormous reach through Facebook, Instagram and its other applications gives it a powerful distribution channel for bringing AI agents to mainstream users.

At the same time, Meta and its AI chief, Alexandr Wang, have continued releasing foundation models and updates at a rapid pace. The company released Muse Wonder last Monday, further expanding its AI lineup.

Meta Plans a New Class of Shares in 2027

Alongside its earnings, Meta announced plans to establish a new class of non-voting Class C common stock, scheduled to take effect in January 2027.

Existing shareholders will receive the new shares through a dividend, creating an arrangement that effectively resembles a two-for-one stock split. The additional class of stock is expected to begin trading in the new year.

Meta said the structure is intended to help finance its growth and preserve its position amid intense domestic and international competition. The company also said the plan would maintain the founder-led model it credits with driving its success.

The announcement included a Founder Alignment Agreement with Zuckerberg. The agreement covers his sales of Class C shares received through the dividend, his continued service at Meta and conditions governing potential future Class C stock dividends.

The new structure adds another dimension to the company’s plans as it seeks to fund an expensive AI buildout while maintaining its existing leadership arrangement.

Data Center Spending Remains a Wall Street Flashpoint

Meta narrowed the range of its full-year capital expenditure forecast for data centers and related infrastructure, now expecting spending between $135 billion and $145 billion. Previously, the company had projected a range of $130 billion to $145 billion.

Analysts had estimated full-year capital expenditures at approximately $138.27 billion, putting their forecast within Meta’s updated range.

The spending matters because Meta needs enormous computing capacity to develop and operate AI agents and foundation models. Yet the scale of the investment has also raised questions about how quickly the company can turn those costs into meaningful returns.

Investors have questioned why Meta needs to spend so heavily on data centers when it does not operate a cloud computing business on the scale of rivals such as Google and Amazon. Those companies can sell computing services directly to outside customers, while Meta’s core business remains advertising across its social platforms.

Still, recent progress in AI may give investors more reason to tolerate the spending, at least for now.

Fourth-Quarter Revenue Outlook Tops Expectations

Meta projected fourth-quarter revenue of $73 billion to $76 billion. The midpoint of that range exceeds analysts’ estimate of $73.73 billion, offering another positive signal for the company’s near-term business outlook.

One key engagement metric fell slightly short of expectations. Meta reported 3.64 billion daily active people, or DAP, across its family of applications, including Facebook and Instagram, compared with the 3.65 billion analysts had expected, according to StreetAccount.

The figure measures the broad reach of Meta’s apps, which remains central to the advertising engine funding its AI ambitions.

Meta Bets AI Can Strengthen Advertising and Unlock New Revenue

Over the past year, Zuckerberg has spent billions revamping Meta’s AI division and appointed Wang to lead the operation roughly 14 months ago. That strategy has begun to improve investor sentiment, according to a Tuesday note from Baird Equity Research analysts, who pointed to optimism surrounding Muse AI agents and related technologies.

However, the analysts cautioned that AI-driven improvements to Meta’s advertising platform are likely to have a greater near-term effect on quarterly results than direct sales of AI products. Better AI systems could help advertisers target audiences more effectively, strengthening the business that currently generates most of Meta’s revenue.

Longer term, the company is looking to create additional income streams.

Paid Muse Plans Target Power Users and Developers

Meta is offering monthly subscriptions priced at $20 and $100 for users who want more from its Muse AI agent. The basic tool is currently free, but access to the latest model, Muse Wonder, requires a subscription.

Developers can also pay $50 or $200 per month to use Meta’s most powerful Muse Wonder Full AI model through Muse Code, a coding agent introduced in early August.

These subscription plans represent an attempt to expand beyond online advertising, a business that remains Meta’s financial backbone. Whether AI products can become a substantial source of revenue, however, remains an open question.

Last week, Meta partnered with longtime rival Apple to make its new open-weight AI model compatible with the iPhone maker’s high-end Mac computers. Unlike Meta’s proprietary models, developers can download the weights of Muse Spark Open at no cost.

The strategy gives developers an opportunity to experiment with the technology before potentially upgrading to more powerful, paid systems.

“These aren’t the most capable models that we’ve developed or released,” Wang told CNBC. “But we think that there will be a lot of companies and developers who sort of use this model and tinker with it at small scale and then seek to ramp up and scale up their usage.”

That approach could help Meta attract developers first and monetize heavier use later. But turning experimentation into recurring revenue will take time, leaving investors to weigh immediate infrastructure costs against potential future gains.

Legal Challenges Add Pressure Beyond AI Spending

Meta’s latest quarter was not defined by AI developments alone. The company is also facing significant legal exposure tied to allegations about the effects of its social media platforms on children’s mental health.

In August, Meta agreed to pay nearly $17 billion to settle a major lawsuit brought by a group of state attorneys general, including California Attorney General Rob Bonta. The case focused on allegations that the company misled the public about the extent of mental health harms linked to certain design features on Facebook and Instagram, including recommendation algorithms.

The financial implications could have affected how investors interpreted Meta’s quarterly results. In a Monday research note, Wolfe Research analysts said a related $10 billion legal charge had not yet been accounted for in Wall Street estimates and could cause the company to miss earnings-per-share expectations.

Meta continues to face numerous personal injury claims, as well as a separate federal case involving a group of school districts from across the country. Those proceedings also concern allegations that platforms such as Instagram have harmed children’s well-being.

The legal disputes add uncertainty at a time when Meta is pouring enormous sums into AI infrastructure and attempting to build new revenue sources. Its latest results show strong revenue growth and an optimistic near-term outlook, but investors must also consider the costs of its AI ambitions, the execution risks surrounding new products and the financial consequences of ongoing litigation.

For Meta, the challenge is no longer simply to build powerful AI systems. It is to demonstrate that those investments can produce sustainable growth while the company navigates mounting expenses, fierce competition and legal scrutiny.