Meta creates non-voting shares to fund massive AI infrastructure push

Meta Platforms is preparing to issue a new class of non-voting stock to finance its aggressive expansion into artificial intelligence, a move designed to raise significant capital while preserving the voting influence of its existing shareholders. The company confirmed the plan on Wednesday during its third-quarter earnings call, validating reports that the tech giant is restructuring its equity strategy to support its heavy investment in AI technologies.

Susan Li, the company’s chief financial officer, stated that Meta intends to create a new class of non-voting capital stock. This new instrument will be distributed to all current shareholders as a stock dividend. Li explained that the primary goal is to establish a publicly traded, non-voting share that can be used for capital raising, employee compensation, and acquisitions. This approach allows the company to grow its funding tools while maintaining its current founder-led structure and protecting the voting power held by Class A and Class B shareholders.

The financial results for the past quarter provided the backdrop for this strategic shift. Meta reported revenue of $64 billion, marking a 26 percent increase from the same period a year earlier. These figures exceeded expectations set by Wall Street analysts. However, the company’s costs and expenses also saw a substantial rise, climbing 68 percent year-over-year. This sharp increase in spending reflects the intensive resources being directed toward new technological initiatives.

The Reality Labs division, which manages Meta’s augmented reality glasses, virtual reality headsets, and associated software, continues to operate at a loss. While the deficit has narrowed compared to previous years, the company projects that losses for this division will remain between $18 billion and $19 billion. This figure represents a reduction from the losses recorded in 2025, indicating a slight improvement in the division’s financial performance.

Over the past year, Meta has accelerated its rollout of AI products. The company released several models, including Muse Spark and Muse Glimmer earlier in the year. In recent weeks, it introduced Muse Wonder and Muse Spark Open. The latter is an open-weight model, meaning its underlying parameters are publicly available for external developers to use and modify. Meta also launched Muse Code, an agent designed to assist software developers with coding tasks.

During the Meta Connect event earlier this year, the company unveiled a new line of avatars for its AI assistant, Meta Muse. These characters are described as cute and cuddly, a design choice intended to make the AI services more relatable and shareable for users. The company believes that this approach will help drive broader adoption of its AI ecosystem.

Mark Zuckerberg, the company’s CEO, noted that these AI products are gaining traction. He reported that more than 10 million people in the United States and Canada have set up Muse personal agents in just a month and a half. Zuckerberg stated that growth remains strong and that the application has been number one in the App Store on nearly every day. He added that the company is entering a more operational phase focused on scaling the product to eventually serve billions of users.

Despite the positive momentum, Li acknowledged that regulatory and legal challenges persist. She noted that the company continues to monitor active matters that could impact its financial results. Specifically, Li pointed to ongoing scrutiny regarding youth-related issues in several markets. The company has multiple trials related to these issues scheduled in the coming months within the United States.