Meta Shifts to Closed AI Models, Raising Investor Skepticism

Meta Platforms Inc. (NASDAQ:META) is doubling down on its artificial intelligence ambitions, with CEO Mark Zuckerberg pushing for a strategic shift toward developing proprietary AI models that can generate revenue. The company is working on a new frontier AI model codenamed "Avocado," which is expected to debut in early 2026 as a closed model that customers will pay to access. This marks a significant departure from Meta's longtime open-source strategy, aligning it with rivals like Google and OpenAI according to Bloomberg.

Zuckerberg has personally recruited top AI talent, including Scale AI founder Alexandr Wang, who now leads the TBD Lab working on Avocado. The decision follows underwhelming reception of Meta's previous open-source model, Llama 4, and growing pressure from investors to justify AI spending. Morgan Stanley recently lowered Meta's price target to $750 from $820, citing concerns about revenue durability and the company's operating expenses.

The new model is being trained using third-party AI systems, including Alibaba's Qwen and Google's Gemma, highlighting a strategic pivot that has raised internal and external eyebrows. Some employees have expressed concerns about micromanagement from Zuckerberg, while former AI research leader Yann LeCun left the company amid disagreements over resource allocation. The move also signals a shift in tone for Zuckerberg, who previously criticized Chinese AI models for potential state censorship.

A Strategic Shift Toward Proprietary AI

Meta's pivot to closed AI models reflects a broader industry trend and a response to investor demands. Unlike open-source models, which allow developers to freely access and build upon the code, closed models are tightly controlled and monetized. This approach has long been favored by competitors like Google and OpenAI, who have achieved significant commercial success through their proprietary AI systems.

Zuckerberg's decision follows a series of strategic changes at Meta. The company has already cut resources for its metaverse initiatives and redirected investments toward AI. These efforts include the development of AI-powered wearables and partnerships with infrastructure firms to build large-scale data centers. The shift is part of a broader $600 billion investment plan over the next three years, much of which will be dedicated to AI research and development according to Bloomberg.

The move toward closed models also has implications for Meta's long-term AI goals. The company aims to achieve what it calls "superintelligence," or AI systems capable of surpassing human-level reasoning and performance. However, this term has raised concerns among policymakers and the public, particularly in Europe, where regulators have expressed skepticism about the risks of uncontrolled AI development.

Market Reactions and Investor Sentiment

Meta's pivot to closed AI models has been met with mixed reactions from investors. While some analysts believe the company is on the right track, others remain cautious about its ability to generate returns from these high-cost investments. Morgan Stanley maintains an Overweight rating on Meta but has lowered its price target due to concerns about revenue growth and operating expenses.

The stock has also been under pressure amid broader market skepticism. In late 2025, Meta's shares traded at a price-to-earnings multiple of 28.65, below its long-term average. The company's PEG ratio of 4.07 suggests that the stock is trading at a high valuation relative to near-term earnings growth. Meanwhile, Alphabet (NASDAQ:GOOGL) has seen its next-twelve-months multiple premium versus Meta reach 33%, the highest in three years according to Investing.com.

Investors are also watching how Meta's AI strategy plays out in the context of the broader market. The AI Voice Generator Market, for example, is projected to grow at a 30.7% CAGR between 2025 and 2031, reaching $20.71 billion by 2031. This suggests significant monetization potential in the AI space, but it also means that competition will be fierce.

Risks to the Outlook

Meta faces several challenges as it transitions to a closed AI model strategy. One of the biggest is the need to balance investor expectations with the long-term nature of AI development. While closed models can generate revenue, they often require significant upfront investment and time to reach commercial viability. This has led to concerns that Meta's AI spending may not yield immediate returns, even as the company continues to ramp up its investments.

The company also has to navigate regulatory scrutiny, particularly in Europe, where AI regulations are becoming more stringent. The European Union has already taken steps to limit the use of AI in certain sectors, and there are concerns that Meta's new AI initiatives could face similar restrictions. In the U.S., regulators are also paying close attention to how AI models are developed and deployed, particularly in areas like content moderation and data privacy according to Investing.com.

Another risk lies in the internal dynamics of Meta's AI team. The high-stakes environment has led to a culture of long work hours, layoffs, and restructurings. While the company has managed to attract top talent, it is unclear whether it can retain these employees in the long run. Some former employees have already left to pursue other opportunities, and there are concerns that the pressure to deliver results could lead to burnout or further departures.

Despite these challenges, Meta remains committed to its AI strategy. Zuckerberg has framed the shift as a necessary step to ensure the company's future in a world increasingly defined by artificial intelligence. The success of this strategy will depend on the performance of Avocado and the company's ability to execute on its long-term vision.