Twelve months after Meta Platforms Inc. committed a staggering $14.3 billion to secure the services of Alexandr Wang and a cohort of elite engineers from Scale AI, the social media giant finds itself in a precarious position. While the investment has successfully re-established Meta as a serious contender in the generative artificial intelligence landscape, the company remains significantly behind industry frontrunners such as OpenAI, Anthropic, and Google. The transition from an open-source advocate to a proprietary model developer has sparked internal friction, alienated long-time developer allies, and left Wall Street questioning the long-term monetization strategy of Mark Zuckerberg’s latest multi-billion-dollar bet.
The centerpiece of Wang’s tenure thus far is the Muse Spark AI model, debuted in April. This release marked a fundamental shift in Meta’s philosophy, moving away from the "open weight" approach that defined its earlier Llama family of models toward a proprietary, closed-system foundation. To house this new direction, Zuckerberg established Meta Superintelligence Labs (MSL), a high-stakes division designed to inject "sizzle" into Meta’s technological portfolio and rival the research output of Silicon Valley’s most advanced AI laboratories. However, as the initial excitement of the Muse Spark launch fades, the focus has shifted from technical milestones to commercial viability.
The Financial Pressure and Wall Street’s Skepticism
Despite Meta’s aggressive pivot toward AI, the company’s stock performance tells a story of investor apprehension. Over the past year, Meta’s shares have declined by 18%, positioning it as one of the worst performers among the megacap technology group. This slump occurred even as the company reported a robust 33% revenue growth in the first quarter—its fastest rate of expansion since 2021. The disconnect between top-line growth and stock valuation suggests that investors are no longer satisfied with AI merely "bolstering" the core advertising business; they are demanding a new, AI-native revenue stream.
Ralph Schackart, an analyst at William Blair, notes that the market is looking for "proof points of both adoption and commercialization." While AI has undeniably improved the efficiency of Meta’s advertising algorithms—helping to protect a $200 billion annual business—investors are eager to see a standalone, profitable AI product. "Investors are looking for Meta to monetize a new AI-first product, beyond the substantial positive impact AI is having on enhancing the advertising models," Schackart said.
The urgency for monetization is underscored by Meta’s history of expensive pivots. The company’s Reality Labs division, which houses its metaverse and virtual reality ambitions, has recorded total losses exceeding $80 billion since late 2020. This massive expenditure has depleted much of the "credibility capital" Zuckerberg holds with shareholders. Consequently, the $14.3 billion spent on the Scale AI deal and the ongoing high capital expenditure required for AI infrastructure are being scrutinized with unprecedented intensity.
A Strategic Rebuild: From Llama to Muse Spark
Meta’s journey to its current state was born out of what many industry experts now describe as a strategic miscalculation. Initially, Meta sought to dominate the AI ecosystem by releasing the Llama family of models under an open-source (or open-weight) license. This allowed developers to use and modify the models for free, a move intended to commoditize the underlying technology of rivals who were charging for access.
However, the strategy faltered in April 2025 when the release of Llama 4 failed to capture the imagination of the developer community. The model was perceived as trailing the capabilities of OpenAI’s GPT series and Google’s Gemini. Recognizing that Meta was losing the "frontier model" race, Zuckerberg made the radical decision to bring in external leadership. The June 2025 deal to acquire a 50% stake in Scale AI and hire its founder, Alexandr Wang, was a clear signal that Meta was abandoning its previous trajectory in favor of a "strategic rebuild."
Under Wang’s leadership, the development of Muse Spark was prioritized as an internal-first solution. Unlike Llama, which was designed for the broad developer community, Muse Spark was engineered to integrate seamlessly with Meta’s existing ecosystem, including Facebook, Instagram, and the increasingly popular Ray-Ban Meta smart glasses. Thomas Randall, an analyst at Info-Tech Research Group, argues that this was a necessary survival tactic. "Meta needs to have a consistent, reliable proprietary model that they themselves own," Randall said, adding that the company would be "lost" had it not made the aggressive hires of Wang and other high-profile figures like former GitHub CEO Nat Friedman.
The Developer Rift and the "Walled Garden" Problem
While the pivot to proprietary models may serve Meta’s internal product goals, it has severely damaged its relationship with the broader AI research and developer community. For years, Meta was viewed as the "champion" of open AI development. By shifting toward a "walled garden" approach with Muse Spark, many developers feel the company has abandoned the very community that helped build its AI reputation.
"I think the AI community largely ignores Meta at this point," said Rob May, CEO of the startup Neurometric. May, who specializes in token engineering, noted that the lack of accessibility to Muse Spark has turned what should have been a landmark release into a "yawn" for many in the industry. The communication channels that were once open between Meta and third-party developers have reportedly gone silent. "I used to be in regular touch with Meta for Llama-related issues, but now I can’t get them to return messages," May added.
Krish Subramanian, CEO of KOI AI and former product head at IBM Consulting, warns that this loss of trust could have long-term consequences. He compares Meta’s current situation to Microsoft’s early struggles with the open-source community during the inception of Azure. "To just focus on a walled-garden kind of an ecosystem and ad revenue as the main source of income, they probably will never become the big player," Subramanian said.
Internal Turmoil and Leadership Tensions
The external challenges are mirrored by internal pressures within Meta’s Menlo Park headquarters. The company has undergone significant restructuring over the past year, including the termination of approximately 8,000 employees in May 2026. These layoffs have affected various departments, including critical teams focused on trust and safety. Insiders suggest that the reduction in safety personnel has raised concerns about the ethical guardrails surrounding future AI development, although Meta has officially stated that model safety remains a top priority for Wang.
Furthermore, reports have surfaced regarding tension at the highest levels of Meta’s AI organization. While Muse Spark was considered a technical success internally, the pressure on Wang and Nat Friedman to deliver "meaningful revenue growth" is immense. Standing in the wings is Andrew "Boz" Bosworth, Meta’s longtime Chief Technology Officer and a close confidant of Zuckerberg. Sources familiar with the matter suggest that if the newcomers fail to turn AI into a financial powerhouse quickly, Zuckerberg may consolidate AI leadership under Bosworth and other company veterans.
Wang has attempted to downplay reports of internal conflict, describing Muse Spark as merely an "appetizer" for more powerful, larger models currently in development. However, the cadence of these releases is a point of contention. Competitors like OpenAI and Google maintain a relentless cycle of updates; if Meta cannot match this frequency, it risks falling back into irrelevance.
The Path Forward: Efficiency as a Differentiator
Despite the hurdles, some experts believe Meta can still carve out a unique and profitable niche. Andrew Moore, CEO of Lovelace and former Google Cloud AI chief, suggests that Meta’s focus on computational efficiency could be its "secret weapon." As the cost of training and running massive foundation models continues to skyrocket, developers and enterprises are increasingly looking for models that offer high performance at a lower "compute" cost.
"If they do proprietary, computationally efficient models, that will be so different from what’s happening in this death match between the big guys," Moore said. If Wang can prove that Muse Spark and its successors are faster, cheaper, and more reliable than the competition, Meta may find the "lane" it has been searching for.
Ultimately, the responsibility for Meta’s AI future rests with Mark Zuckerberg. Having spent tens of billions of dollars on a vision that has yet to yield a standalone profit, the CEO is running out of room for error. The success or failure of Alexandr Wang’s tenure at Meta Superintelligence Labs will likely be the deciding factor in whether Meta becomes a dominant force in the AI era or remains a company perpetually "protecting the machine" of its legacy advertising business. As Howard Yu, a professor at the International Institute for Management Development, aptly put it: "He’s running out of the space for his credibility to last. The virtual reality foray may have burned up a lot of his goodwill in front of investors." The next twelve months will determine if Muse Spark was the beginning of a renaissance or an expensive footnote in Meta’s history.
