Meta has spent $14 billion reshaping its artificial intelligence strategy and regained some competitive ground, but the company still trails OpenAI, Anthropic, and Google in both market perception and developer enthusiasm. The real challenge now is proving that all this investment can actually make money, not just improve ads.

Why Did Meta Abandon Its Open-Source AI Strategy?

For years, Meta pursued a different path than competitors by releasing Llama, an open-weight AI model that developers could freely modify and experiment with. The approach was initially praised as a challenge to the industry's closed ecosystems. However, when Llama 4 launched in April 2025, it failed to generate significant excitement among developers, prompting CEO Mark Zuckerberg to reassess Meta's entire AI direction.

Just two months later, Zuckerberg stunned Silicon Valley by announcing a $14.3 billion investment in Scale AI and bringing founder Alexandr Wang into the company. The move signaled that Meta needed a fresh start. This time, instead of building models for outside developers, Meta created Meta Superintelligence Labs (MSL), a division designed to inject urgency and credibility into the company's AI ambitions.

The shift away from open-source models reflects a broader strategic pivot. Rather than competing with rivals on developer relationships, Meta is now focusing on building proprietary foundation models that integrate deeply across its own ecosystem, powering experiences within Facebook, Instagram, Meta AI, and the company's growing portfolio of AI-enabled hardware, including Ray-Ban Meta smart glasses.

What Has Meta Actually Delivered So Far?

Wang's biggest achievement has been the launch of Muse Spark in April 2026, Meta's first major proprietary AI model under the new strategy. Unlike Llama, Muse Spark was not primarily built to win over outside developers. Instead, the model was designed to integrate deeply across Meta's ecosystem and reportedly received a positive reception inside the company.

Since Muse Spark's launch, Meta has introduced new AI-focused subscription offerings and business services, hoping to diversify a company that still derives approximately 98 percent of its revenue from advertising. However, the reception from the broader AI community has been lukewarm. Rob May, CEO of startup Neurometric, described the developer response to Muse Spark as a "yawn," largely because access remains limited.

"I think the AI community largely ignores Meta at this point," said Rob May, CEO of startup Neurometric.

Rob May, CEO, Neurometric

The contrast with Meta's earlier approach is stark. During the Llama era, the company actively cultivated relationships with outside developers. According to May, that engagement has largely disappeared. "I used to be in regular touch with Meta for Llama-related issues," he said. "I can't get them to return messages".

How Can Meta Differentiate Itself in a Crowded AI Market?

Rather than competing directly with rivals in building ever-larger models, some experts believe Meta could carve out a unique position by focusing on computational efficiency and affordability. Andrew Moore, CEO of enterprise startup Lovelace and former head of AI at Google Cloud, argues that developers care deeply about factors such as operating costs, latency, and efficiency.

"If they do proprietary, computationally efficient models, that will be so different from what's happening in this death match between the big guys," said Andrew Moore. "They might really benefit."

Andrew Moore, CEO, Lovelace

However, the biggest obstacle may not be technology; it may be trust. Several developers and AI entrepreneurs argue that Meta has lost credibility with the broader AI community after changing direction so dramatically. Krish Subramanian, CEO of consulting firm KOI AI and former IBM Consulting product executive, warned that developers currently show far greater enthusiasm for Google's AI offerings than Meta's.

"The lack of developer trust will come back to hit them if they don't focus on third-party developers," warned Krish Subramanian. "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."

Krish Subramanian, CEO, KOI AI

What Wall Street Expects From Meta's AI Investment

Investors increasingly want to see evidence that AI can generate direct revenue through products and subscriptions rather than simply making advertising more efficient. Despite reporting first-quarter revenue growth of 33 percent, its strongest pace since 2021, Meta's stock has fallen 18 percent over the past year, making it one of the weakest performers among major technology companies.

Ralph Schackart, an analyst at William Blair who recommends buying the stock, explained the market's skepticism: "Meta needs to provide more proof points of both adoption and commercialization. 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".

"Meta needs to provide more proof points of both adoption and commercialization," said Ralph Schackart, analyst at William Blair. "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."

Ralph Schackart, Analyst, William Blair

Steps Meta Must Take to Prove Its AI Strategy Works

• Demonstrate Direct Revenue Generation: Move beyond using AI to improve advertising efficiency and show that AI products can generate standalone revenue through subscriptions, licensing, or enterprise services that don't depend on Meta's ad platform.
• Rebuild Developer Trust: Actively re-engage with the third-party developer community through transparent communication, accessible APIs, and clear documentation, reversing the perception that Meta has abandoned external partners in favor of a closed ecosystem.
• Establish Computational Efficiency Leadership: Differentiate from competitors by proving that Meta's proprietary models are more cost-effective and faster to run than larger models from rivals, creating a compelling value proposition for enterprise customers.
• Release Multiple Proprietary Models: Move beyond the single Muse Spark release to demonstrate that the new strategy can consistently produce competitive AI models across different use cases and industries.

Inside Meta, pressure is mounting to prove that the enormous investment in AI can generate measurable business results. The company has spent much of the year cutting jobs, including approximately 8,000 layoffs in May, with reductions reportedly affecting several divisions, including some teams involved in trust and safety functions. Sources familiar with internal discussions say expectations are especially high for Wang and former GitHub CEO Nat Friedman, who also joined Meta's AI efforts during the hiring blitz.

Thomas Randall of Info-Tech Research Group believes the strategic shift was necessary. "There'll be a lot of these frontier model providers that will fundamentally change in lots of different ways, and Meta needs to have a consistent, reliable proprietary model that they themselves own," Randall said. He argued that Meta would be "lost" if Zuckerberg had not invested heavily in Wang and other prominent AI recruits, describing the effort as a "strategic rebuild".

The next 12 to 18 months will be critical. Meta has the resources, the talent, and the infrastructure to succeed in AI. What remains uncertain is whether the company can overcome the trust deficit it created by abandoning its open-source strategy and prove to Wall Street that AI can become a meaningful business in its own right, not just a tool to make ads more effective.