Nearly ten months after Meta invested billions to recruit Scale AI's Alexandr Wang as the cornerstone of Mark Zuckerberg's AI transformation, the company introduced Muse Spark on Wednesday—its first new model since the strategic shift. A central question remains whether users will be willing to pay for the offering.

While competitors such as OpenAI, Anthropic, and Google have led the AI surge with advanced models, popular chatbots, and other services, Meta has spent heavily on artificial intelligence without yet demonstrating new revenue sources from these efforts.

In June, Meta allocated over $14 billion to hire Wang along with several of his top engineers and researchers, quickly establishing Meta Superintelligence Labs as a new elite division. Then, in January, the company informed Wall Street that it expects to invest between $115 billion and $135 billion in capital expenditures this year—nearly double its 2025 capex projection.

Morningstar analyst Malik Ahmed Khan noted, "It's been a year with essentially no product releases but significant hiring, and capex concerns for this year are evident. Meta needed to demonstrate to investors and operators that they have been developing something substantive. This is the initial step."

According to Khan, the next phase for Meta involves ensuring the model functions effectively and devising a monetization strategy.

Muse Spark, Meta's newly launched model, is proprietary—a marked departure from its predecessor Llama, which was open-source. However, the company indicated it still intends to release certain open-source versions in the future. Zuckerberg overhauled the company’s strategy following the April launch of Llama 4, which failed to attract significant developer interest.

Gartner analyst Arun Chandrasekaran characterized the move as a "major shift" and said it "signals an intention to move away from the Llama brand."

Following the approach of other leading AI labs, Meta plans to eventually offer paid API access to Muse Spark for third parties, starting with a "private API preview" available to select partners.

However, Meta is entering the field late. OpenAI and Anthropic are jointly valued at well over $1 trillion due to the popularity of their models and services. Google has integrated its Gemini model across its app and product portfolio and also sells access to Gemini models through its cloud division.

For Meta's AI technology to succeed, it must be competitive with leading models while also creating a unique business opportunity.

'Crown Jewel'

Citizens analyst Andrew Boone pointed out that Meta's clear advantage lies in its monthly user base of over 3 billion people across Facebook, Instagram, and WhatsApp. The primary business opportunity for Meta, he suggested, is not in attracting developers—who currently favor OpenAI, Anthropic, Gemini, and various Chinese models—but in focusing on its core market: advertising.

"That's the crown jewel, that's what needs to continue to improve," said Boone, who recommends buying Meta stock.

Khan expressed a similar view.

"I believe that would be the killer use case from Meta's perspective," Khan stated, with the objective being to "make ads more engaging and improve targeting."

Advertising represented 98% of Meta's $200 billion in revenue last year. Although the company has made numerous attempts to diversify—most notably by spending tens of billions of dollars to advance the metaverse—its advertising model remains its consistently successful venture. Meta's AI investments have primarily enhanced its targeting capabilities and provided better tools for marketers.

Khan explained that as advertisers see returns on their Meta ad spend, they reinvest those gains into additional ads on the platform. Therefore, it is logical that they would be willing to pay for AI services if they can achieve even better outcomes.

Meta declined to comment on its API plans beyond the initial announcement.

Based on technical benchmarks released by Meta comparing Muse Spark to rival models, the new AI appears to excel in image and video processing, according to Doris Xin, CEO of AI startup Disarray. These capabilities are crucial for advertisers aiming to create dynamic campaigns for audiences accustomed to short-form videos on Reels or viewing photos on Facebook and Instagram.

"Compared to models like Claude and Gemini, I think it definitely feels like it has more of a consumer bent," Xin remarked about Muse Spark.

Zuckerberg, however, has long held ambitions that extend beyond advertising. His strategy with Llama targeted developers, aiming to attract top AI talent to use Meta’s tools—even if they weren't paying customers.

Shifting to proprietary models makes the appeal to developers more challenging. Joseph Ott, CEO of AI startup Samu Legal Technologies, expressed uncertainty about where he would find value in the new offering.

"The only reason I would use Llama is that I could fine-tune it," Ott said, referring to the practice of customizing AI models.

Many developers rely on open-weight AI models, such as those offered by Chinese tech firms, as a foundation to train models for specific applications. Ott noted that it is unclear what would differentiate Meta’s Muse Spark from free or lower-cost alternatives, as well as leading proprietary AI models.

Ulrik Stig Hansen, co-founder of AI and data training startup Encord, emphasized the importance of Meta developing its own foundation models to avoid future reliance on third parties. As one of the few companies with the resources and computing infrastructure required to build and maintain large AI models, Meta aims to maintain its relevance in the highly competitive AI market.

"It is about AI sovereignty and being a player in the game," Hansen stated. "They want to be perceived and known as an AI company."

Regarding Meta's substantial investment in Wang and his team, Boone suggested that recent benchmark results indicate Zuckerberg obtained what he sought, and now the responsibility returns to leadership.

"We just gave you a state-of-the-art frontier model," Boone said, referring to the team behind Muse Spark. "What are you going to do with it?"