Meta Platforms Inc. unveiled its first major new artificial intelligence model in nearly a year on Wednesday, signaling a shift from open-source experimentation toward proprietary AI with potential business applications.
The model, called Muse Spark, represents Meta’s biggest push yet to monetize its multibillion-dollar investment in AI, including the hiring of Alexandr Wang, co-founder of Scale AI, for over US$14 billion last year. Wang leads the newly created Meta Superintelligence Labs, tasked with developing frontier AI technology.
Meta’s CEO Mark Zuckerberg has long emphasized AI as central to the company’s future. While rivals such as OpenAI, Anthropic, and Google have dominated public attention with popular chatbots and cloud services, Meta has focused on internal development with limited new revenue. Analysts say Muse Spark marks the first step toward changing that.
“It’s been a year of basically no releases and a lot of hiring,” said Malik Ahmed Khan, analyst at Morningstar. “Meta had to show investors and operators they have been working on something of substance. That’s the first step. The second step is figuring out how to monetize it.”
Muse Spark is proprietary, a sharp departure from Meta’s previous Llama models, which were open-source. The company said it plans to release some open-source versions eventually but will initially limit access through a “private API preview” for select partners. Paid API access may follow, echoing strategies used by leading AI labs.
More Read
Industry analysts note that Meta enters the commercial AI market late. OpenAI and Anthropic are valued at over US$1 trillion combined, and Google has already integrated its Gemini models across its cloud services and consumer apps. To succeed, Meta must deliver a model competitive in quality and provide tangible business value.
Analysts say the most promising avenue for Meta is its core advertising business. With more than three billion monthly users across Facebook, Instagram, and WhatsApp, AI can enhance ad targeting and engagement, potentially driving higher returns for advertisers. Advertising accounted for 98 percent of Meta’s $200 billion revenue in 2025.
“From Meta’s perspective, improving ads is the killer use case,” said Khan. “As advertisers see better ROI, they are willing to reinvest, which makes AI monetization viable.”
More Read
Technical benchmarks released by Meta suggest Muse Spark excels in image and video processing, areas critical for dynamic ad campaigns, short-form content, and visual storytelling. Analysts note that these strengths align with Meta’s strategy to boost engagement on platforms like Instagram Reels and Facebook feeds.
However, some developers question the model’s value outside Meta’s ecosystem. Proprietary models limit fine-tuning compared to open-weight alternatives, which many AI researchers and startups rely on for custom applications. Joseph Ott, CEO of Samu Legal Technologies, said it is unclear how Muse Spark will compete with cheaper or free AI tools.
Meta’s shift reflects a broader goal of AI sovereignty. Analysts like Ulrik Stig Hansen, co-founder of Encord, say Meta aims to maintain independence from third-party AI providers and remain a major player in the global AI market.
With Muse Spark now launched, the pressure is on for Meta to translate technical capabilities into revenue. Andrew Boone, analyst at Citizens, summarized the challenge: “We just gave you a state-of-the-art frontier model. What are you going to do with it?”
The coming months will test whether Muse Spark can do more than impress benchmarks and deliver a profitable AI strategy for one of the world’s largest tech companies.
