Meta CEO Mark Zuckerberg at the Meta Connect event in Menlo Park, California, September 17, 2025.
David Paul Morris | Bloomberg | Getty Images
As Mark Zuckerberg relies on Meta’s new AI model, Muse Spark, to reposition his company within the booming AI landscape, insightful commentary is expected following the first-quarter earnings report on Wednesday.
The introduction of this model, previously known as Avocado, came just as the second quarter began. Muse Spark signifies a pivotal shift in Meta’s AI approach, distinguishing it from earlier models such as the Llama models, which were distributed freely within the open-source community.
Meta has expressed its intention to monetize this technology by providing paid access to developers, mirroring strategies employed by OpenAI, Anthropic, and Google. Analysts suggest that it’s crucial for Meta’s AI tools to enhance its already significant advertising business, while demonstrating capabilities that compete with leading market players.
According to Arena.AI, which monitors the performance and quality of leading AI models, Meta AI is currently ranked below Anthropic’s Claude and Google’s Gemini in text generation but only trails Claude in visual tasks. However, it surpasses OpenAI’s GPT in both domains. Claude is ahead in document and coding tasks, where Meta’s rankings are less favorable.
In a client report last week, Citizens analysts labeled AI as a “complementary good” for Meta, anticipating additional insights during the upcoming earnings call.
“We are impressed with Meta’s Muse Spark model,” the analysts stated, recommending the stock for purchase due to its strengths in text and visual capabilities. “While the company has integrated Meta AI into its core applications, we await a strategy that drives consumer engagement akin to other prominent AI chatbots like ChatGPT and Claude, believing this could unlock new data and advertising budgets.”
Meta’s advertising segment continues its robust growth, propelled by enhanced targeting capabilities resulting from advancements in AI technology. Analysts predict a year-over-year revenue increase of 31% for the first quarter, reaching $55.6 billion—a rate of growth not seen since 2021.
However, Wall Street seeks AI growth beyond advertising, as OpenAI and Anthropic’s combined valuations have surged past $1 trillion, fueled by the popularity of their AI models and services. Meta’s stock has risen by 24% over the past year, while Alphabet’s shares have soared by 116%, aided by the expansion of Gemini.
When Meta launched Muse Spark earlier this month, it was promoted as the first significant AI model emerging from Meta Superintelligence Labs, led by Alexandr Wang, the company’s chief AI officer. Wang was formerly the CEO of Scale AI and joined Meta in June as part of a $14.3 billion acquisition deal for the data-labeling startup.
Zuckerberg’s recruitment of high-profile talent continued with the hiring of former GitHub CEO Nat Friedman, alongside business partner Daniel Gross, who was previously the CEO of AI startup Safe Superintelligence, founded by Ilya Sutskever in 2024 after leaving OpenAI.
“This shift in leadership and the subsequent nine-month overhaul of Meta’s AI infrastructure indicate a committed effort to catch up with rivals like OpenAI (currently private) and Google,” Truist analysts noted in a report on April 21. “Importantly, Muse Spark is closed-source, marking a departure from the open-source approach of Llama and a transition towards high-performance, specialized infrastructure.”
‘Back into the AI conversation’
Meta highlighted that its internal evaluations, released alongside Muse Spark, showed it to be less powerful compared to cutting-edge AI models from competitors like Anthropic, a strategy to moderate early expectations.
Nevertheless, analysts have expressed relief that Meta is finally progressing, anticipating more models to follow. JPMorgan Chase analysts stated in a recent report that Muse Spark “has reintegrated Meta into the AI dialogue.”
“Investor sentiment regarding Meta is becoming increasingly positive,” the analysts remarked. “The stock has faced pressures from high expenses and capital expenditures, concerns over delays in AI model launches, and adverse legal decisions related to social media.”
In parallel, Meta is trimming its workforce to sharpen its focus on AI.
The company announced Thursday that it would lay off 10% of its workforce, approximately 8,000 employees, on May 20, in a bid to enhance operational efficiency. This move comes as Meta invests heavily in AI infrastructure, recently informing investors that projected AI-related capital expenditures for 2026 could range from $115 billion to $135 billion, a significant increase from $72.2 billion in 2025.
Analysts from Loop Capital have noted that Meta’s substantial investments have fostered a negative perception, depicting the company as “desperately spending to rectify problematic AI initiatives.” However, the launch of Muse Spark implies that Meta is developing AI models capable of enhancing its core online advertising business.
Even if Muse Spark and future models underperform compared to competitors, their testing remains “of mixed importance,” as Meta holds a distinct advantage in the advertising sector, according to Loop analysts.
“While foundational LLM/agentic reasoning models are vital for Meta, we perceive image/video generation models as strategically significant, with more immediate engagement and monetization potential,” they stated. “The primary measure of success is developing models that produce excellent products for users, creators, and advertisers.”
WATCH: Meta and Broadcom deepen AI chip partnership through 2029.
