Meta Platforms shares have dipped roughly 1% year-to-date, trailing the S&P 500's 12% gain over the same span. Despite that underperformance, the stock has already rebounded about 20% from its recent trough, while the broader index has slipped 1% during that stretch.
JPMorgan has issued a research note arguing that Meta Platforms (META.US) still has substantial upside ahead, as the company sits in the early innings of launching frontier models and AI-driven products beyond its core advertising business. The bank is particularly focused on the Muse AI agent and access to the Meta Model API as key catalysts.
Where the opportunity lies
The bank has lifted its rating on Meta from Neutral to Overweight, while raising its December 2027 price target from $640 to $820, implying roughly 30% additional upside through the end of next year. JPMorgan believes frontier models will serve as the backbone of Meta's product and monetization pipeline over the coming years, charting a course toward superintelligence.
When Meta began rebuilding its Superintelligence Labs unit in the summer of 2025, the stated goal was to deliver frontier-level models within a year. JPMorgan assesses that the team has largely hit that target, and the cadence of model releases is accelerating — from Muse Spark 1.1 in July to the more recent Muse Spark 1.3, which now holds its own against Claude and GPT-class models. The bank anticipates the next-generation model, codenamed Watermelon, will unlock fresh opportunities across consumer products, business intelligence, Family of Apps engagement, advertising, and internal operational efficiency.
Why distribution scale matters
JPMorgan has long argued that Meta is exceptionally well positioned to deliver consumer-driven AI products to an audience of roughly 4 billion users, with massive distribution capabilities serving as a formidable competitive edge. The Muse AI agent is already demonstrating strong early traction, briefly climbing to No. 3 on the U.S. App Store just one day after launch, with early usage reaching ten times the size of the training cohort. While monetization is not an immediate priority for Muse, JPMorgan sees meaningful opportunities in both take-rate and subscription models, with the potential total addressable market reaching tens of trillions of dollars.
A key differentiator for Muse is its privacy-first architecture, which JPMorgan views as a genuine advantage. The bank suggests this could become a critical hurdle for competing agents seeking to earn user trust when handling sensitive tasks involving passwords, credit cards, and other personal data. Over time, the bank expects agent-to-agent interactions to become more commonplace, granting merchants and enterprises smarter capabilities and higher transaction volumes.
Beyond the consumer-facing agent, JPMorgan also sees considerable potential in the Meta Model API service aimed at developers and businesses. As Meta's model capabilities advance from Muse Spark 1.3 to Watermelon and beyond, the bank believes the API offering will grow increasingly compelling, supported by aggressive pricing, driving further adoption and usage expansion.
Capital spending and the long-term picture
Although Meta is heavily focused on frontier model development and AI product innovation, JPMorgan stresses that its AI infrastructure capacity carries substantial standalone value and will underpin future compute demand. Meta has indicated it intends to maximize compute utilization through 2026 and 2027, while retaining greater flexibility from 2028 onward. JPMorgan projects Meta's capital expenditures will reach $243 billion in 2027, up 70% year over year, followed by $284 billion in 2028, a 17% increase — both figures running well ahead of consensus estimates.
The bank points to data center construction timelines and associated headwinds, AI product adoption rates and monetization progress, and supply constraints as the key variables that could influence 2028 capex. JPMorgan currently expects Meta to face meaningful free cash flow pressure in 2027 and 2028, with negative FCF running in the range of $65 billion to $70 billion annually. Notably, however, the bank has not yet incorporated revenue from new AI product commercialization into its forecasts.
The core advertising engine
What remains critical, in JPMorgan's view, is that Meta's core advertising business still has massive room to grow, driven by AI-enhanced content recommendations, deeper user engagement, sharper ad targeting and retrieval, and AI-assisted content creation. The bank believes Meta's scale distribution and engagement continue to reinforce powerful network effects, while its precision targeting delivers exceptional value to advertisers. Improvements in AI-driven ad ranking and recommendation capabilities are already translating into higher engagement, stronger return on ad spend, and increased revenue.
JPMorgan concludes that Meta has laid out a credible path to monetize its AI-driven capital expenditures through consumer-facing and enterprise agents, as well as the Model API offering.
Comments