Meta Platforms delivered a strong, albeit somewhat noisy, third quarter in its first reporting period since the release of the viral Muse AI agent. Investors responded positively. Revenue rose 26% year over year to $64.6 billion, exceeding the $63.37 billion analyst consensus compiled by LSEG by more than $1 billion.
Reported earnings complicated the picture. Diluted earnings per share were $3.83, but would have been $7.40 excluding a $9.89 billion, or $3.57 per share, legal settlement charge related to teen addiction litigation involving state attorneys general. Analyst estimates ranged from $2.61 to $7.57, suggesting that some included the settlement in their calculations while others may have excluded it.
Strong guidance supports investor confidence
Wall Street appeared to look through the complexity surrounding earnings. Meta shares rose 2% in after-hours trading following the report, supported by a strong revenue outlook and a measured capital-expenditure forecast.
Shares had climbed sharply since Muse launched on Sept. 8, closing at a 52-week high of nearly $780 on Sept. 24. The position had been trimmed at just under $750 two days earlier because of the rapid rally. Although Meta has since pulled back from those highs, it remains well above its pre-Muse level.
Muse shows early signs of adoption and retention
Muse was the central focus of Meta’s earnings conference call. Zuckerberg described the product as an early success with a promising path toward monetization, noting that more than 10 million people in the United States and Canada had set up Muse personal agents in just a month and a half.
The number of downloads alone is not the key metric, however. Sustained usage will determine whether Muse can become a meaningful platform. Meta reported relatively solid retention, with about 25% of people who set up Muse using it on any given day.
“For our earliest adopters, the longer that they use Muse, the better that retention seems to get. To put this in perspective, this is more than twice the level that Threads started at, or where we believe that other AI apps started at,” Zuckerberg said.
Shopping has emerged as the primary use case so far, with more than 25% of Muse users engaging in agent-led shopping activities. Plaid told Meta that it had not seen activity comparable to Muse’s on any other service. During Muse’s first week, Meta also worked with Stripe to raise credit limits because transaction volumes exceeded expectations. Plaid and Stripe provide technology that facilitates online payments.
A multi-pronged monetization strategy
Meta still has substantial work ahead if it hopes to capitalize on Muse’s viral launch. The company must scale the product for a global audience, develop new revenue streams and bring down the cost of serving increasingly sophisticated AI models.
Zuckerberg said Meta intends to build a network of Muse agents as quickly as possible. The approach would give the system more opportunities to interact with users, connected services and the wider online ecosystem, potentially improving its capabilities over time.
The planned monetization model includes several components:
- Free access for most users: Zuckerberg expects the free Muse tier to be sufficient for the core audience. Meta plans to develop e-commerce tools that allow the company to earn a share of transactions facilitated through Muse.
- Paid plans for power users: More demanding users will be able to pay for access to the latest models and higher token budgets.
- Business and enterprise offerings: Meta also plans to provide additional features for companies seeking to grow their businesses or improve employee productivity.
Compute efficiency is central to the strategy
Zuckerberg acknowledged that Muse remains highly compute-intensive, but expects Meta to improve its cost efficiency by more than 10-fold over the next year. He said this could give Meta an advantage over pure-play AI developers such as OpenAI and Anthropic, as the company can continue building its agentic products before they become profitable, supported by a highly profitable core advertising business.
“If compute efficiency continues improving quickly for another few years, which we think is highly likely, then Muse will be highly profitable over multiple years. In the near term, the strength of our core business enables us to scale the Muse network before then, and we believe that we are the best positioned company to pursue this strategy,” Zuckerberg said.
Rising costs associated with bringing Muse to full scale were a concern ahead of the report, making Meta’s decision to raise the lower end of its annual capital-expenditure range while leaving the upper end unchanged a positive signal. The company is also seeking to turn its AI infrastructure into a broader service through the Meta Enterprise Platform, led by CJ Desai, the former CEO of MongoDB.
“One other note on compute economics is that since training is a large fixed cost, now that we have leading models, we want to serve as much high-margin inference demand as possible. That is why we are building the Meta Enterprise Platform led by CJ Desai,” Zuckerberg said.
He added: “Muse Wonder and even more advanced models that we’re training can serve additional high-margin services beyond the Muse network, including our API, Muse Code and other enterprise services. A fundamental feature of our business is that we have many good uses for this compute.” Muse Wonder is a new AI model.
Meta is also preparing to deploy more advanced models across a wider range of services. Zuckerberg said the company expects to use AI internally to improve efficiency as the technology advances, while the range of external applications remains difficult to forecast.
“It is difficult to forecast the exact growth rates for the Muse network, for our enterprise platform and for how Muse, Wonder and beyond will accelerate our core business. But through building up multiple ways to deploy these advanced models, we increase the chances that one or more of them significantly outperforms our expectations,” he said.
Ad business and Reality Labs performance
Meta’s Family of Apps, the division responsible for most of the company’s revenue through digital advertising, continued to benefit from stronger video consumption. Facebook and Instagram recorded record engagement during the quarter, while Threads, the newest addition to the family of apps, surpassed X in monthly active users. WhatsApp also reached an all-time high for messages sent.
Global average price per ad increased 14% year over year, accelerating from the 12% increase reported in the previous quarter. Ad impressions across the Family of Apps rose 10%, although that represented a sequential deceleration. Daily Active People fell short of expectations, but better-than-expected average revenue per person across the family of apps offset the shortfall.
Reality Labs, which houses Meta’s virtual- and augmented-reality projects and smart-glasses business, reported a $4.13 billion quarterly loss. The loss was smaller than analysts expected but greater than in the year-earlier period. Zuckerberg said Reality Labs would introduce new products only when the company achieves a meaningful technological breakthrough. He forecast operating losses of $15 billion for the coming year, followed by a trend toward profitability thereafter.
Meta also announced plans to create a new class of non-voting common stock, Class C, effective January 2027. The shares are intended for capital raising, employee compensation and potential acquisitions. The class will also be used in a two-for-one stock split, under which shareholders receive one Class C share for every Class A share they own.
Price target and remaining risks
Given the strong quarterly results, Meta’s execution and its position to capture a larger share of the AI market, the analysis reiterates an $800 price target for the stock. It also maintains a hold-equivalent rating of 2 for the time being.
Several broader risks remain. Meta is facing a court-ordered $25 billion penalty from New Mexico related to the Cambridge Analytica scandal from nearly a decade ago and has said it will appeal. The upcoming midterm elections also introduce uncertainty for Meta and the broader market.
Guidance
Meta projected fourth-quarter revenue of $73 billion to $76 billion, above the $73.88 billion LSEG consensus estimate. For full-year 2026, the company increased projected total expenses to a range of $175 billion to $179 billion from the previous $165 billion to $169 billion range. The revision largely reflects the nearly $10 billion charge recorded in the third quarter and was largely overlooked by investors.
Full-year capital expenditures are now expected to range from $135 billion to $145 billion, compared with the previous $130 billion to $145 billion forecast. The $139.2 billion midpoint is broadly consistent with analyst expectations, according to FactSet.