The Zhitong Finance App learned that Pivotal Research sharply raised the target price of Meta Platforms (META.US) from $790 to $930 on Tuesday, maintaining a “buy” rating, implying an upward margin of about 25%. Analyst Jeffrey Wlodarczak's core logic is: Meta's return on investment in the AI field is “coming into view” — Muse personal AI assistants were strong in the early stages of adoption, and the Muse Wonder model was at the cutting edge after release.
This judgment is not isolated on Wall Street. The 63 analysts compiled by S&P Global agreed that the target price was $799, with a consensus rating of “Strong Buy,” but the average target price only implied a 7.5% upside compared to Meta's closing price of $742.96 on Monday. Pivotal's $930 is about 16% higher than consensus, which is a high-end reading.
Two pivots Pivotal is bullish on
Pivotal's multi-view logic is based on two verifiable pieces of data.
First, the installation speed of Muse. According to the research report, the cumulative number of installations of Muse reached 2.8 million in the 12 days before launch, which surpassed the performance of ChatGPT when it was released on mobile in the early days. This figure is supported by third-party data — Sensor Tower estimates that the data for the same period was 3.7 million global downloads of ChatGPT, and the “beyond” cited by Pivotal is based on the caliber comparison of Apple devices in North America.
Second, substantial catch-up of model capabilities. Pivotal believes that Muse Wonder is the first cutting-edge model released by Meta since Alexandr Wang restructured the AI business in August 2025, and the performance is at or close to that of leading laboratories. Independent evaluation agency Artificial Analysis confirmed this judgment: Muse Wonder Full scored 56 on the Intelligence Index, on par with Anthropic's Claude Sonnet 5.5, surpassing OpenAI's GPT-6 Astra and Google's Gemini 4 Argon, and second only to Claude Opus 5.5 with 58 points. The standard version of Muse Wonder scored 52 points.
At the product level, Muse Wonder introduces two core capabilities for Agentic scenarios: Active Memory is used to preserve user information across conversations, and Active Focus allows the model to continue advancing tasks rather than abandoning goals when obstacles are encountered. The model supports a contextual window of approximately 1 million tokens, and the training was completed at Meta's Prometheus data center in Ohio.
Download first, save not necessarily
However, behind the brightness of the downloaded data, the retention curve is not optimistic.
According to Sensor Tower data, Muse's first-day retention rate was as high as 73.3%, far higher than ChatGPT's 41.3%. However, the advantage narrowed rapidly after the 7th day, and the Muse retention rate dropped to 22.5% on the 14th day, down from 26.8% in the same period of ChatGPT, and lower than the level of Meta's mature products such as WhatsApp, YouTube, and Instagram.
BNP Paribas analyst Nick Jones has previously pointed out that although Muse's “app open rate” is stabilizing, it is still lower than mature apps under Google, ChatGPT, and Meta, indicating that the product has not really been integrated into users' daily usage habits. The bank maintained Meta's “outperforming market” rating and a target price of $885, but lowered Muse's long-term monetization expectations. Jones's core judgment is that Muse's initial download advantage came largely from the channel distribution capabilities of the Meta social ecosystem. “Easy to download and difficult to store” was a common dilemma for consumer-grade AI products.
The historical baggage of monetization
Needham analyst Laura Martin maintains a “holding” rating without setting a target price. Her argument is that Meta “is known for being slow in monetizing new products.”
This judgment is supported by historical records. Meta bought Instagram for 1 billion dollars in 2012, and the first advertisement did not appear until the end of 2013; in 2014, it bought WhatsApp for 19 billion US dollars, and did not start advertising until June 2025, with an interval of more than ten years. Martin pointed out that Muse is currently free, and Meta has made it clear that it will not share user conversations and data with the advertising business, “which suggests low monetization potential.”
More realistic constraints come from the capital expenditure side. Meta's 2026 capital expenditure guideline is US$130 billion to US$145 billion. Q2 capital expenditure reached US$31.1 billion in a single quarter, and free cash flow plummeted to US$784 million from US$8.55 billion in the same period last year. What Pivotal said “the return on AI investment is coming into view” needs to be examined at this rate of cash consumption. Whether Muse can move from a “user growth story” to a “revenue growth story” will determine whether the target price of $930 is forward-looking or ahead of schedule.
Meta will release its Q3 earnings report after the market on October 21, EST. The market expects quarterly revenue of about US$63.15 billion. At that time, investors will have the first opportunity to observe Muse's actual revenue contributions after launch, as well as management's specific guidance on capital expenditure and AI monetization paths in 2027. Pivotal's aggressive pricing will be tested for the first time in two weeks.