Meta is winning over Wall Street with its new Muse AI agent

By Britney Nguyen

A J.P. Morgan analyst recommends Meta's stock following signs of progress on consumer-focused AI applications

J.P. Morgan upgraded Meta's stock on Thursday.

Meta Platforms hasn't had as clean an artificial-intelligence narrative as some of its Big Tech rivals, but that may now be changing.

J.P. Morgan analyst Doug Anmuth is ready to recommend Meta's stock (META) following the recent launch of Muse, which the company said "proactively helps with people's goals" while running privately and securely. The AI agent, powered by Meta's Muse family of models, reached the No. 3 spot on Apple's App Store in the U.S. as of Thursday.

Meta may be in the early stages of rolling out AI products to consumers, said Anmuth, who sees "meaningful upside potential" as the company makes headway with the personal agent and frontier models.

Anmuth upgraded Meta's stock to an overweight rating - a buy equivalent - from a hold-equivalent neutral, and he raised his price target to $820, implying the stock has room to run about 26% higher than its $651 price as of Thursday morning.

The social-media company has so far focused much of its AI efforts on improving its advertising business, Anmuth said, but its AI models will be the foundation of its "product and monetization pipeline over a multiyear period, and its path toward superintelligence." Superintelligence is the point at which AI will have surpassed even the smartest humans across several domains.

Anmuth noted the launch of Meta Superintelligence Lab last June, which has now "essentially delivered" with its Muse Spark 1.1 and 1.3 frontier models that he said are competitive with cutting-edge offerings from OpenAI and Anthropic.

Meanwhile, he said, the Muse agent's strong debut reflects his belief that Meta can deliver consumer AI products to its base of four billion users and that this level of reach "is a significant competitive advantage" for the company.

That the AI agent runs on a virtual machine, which houses it with a person's data, is another advantage, according to Anmuth. He said that security "could be a key hurdle for other agents to gain trust" from users when it comes to more ambitious tasks that include passwords and other sensitive personal information. He expects agent-to-agent activity to increase as consumer adoption of AI grows.

Although Meta is not focused on the monetization of Muse in the near term, Anmuth said he sees "meaningful opportunities" stemming from commissions and subscriptions that could unlock a total addressable market worth "tens of trillions of dollars," he said.

Additionally, access to Meta's models through its application-programming interface for developers and enterprises "should become increasingly compelling" given the strength of its offerings, including an upcoming model code-named Watermelon. Alexandr Wang, who heads MSL, reportedly told employees in July that Watermelon had caught up to OpenAI's GPT-5.5, which had launched in April, according to Business Insider.

Anmuth said Meta's spending on AI computing power will reach $243 billion next year and $284 billion in 2028 as the company looks to meet the demands of running increasingly powerful models.

-Britney Nguyen

This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.

(END) Dow Jones Newswires

09-10-26 1123ET