Meta shares jump 6% as Zuckerberg's new AI agent Muse wins over Wall Street

Meta Platforms shares climbed as much as 6.8% Wednesday, their sharpest one-day move in months, after the company unveiled Muse, a stand-alone personal AI agent that Chief Executive Mark Zuckerberg says can book travel, negotiate purchases and manage a user's inbox without being asked twice. The rally pushed Meta's stock above $655 and added tens of billions of dollars to its market value, as Wall Street bet that the product marks the start of Meta finally turning its enormous artificial intelligence investment into a subscription business.

The stock had opened Wednesday already up more than 5% in premarket trading, extending gains that began Tuesday night when Zuckerberg posted that Muse is "the personal agent that understands your goals and works 24/7 to get things done for you." By midday the shares were trading around $649 to $655, according to intraday pricing tracked by StockAnalysis.com's live data on Meta, versus a close near $610 the day before.

What Muse actually does

Meta described Muse, in a post on its official newsroom announcing the launch, as "the world's first personal AI agent built for everyone" — a program that "doesn't just answer questions, it actually does the work." Rather than responding to one-off prompts, Muse is designed to build a running action plan around a user's stated goals, track progress on it, and proactively chip away at tasks: filling out a permission slip for a class field trip, scheduling a tennis lesson, buying movie tickets, or checking out online using Link by Stripe. It can also mine content users have already saved — Meta's example is turning a recipe bookmarked from an Instagram reel into an actual shopping list and cooking plan.

The agent runs on Muse Spark, a reasoning model Meta Superintelligence Labs introduced this year and has described as requiring "over an order of magnitude less compute" than the company's prior flagship model to hit comparable benchmark performance. Muse is rolling out this week to U.S. users 18 and older as a stand-alone iOS and Android app and at muse.ai, and is also reachable inside WhatsApp; Meta says it is "coming soon" to the company's AI glasses line. Pricing is free for most everyday use, with two paid tiers — $20 and $100 a month — for people who want the agent to handle longer or more complex chains of tasks.

Meta has leaned hard into safety messaging around the launch, aware that a product designed to send money and messages on a user's behalf invites obvious risks. The company says Muse runs inside a "Muse Secure VM," an isolated virtual machine that houses both the agent and a user's data, watched over by a separate monitoring system Meta calls Sentinel that must approve anything before it reaches the open internet. A "Secure Credential Store" is meant to keep the agent from ever seeing a user's actual passwords or payment card numbers.

"We built a Secure Credential Store for your passwords and credit cards so Muse can't read this information. Muse also checks directly with you before taking sensitive actions like making payments or sending messages."

Meta AI chief Alexandr Wang, who leads Meta Superintelligence Labs, added that the agent operates in "its own isolated environment" within the company's infrastructure and "never sees your actual passwords or payment details." Users are opted in by default to having their Muse interactions used to train Meta's AI models, and must actively opt out if they don't want that — a design choice already drawing scrutiny from privacy advocates, given the sensitivity of tasks the agent is built to handle.

Why investors care about the money, not just the gadget

The enthusiasm on Wall Street is less about the novelty of an AI assistant — OpenAI, Google and Anthropic all have competing agent products — and more about Meta's balance sheet. In its second-quarter results, filed with the Securities and Exchange Commission on July 29, Meta reported revenue of $60.8 billion, up 28% year over year, but diluted earnings per share fell to $6.18 from $7.14 a year earlier, and free cash flow collapsed to $784 million from $8.5 billion, as capital expenditures hit $31.1 billion for the quarter. The company told investors it now expects full-year 2026 capital spending of $130 billion to $145 billion, narrowed upward from a prior range of $125 billion to $145 billion, with nearly all of the increase going toward data centers and AI infrastructure.

That spending has left investors waiting for proof that Meta's AI bet will eventually generate revenue beyond advertising, rather than simply weighing on free cash flow indefinitely. Muse's paid subscription tiers are the company's clearest attempt yet at a direct-to-consumer AI revenue line, distinct from the ad-supported model that has defined Facebook and Instagram for two decades.

Mizuho Securities reiterated an Outperform rating on Meta with a $750 price target, telling clients that "Meta's Muse consumer AI agent marks the beginning of a substantial product cycle for Meta that is not priced into shares." KeyBanc Capital Markets kept an Overweight rating with a $780 target, writing that it continues "to believe the market underestimates Meta's AI positioning and product cycle," sentiment echoed across Wall Street's reaction to the launch.

The optimism was not universal across the sector. Alphabet shares fell roughly 2% to about $330 the same day, with some traders reading Muse as a direct challenge to Google's Gemini assistant, which Alphabet says has around 950 million monthly active users.

What comes next

Meta has already signaled Muse is an early step rather than a finished product: the company says a "Muse Confidential VM" with end-to-end encryption controlled by user-held keys is coming, and the agent's rollout beyond the U.S. and beyond phones — into Meta's AI glasses — has not yet been dated. The real test for investors will arrive with Meta's third-quarter results, when analysts will look for early signs of subscription uptake and whether Muse can move the needle on the free-cash-flow pressure created by the company's AI infrastructure spending. Regulators and privacy groups are also expected to press Meta on the default opt-in for AI training data and on how the Sentinel monitoring system performs once millions of users start handing the agent real purchases, real messages and real money to manage.