It hasn’t been a fun year to own shares of Meta.
The stock is down over the past 12 months, and the reason isn’t all that complicated. Investors have spent nearly the whole year watching the company pour tens of billions into AI infrastructure with very little to show for it on the product side. Analysts now model Meta’s free cash flow swinging negative this year, to roughly ($6.3) billion, and getting worse in 2027 at ($24.3) billion. That’s a brutal thing to underwrite when the payoff is still theoretical.
However, today, the stock jumped 11.43%, its biggest single-day gain since April 2025.
What happened
Two weeks ago, Meta released Muse, its personal AI agent. Over the weekend, the app hit No. 1 on Apple’s U.S. App Store. Wells Fargo analyst Ken Gawrelski responded by raising his price target to $796 from $640. Truist’s Youssef Squali went further, estimating Muse could add $28.5 billion in revenue by 2030 and calling it Meta’s most consequential consumer product since the smart glasses launched in 2023.
While Muse has paid tiers, Mark Zuckerberg doesn’t appear to be counting on them.
In a September 8 interview with Alex Heath on the Sources podcast, Zuckerberg laid out the potential economics directly. Meta is giving away roughly 100 million tokens a week for free, he said, along with a cloud virtual machine. He framed free access as a requirement rather than a promotion, arguing that if you want to build a future where everyone has a capable AI agent, you have to make it affordable enough that everyone can actually have one.
Then he explained how the company expects to get paid:
“We’re basically just standing behind that and saying, we think that this thing is actually going to make you money and save you money. And that is how it’s going to pay for itself.”
This is not how the company is monetizing the product right now, but it does present an interesting path that many investors might not have been considering. An agent that books, buys, plans, and transacts on your behalf sits directly upstream of commerce. Later in the same conversation, Zuckerberg said Muse can connect to Meta’s ad systems, help a user build a product, and then help run the business around it in a continuous loop. Meta already serves hundreds of millions of small businesses.
In other words, Meta could be building a toll booth rather than a paywall. That’s a far bigger addressable opportunity than $20 a month, and it’s a large part of why today’s move was the size it was as more investors open up to the possibility.
Is the turn real?
Maybe. A few things to sit with.
Meta trades at around 6x next-twelve-month sales, roughly in line with its five-year average and well below the almost 10x peak it hit in 2025. Consensus has revenue compounding 16.6% a year through 2030 to $468 billion, with normalized earnings per share reaching $52.25.
Zuckerberg’s answer to the spending question, in that same interview, was the balance sheet. Asked about funding many gigawatts of compute, he noted that unlike some of the other AI labs, Meta is an extremely profitable business, which makes those investments considerably easier to carry. He isn’t slowing down.
One App Store ranking doesn’t prove a business model works. But today was the first session all year where the market treated Meta’s AI spending as an investment with a visible return instead of a hole in the cash flow statement. Watermelon, Meta’s next frontier model, ships soon and might tell us more about how Zuckerberg is gearing up to play the game in 2027.
