Muse launched in September in the U.S. and Canada with an unusually forceful advantage: Meta’s vast distribution network and a substantial advertising push. Sensor Tower counted 10.9 million downloads in its first 40 days, including 10.2 million in the U.S. — a threshold reached by only 11 other apps in that timeframe. Meta reportedly devoted 39% of its October ad spending to Muse, while early retention also outpaced rival AI apps.¹

By Oct. 16, daily active users had reached 2.5 million. A week later, Mark Zuckerberg told investors that more than 10 million people had set up Muse agents, with roughly a quarter using them each day. The most consequential early behavior was not email or calendar management but shopping: Zuckerberg said the average purchase exceeded $200, and Meta had to ask Stripe to raise credit limits after transaction volumes beat expectations.²

That gives Meta a clear commercial case. It plans to keep Muse broadly free, charge for heavier model use and take a cut when the agent completes sales. Zuckerberg’s larger ambition is a “Muse network,” in which consumers’ agents deal directly with business agents rather than navigating websites — a system Meta says could ultimately earn more per user than Facebook and Instagram.³

But the speed of the launch has sharpened the objections. Retailers may resist ceding the customer relationship and product discovery to Meta; consumers, meanwhile, are being asked to grant an agent access to sensitive services. Forrester’s Kate Winick called trust the central unknown, arguing that Muse becomes more useful as it gets more information even as “overall trust in Meta remains low.”⁴

The financial pressure is equally stark. Meta expects $135 billion to $145 billion in annual capital spending, while Muse remains too compute-intensive for a global rollout. Zuckerberg says the company must improve cost efficiency by more than tenfold next year. The bet is that convenience will overcome those frictions before the bill for the AI future does.⁵