Meta says Muse has over 10M users, with shopping a top activity

Meta disclosed on its third-quarter earnings call on Wednesday that its AI app Muse has passed 10 million users, matching an estimate published earlier in the week by market intelligence firm Sensor Tower. CEO Mark Zuckerberg used the call to describe not just reach but behavior: roughly a quarter of the people who set up Muse use it daily, a figure he tied to improving retention.
The more striking number concerned commerce. Zuckerberg said that among Muse users who are engaged and shopping through the agent, more than a quarter had an average purchase size above $200. He presented this as a surprise even to people inside the industry. Muse handles payments through a connector from Plaid, the financial technology company that links payment apps to bank accounts, and Zuckerberg said Plaid reported seeing activity with its connector on Muse at a level it had not seen on any other service.
That detail is worth pausing on for developers building agent products. Agentic shopping means an AI assistant does more than recommend a product; it carries out parts of the transaction on the user's behalf, which requires a secure link between the agent and the user's financial accounts. Plaid's connector is one common way to establish that link in fintech. A $200-plus average order size, if it holds as more users arrive, suggests buyers are willing to delegate meaningful spending rather than just small trial purchases — though the company's figures describe engaged shoppers specifically, not the whole user base.
Beyond shopping, Zuckerberg said popular uses include personal productivity, such as connecting the agent to a user's email or calendar. He framed that adoption as evidence users trust Muse's security setup, which includes a feature called Muse Secure VM, described as protecting user data and privacy. He also pointed to a forthcoming Muse Confidential VM, which he said will keep user content hidden even from Meta itself. In practice, confidential computing typically relies on hardware-isolated environments, or enclaves, where data is encrypted while being processed rather than only at rest or in transit; such approaches are attractive for agents that read mail and calendars, but they add engineering complexity and can constrain which workloads run inside them.
With Muse at what Zuckerberg called product-market fit, he said Meta's focus shifts to scaling the agent, growing revenue and improving compute costs. The monetization plan Meta has described keeps Muse free for most users while taking a percentage of commerce conducted through the platform, with paid tiers for power users who want the most advanced models or more tokens.
The ambition is large. Meta forecasts that Muse will eventually generate more revenue per user than its family of apps — Facebook, Instagram, WhatsApp and Messenger — do today, a figure Zuckerberg put at around several hundreds of dollars annually. He also said Meta believes Muse will be profitable once it scales. On the cost side, he told analysts Meta is building capacity to serve more than a billion people over the next few years, that Muse is very compute-intensive to run near term, and that the company expects to improve its cost efficiency by more than 10x in the next year, which he said would make it profitable once revenue lines reach scale. A 10x efficiency gain in a year is an aggressive target; inference costs have been falling across the industry as models and serving stacks improve, but whether that pace is achievable for a product with heavy agentic workloads is an open question.
Muse is also positioned at the center of Meta's smart glasses strategy. Reality Labs, the division behind glasses, VR and the metaverse, continues to lose billions, though the company expects those losses to narrow from $19.1 billion in 2025 to $15 billion this year.
Zuckerberg additionally teased a forthcoming "Muse network," describing it as moving Muse from a single-player experience to one where a user's Muses can work with those of friends and family. In that vision, buying something would not mean navigating to a business's website but talking to that business's agent. This suggests Meta is aiming at an interoperability layer between personal assistants and merchant-side agents, an area where standards and trust boundaries are still unsettled.
Why it matters: For product teams, Meta's numbers are a signal that consumers will let an AI agent touch money, calendars and inboxes at meaningful scale, which raises the bar for secure execution and identity in agent design. For developers, the shopping and productivity usage patterns plus the promised cost reductions suggest agent platforms will compete on both trust infrastructure and inference economics. Whether the per-user revenue target and the 10x efficiency goal hold is, by inference, the central question for anyone building on or against Muse.
Based on reporting from the original publisher. Visit the source for full context and later updates.
Publisher excerpt
Meta Muse users are adopting the app for shopping and personal productivity, the company said during its Q3 earnings call. The AI app now has over 10M users.