Back in April we wrote that the market still thinks META is an ad company. We called Muse Spark a force multiplier across every revenue line and said the consensus was treating the AI spending as a cost center. Five months later Meta shipped the consumer product built on that model, and the numbers are coming in faster than anything we put in the original thesis.
Muse launched September 8. By September 18 it was the #1 free app on the U.S. App Store, and it has stayed at or near the top of the iOS charts for roughly three weeks since. The latest daily active user estimates we're tracking put it around 2.2 million DAU. For an app that is only live in two countries and only open to adults, that is a serious number.
How It Compares to ChatGPT's Launch
Every AI app launch gets measured against ChatGPT's mobile debut in 2023, which was the fastest consumer adoption curve anyone had seen. Muse's early numbers beat it.
| First ~12 days, U.S. | Muse | ChatGPT |
|---|---|---|
| iOS downloads (U.S. + Canada) | ~1.8M | ~1.3M |
| U.S. mobile daily actives | ~642K | ~231K |
| Avg. daily download growth | 55% | 24% |
These are estimates from outside trackers like Apptopia and Sensor Tower, not Meta's internal numbers. Meta hasn't published official figures. But every tracker points the same direction and the gap is wide. Muse passed 5 million downloads and roughly 3 million weekly actives in about three weeks. Only a couple dozen apps in history have hit 5 million downloads that quickly.
Going from roughly 642,000 daily users at day twelve to an estimated 2.2 million a few weeks later suggests the app is still growing, not fading after a launch spike.
The House Ads Argument
The bear response we keep hearing is that Meta forced this. Muse got the majority of Meta's internal ad slots within ten days of launch, ahead of Facebook, Instagram and WhatsApp themselves. Meta also bought placements on Reddit, TikTok and YouTube.
We see that as a strength. No other company can do it. OpenAI had to earn every download. Google has to work around its own search business to push Gemini. Meta has more than 3.6 billion daily users across its apps and can point that attention at one product whenever it wants. In April we said every new Meta product gets instant distribution to an audience no company in history has assembled. Muse is a clear example.
The ads also aren't carrying it alone. Sensor Tower's read is that paid ads were only a small share of Muse's impressions in the first stretch after launch. Apptopia found that over 95% of Muse users also use Facebook. Meta didn't have to buy a new audience. Its existing users tried Muse because it showed up in apps they already use every day.
Why an Agent Is Different From a Chatbot
Muse isn't a chat window. It's an agent. It books reservations, manages email, handles subscriptions, sends messages and makes purchases for you. It runs on its own virtual machine and keeps working after you close the app. It connects to Facebook, Instagram and WhatsApp, and Meta has already announced Mac computer use, a dedicated Muse email address, avatar video chat, more connectors and smart glasses integration.
That matters for the stock because of where the money is. A chatbot answers questions. An agent sits between the user and the purchase. Zuckerberg has said Muse will stay free and make money by taking a small fee on the transactions it completes. Meta already has a Walmart partnership, and Shopify opened agentic checkout across its merchant network.
This connects to the Instagram Shop thesis from April, where we argued Instagram would beat TikTok Shop because trust converts better than attention. Now add an agent that already knows your Instagram follows, your WhatsApp threads and your purchase history, and can do the buying for you. That puts Meta in a position to take a cut of a large share of online purchases, which is a much bigger opportunity than chatbot subscriptions.
The Monetization Math
Muse has three tiers: a free tier with weekly usage limits, Power at $20 a month and Maximum at $100 a month. Meta's AI leadership has reportedly said most users will stay free and that the paid tiers are mainly there to cover compute costs.
Analysts are putting numbers on it anyway. One widely circulated estimate pegs 10 million paying users at roughly $2.4 billion a year. Wedbush has floated about $5 billion in subscription revenue by 2027. JPMorgan upgraded the stock two days after launch and said Muse could become the most widely used consumer AI app since ChatGPT.
We think the subscription number matters least. Meta did nearly $61 billion in revenue in Q2 alone, about 98% of it from ads. A few billion in subscriptions is nice. The bigger opportunities are the transaction fees and what the agent's data on user intent does for ad targeting across all of Meta's apps. As far as we can tell, nobody is modeling either one yet.
What Could Break This
We always lay out the bear case honestly.
Retention. Downloads are easy when you own the ad inventory. The real test is whether people come back every day. Early estimates showed open rates well below ChatGPT and Instagram. The climb toward 2.2 million DAU is the best counter we've seen, but this is the number to watch every week.
Trust. An Oppenheimer survey found only 8% of Americans would trust Meta to store passwords for other apps, compared with 30% for Google. An agent needs access to your accounts to work, so this is a real headwind, and Meta's history doesn't help.
Other platforms pushing back. Amazon blocked Muse from its shopping site before Meta Connect, citing security concerns. Meta says Muse never sees passwords or payment methods. Expect more platforms to try the same thing. Those that do risk losing customers who would rather shop through the agent.
Reliability. Reuters reported Muse had real problems in testing before launch, including a case where it exposed a user's private photos. One bad incident at scale could slow adoption quickly.
Geography. It's only available in the U.S. and Canada right now. That also means everything above is happening in just two countries, and the international rollout is still ahead.
Where We Stand
We added META at $605 in April after the capex scare. The stock ran through September on Muse. It rose 6.5% the day after launch and more than 11% in a single session ahead of Connect, its biggest daily gain since April 2025, and traded into the high $700s by late September.
We're not chasing it here and we're not selling. Our thesis was never about Muse reaching #1 on the App Store. It was that Meta's distribution turns the products it ships into real businesses, while the market keeps valuing the company as only an ad network. Muse is the first strong evidence of that on the AI side.
We're holding the $605 shares. If the stock sells off on another capex scare, we'll look to add again like we did in April.