Meta has spent 20 years as an advertising company, and advertising was nearly all of its $60.8 billion in second-quarter revenue this year. On September 23 at Connect, Mark Zuckerberg said Muse would be something different. No ads inside the agent. Instead, Meta would "profit by taking a small fee from transactions" Muse completes on your behalf, with the free tier kept generous to get the agent into as many hands as possible.
That sentence is the whole business model, as far as Meta has stated it. Here is what is confirmed behind it, what is still missing, and why the missing parts are the interesting parts.
What Meta has actually said
The quote, from the Connect keynote: "Muse also has a novel business model. We believe that Muse will make you money. We are standing behind this by making Muse free for a huge number of tokens with the expectation that over time we will profit by taking a small fee from transactions."
Two things stand out. First, "no ads inside Muse" is the policy as stated; Zuckerberg did not utter the word "advertising" even once in that keynote, and Meta's launch materials said there is no advertising within the agent. Second, the fee comes from merchants, not from you. TheStreet's reporting on the keynote pinned the consumer question: users do not pay the fee; the merchant absorbs it.
The model is: Muse shops for you, Muse checks out for you, and Meta takes a cut of the sale from the store. It only works if the stores agree to be checked out on, which is why the next section is really the business plan.
The commerce rails already built
Meta spent the first three weeks of Muse's life wiring up the checkout infrastructure:
Payments. Link by Stripe, the wallet Meta started with at launch, creates a one-time-use card number so the merchant never receives your actual card details. Shopify's Shop Pay went live in Muse on September 24. PayPal was announced on September 22 but is not available inside the agent yet.
Retailers. Walmart, Best Buy, Gap, Sephora, Ulta, Wayfair, Michael Kors, American Eagle Outfitters, DICK'S Sporting Goods, and Fanatics were named as connectors at Connect. Travel and groceries came along too: Expedia and Instacart. Shopify is the big one underneath: Meta said Muse had access to Shopify's entire catalog at launch, which means thousands of independent stores are reachable through one partnership.
The receipt layer. On October 9, Rezolve Ai launched RezolvePay, infrastructure that lets a consumer agent like Muse transact with a merchant-side agent and complete a purchase. The launch demo showed Muse placing a coffee-shop order for in-store pickup, and the receipt identifies Muse as the agent that placed the order. That matters: merchants can see which agent channel the sale came through, which is the beginning of agent-commerce accounting.
How checkout works in practice: you ask Muse to find something, it returns listings inside the app or sends you to the retailer, and for purchases it completes inside Muse, it prepares the order and shows an approval card with the order details before anything is finalized. Nothing is charged without your explicit confirmation.
Who said yes, who said no, who is hedging
The door list is where the business model becomes visible.
Said yes: Shopify (full catalog plus Shop Pay), Walmart, Best Buy, Gap, Sephora, Wayfair, Expedia, Instacart, PayPal (announced). Rezolve Ai built the first purpose-built "welcomed" rail on October 9.
Said no: Amazon. In mid-September Amazon blocked Muse from browsing and buying on its site, accusing the agent of violating its terms around credentials and account data. Insurify, an insurance-shopping site, has since joined the blocked row. The full merchant-refusal tracker is its own post on this site.
Hedging: Ticketmaster. Muse can surface Ticketmaster listings, but the purchase stays inside Ticketmaster's own marketplace. "Fans remain in complete control of their purchase decisions, as final purchases are all made within Ticketmaster's secure marketplace," a Ticketmaster spokesperson told CNBC.
The yes/no split is the fee logic made legible. Walmart said yes to a fee coming out of the merchant's margin because a new sales channel is worth it. Amazon said no because it has spent decades building the world's most sophisticated shopping environment, and an outside agent that browses, extracts the answer, and leaves takes the value of that environment without entering Amazon's monetization. The fight is over who sits between intent and transaction, and that is the most valuable seat in commerce.
My read: the fee is not really the product. The product is the seat. Whoever owns the agent owns the relationship with the shopper, and Meta just announced it intends to collect rent on that seat. Watch the door list. Every "welcome" is a new tax base; every "blocked" is a lawsuit-shaped negotiation.
What is missing: the price sheet
Nothing in this section is a number Meta has published. The transaction-fee plan is still a speech, not a price sheet. Say that plainly, because a lot of coverage does not.
What Meta has not disclosed:
- No standard commission. Meta has stated no percentage, no fee schedule, no "per-order" figure anywhere, not at Connect, not on its platform pages.
- No count of paying users. Meta has disclosed no number of subscribers and no separate revenue line for Muse.
- No definition of "small." The word is doing all the work, and nobody outside Meta knows its size.
Until those numbers appear, the entire plan is a promise from a keynote. Every downstream figure, including the analyst math below, is a guess built on that promise.
The second revenue lane: subscriptions
Subscriptions are the other, quieter half of the playbook, and they have a distinctive shape: the tiers buy volume, not features.
| Tier | Price | Weekly tokens |
|---|---|---|
| Free | $0 | ~100 million (per Zuckerberg's own Threads post; Meta's help page says only "limited and refreshes") |
| Power | $20/month | 500 million |
| Maximum | $100/month | 3 billion |
Multiple independent guides have now verified the same thing: the plans share the same access channels and the same features. Power and Maximum buy more usage for busier delegation, not exclusive capabilities. The Motley Fool's October framing puts the strategy plainly: the free tier carries a relatively low monthly usage cap, and Meta is betting that enough people find the agent useful that they burn through it and start paying. The free tier is a customer-acquisition funnel, not charity.
One pricing wrinkle worth knowing: Power costs $16 a month and Maximum $80 on the web, versus $20 and $100 inside the iPhone app, with identical token allowances either way. That is almost certainly the Apple in-app commission passed through, though Meta has not said so.
The ad contradiction
The trickiest row in the playbook is the one Meta is not supposed to have. Meta says Muse's conversations and virtual-machine data are not shared with its ad systems. But the wall has leaks, and Meta's own materials admit two of them:
- Meta says websites see Muse's browsing as the user's own activity, which means a store Muse visits can retarget that user later, "the same as if you had visited it yourself."
- Meta's own safety guidance concedes that the agent's actions could "indirectly influence" the ads you see.
Wall Street has noticed the gap. Deutsche Bank's analyst sees Muse driving up to 8 percent of Meta's revenue by 2030 through enhanced ad targeting, with a bull-case figure around $36 billion (base case $2.4 billion, middle case $10.7 billion). Truist's base case is $28.5 billion. Citi's is about $27 billion, split roughly $23 billion from transactions and $4.5 billion from subscriptions.
These are analyst estimates, not Meta's, and they rest on the undisclosed fee math. The interesting tension is the one between them: Meta says no Muse data flows to ads, while the Street models the agent's value as better ads anyway. If the targeting value turns out to be real, it becomes the cheapest possible defense of the whole enterprise: the fee funds the agent, the data sharpens the ads, and the ads remain the business.
Treat every revenue number in this section as a forecast from someone who does not work at Meta. Meta has published exactly zero dollars of Muse revenue and zero details of the fee. The analysts are modeling a speech.
Why this matters for you
If you are a shopper: you do not pay the fee. But two things follow from the model. First, an agent that earns a cut of what it buys benefits from every second shaved off the path between a thought and a purchase, so keep reading those approval cards. Second, the "who said no" list is your map of where the agent cannot reach: if you want Muse to price-shop Amazon, it cannot.
If you are a merchant: the question is whether to welcome the agent (like Shopify and Rezolve's merchants) or block it (like Amazon). The welcome math says a new sales channel with identified-agent receipts is worth a cut of margin. The block math says letting an outside agent stand between your storefront and your customer surrenders the most valuable position in commerce. There is no consensus yet, and Q3 earnings on October 28 is the next datapoint where Meta might have to say something concrete.
If you are just watching Meta: the pivot is real even if the numbers are not. A company that earned 98 percent of its revenue from ads just announced, onstage, that its flagship AI product will make money without them. Whether the fee is 1 percent or 10 percent will decide how aggressive Meta is about pushing agent checkout into every store, and that will decide what shopping looks like in two years.
Quick answers
How does Meta make money from Muse?
Per Meta's own keynote: a "small fee" on transactions the agent completes, paid by merchants, not by users. No ads inside the agent. Subscription tiers (Power $20, Maximum $100) buy more weekly usage, not extra features.
Do I pay extra when Muse buys something for me?
No. The fee comes from the merchant's side. What you see at checkout is the price the store set; Meta's cut is between Meta and the store.
Why did Amazon block Muse?
Amazon blocked Muse from browsing and buying on its site in mid-September, accusing the agent of violating its terms around credentials and account data. Underneath the security argument is the commercial one: Amazon does not want an outside agent standing between shoppers and its storefront.
Can Muse check out anywhere?
No, only at partner retailers so far: Walmart, Best Buy, Gap, Sephora, Wayfair, and others via connectors, plus Shopify's full catalog, with Stripe Link, Shop Pay, and (announced) PayPal as payment rails. Ticketmaster keeps purchases inside its own marketplace.
Has Meta said how big the fee is?
No. Meta has published no percentage, no fee schedule, and no Muse revenue figures at all. "Small" is the only word on the record.
Frequently Asked Questions
Q: Does Muse show ads?
A: Meta says no: there is no advertising inside the agent, and Muse's conversation and virtual-machine data are not shared with Meta's ad systems. But Meta also says websites see Muse's browsing as your own activity, and its safety guidance concedes the agent's actions could "indirectly influence" the ads you see. The wall covers conversations, not actions.
Q: Why would a merchant agree to pay Meta a fee?
A: For a new sales channel and, increasingly, identified-agent receipts (see RezolvePay, October 2026) that let the merchant see which agent drove the order. It is the same logic as credit-card interchange: margin in exchange for the customer showing up.
Q: Is the subscription fee the same thing as the transaction fee?
A: No. The subscription tiers (free, Power, Maximum) pay for weekly usage volume; they buy more agent time, not extra features. The transaction fee is a separate, merchant-side cut on purchases Muse completes. Meta has published exact terms for the first and nothing for the second.
Q: How much revenue could Muse generate?
A: Analysts, not Meta: Deutsche Bank's bull case is roughly $36 billion by 2030 (base $2.4 billion); Truist's base case is $28.5 billion; Citi's is about $27 billion, mostly transactions. Meta itself has published no revenue figures, no subscriber count, and no fee size.
Q: What changes the picture from here?
A: Meta's Q3 earnings on October 28, the first report since launch; any published fee schedule; any move on the blocked/welcomed list; and whether Meta ever opens the agent to advertising. Until one of those lands, the plan is a keynote sentence, not a price sheet.
Comments