Meta Connect 2026: Muse Aims to Be the Non-Amazon Storefront
Meta unveiled $449 Ray-Ban Meta Gen 3 glasses and $1,299.99 VR glasses at Connect, but its merchant list tells more about its plans. Muse is signing up the retailers that compete with Amazon, and Meta is building the glasses to reach shoppers without going through Apple or Google.
The hardware at Meta’s Connect matters because it gives the Muse assistant more places to live. The business model sits in the list of companies Muse now works with, and in the one company that refused to join.
Meta unveiled Ray-Ban Meta Gen 3 at $449 and Meta VR Glasses at $1,299.99 for release in spring 2027. It also previewed Muse Charm, a keychain device for voice interaction with the assistant due to ship in December, and an audio-focused Ray-Ban model.
Muse is integrating Walmart, Best Buy, Gap, American Eagle, Dick’s Sporting Goods, Fanatics, Michael Kors, Sephora, Ulta and Wayfair, with PayPal and Shop Pay handling payment. Expedia is due to join, and further integrations cover Instacart, Notion, Granola, GitHub and Box.
Amazon began blocking Muse from shopping on its platform on September 20, after Meta declined to remove the bot.
The Amazon block is the strategy
Read those two facts together. Almost every retailer on Muse’s list competes with Amazon for the same customer. Walmart is Amazon’s largest general-merchandise rival. Best Buy fights it on electronics, Wayfair on furniture, Sephora and Ulta on beauty, Dick’s and Fanatics on sporting goods. Shop Pay is the checkout of the Shopify merchants that spend heavily to avoid depending on Amazon.
These companies have the same problem. Amazon owns the default starting point for online product searches in the US, and it charges merchants for visibility through its advertising business. For them, a third-party assistant that sends customers to their own checkouts is worth paying for, provided it doesn’t become a second Amazon.
Amazon’s refusal shows it understands the threat. An agent that sits between a retailer and a customer can redirect discovery, own the relationship and eventually charge for access. Amazon has its own assistant and no reason to let Meta’s stand between it and its customers. The rest of retail has every reason to welcome one.
Those same retailers are not signing exclusively with Meta. Walmart put about 200,000 items into ChatGPT’s Instant Checkout in November 2025, then replaced it in March with its own Sparky agent running inside ChatGPT. It said in-chat purchases had converted at about a third of the rate of click-throughs to its own site. Sparky is also available in Google’s Gemini, and Sephora and Best Buy have signed up to both OpenAI’s and Google’s agentic-commerce protocols. Every large non-Amazon retailer is spreading its bets across agents and keeping control of its own checkout.
That changes how Meta can win. It will not win by owning the checkout, because the retailers will not give it up. It can win the default through two things its rivals lack. The first is the glasses, which put Muse in front of a shopper without a phone, a browser or an app store in between. The second is Meta’s ad auction, which already sells purchase intent to many of the same merchants. Muse will not be the universal shopping agent. It could become the default starting point for shopping that is not on Amazon, and that is a large enough market to matter.
Sizing the prize
Mark Zuckerberg has described the monetisation plainly: Muse stays free to consumers, and Meta collects a small fee on transactions it facilitates. Rough numbers show how far that fee alone goes.
US online retail sales were about $1.23 trillion last year. Suppose Muse eventually facilitates $50 billion of purchases, roughly 4% of that market and an ambitious target for a new interface. A 2% fee would earn Meta about $1 billion a year. Meta’s revenue passed $200 billion last year, nearly all of it from advertising. On any plausible timeline, the transaction fee is a rounding error.
The bigger opportunity is advertising tied to purchase intent. Amazon earned more than $56 billion from advertising in 2024, most of it from merchants paying to appear when customers are ready to buy. An assistant that handles shopping requests for Walmart, Best Buy and Sephora customers produces exactly that signal, and Meta already runs the world’s second-largest ad auction. Once Muse routes real purchases, the fee becomes an entry ticket and sponsored placement inside the agent becomes the product.
Amazon also sets the limit on what merchants will tolerate. Third-party sellers on Amazon typically pay referral fees of around 15% of the sale price in most categories, before they spend anything on advertising. That is the benchmark Meta is competing against. Merchants will pay Muse a low single-digit fee plus sponsored placement to escape that cost. They will not pay Meta an equivalent cost under a different name, and the retailers on Muse’s list are large enough to leave if Meta tries to charge one.
Downloads are here; transactions are not
The distribution has started strongly. Sensor Tower data reported by Reuters showed Muse overtaking ChatGPT as the most downloaded free app in the US and Canada, with about 2.8 million downloads in its first two weeks. The missing metric is how many of those users let the assistant complete a purchase rather than answer a question.
The glasses matter here because they bypass the platform owners. A user wearing Meta’s camera, microphones and speakers can call up Muse without opening an app controlled by Apple or Google, whose operating systems otherwise decide how assistants reach shoppers. Reuters reported that Meta sold more than seven million smart glasses in the prior year and held roughly three-quarters of the category. That installed base gives Muse a distribution channel no other shopping agent has.
Privacy is the constraint on that channel. Meta says covering or disabling the capture LED disables the camera on current glasses, and IPVM, an independent testing group, reported on September 2 that a firmware update now stops recording within about five seconds when the light is covered. But on September 10 the Hamburg Commissioner for Data Protection and Freedom of Information published a lab report on first-generation Ray-Ban Meta glasses. It found the capture LED “not easily visible”, particularly outdoors and in direct sunlight. France’s data-protection regulator, the CNIL, issued an action plan and best-practice guidance on connected glasses in May. Commerce requires the glasses to be worn in shops and at home, not just bought, so social acceptance matters as much as specifications.
The checkpoint
Meta reports third-quarter results next month. The test is whether it discloses any Muse transaction metric, such as purchases completed, gross merchandise value or active merchants. A company confident of the storefront thesis will want investors to see early numbers. Silence would suggest Muse is still mostly answering questions.
Our view: Muse will not dent Amazon’s share within two years, but it has a credible route to becoming the default shopping agent for the rest of US retail. If Meta shows transaction data by early 2027 and adds a second tier of large merchants, commerce advertising becomes a new revenue line that deserves its own segment disclosure. Until then, the glasses are distribution waiting for a business to carry.