Inside Meta Connect 2026: How Much Revenue Can Muse AI Bring for Meta Stock?

Meta put price tags on new glasses at Connect 2026. The more consequential product, Muse, came with no fresh revenue number. Mark Zuckerberg showed an agent that can see through glasses, speak naturally and complete more tasks across shopping and work. 

For investors, the intriguing question is whether all that extra reach can turn a popular AI assistant into a business large enough to matter beside Meta's advertising empire.

Let's break down what Meta actually unveiled, which announcements have a path to revenue, and how much activity Muse would need to generate a meaningful financial result. We will use disclosed prices and company figures where possible, then label every assumption in the parts that remain unproven.

What's Covered

• The main Muse, AI glasses and VR announcements from Meta Connect 2026.
• What Zuckerberg's vision changes for Muse's reach and everyday usefulness.
• The difference between a product integration, a purchase made through Muse and revenue Meta can recognise.
• The subscriber and shopping volumes needed for $1 billion a year.
• Why the cost of running live voice and video could change the profit picture.
• The Meta figures and milestones investors can track after the keynote.

Meta Connect 2026: What Did Meta Announce for Muse AI?

The September 23 keynote made Muse the common thread between Meta's software and hardware. Muse itself was introduced on September 8, so Connect was not its launch. It was Meta's demonstration of what happens when the agent moves beyond an app and gets access to more of a person's day.

Sources: Meta's full announcement, AI glasses specifications and prices and newsroom recap.

The distinction between Meta AI and Muse matters here. Some glasses already use Meta's AI models to answer questions. Zuckerberg said in Meta's Q2 earnings call that the company's own Meta Glasses shipped with Muse Spark, the model, out of the box. Connect's new promise is access to Muse, the personal agent, on AI glasses in the coming months. A model answering a question about a street sign and an agent remembering to act on that sign later are different products. Meta has not published a complete compatibility list or a Muse-on-glasses paid conversion figure.

The VR headset, new games, sports and cinema partnerships were also part of Connect. They matter to Reality Labs, but counting them as evidence of Muse revenue would muddle the central question. The same applies to hearing enhancement. Its $149.99 price is useful evidence that Meta is willing to charge for glasses software, yet that feature is distinct from Muse.

On the first full US trading day after the keynote, Meta shares closed September 24 at $777.59, up 4.5%, according to Yahoo Finance's. That is a measure of market enthusiasm, not a disclosed Muse sales figure, and the day's entire price move cannot be assigned to one announcement.

What Did Mark Zuckerberg Announce About Muse AI at Meta Connect 2026?

Before Connect, a user could message Muse in its app or WhatsApp, ask it to work through a task, and approve sensitive actions such as payments. The fuller explanation of those original capabilities is in our earlier Muse and Meta stock analysis. The new point from Connect is access at the moment a task appears.

Imagine seeing a school-supplies list on a noticeboard. On a phone, you might photograph it, open an app and explain the request. With compatible glasses and the promised integration, Muse could see the list when asked, search for the items and come back for approval before a purchase. Meta described similar examples with a product on a shelf and a flyer on a wall. That cuts steps out of a task. It does not tell us how many tasks people will complete or whether a merchant will pay Meta for one.

The connector expansion is just as important as the glasses. A personal agent can be clever and still fail if it cannot reach the retailer, travel service or work app the user needs. Meta said Muse began with dozens of partners and access to the Shopify catalogue. At Connect it named Walmart, Gap, Sephora, Expedia, Instacart, Notion, GitHub, Box and others. Shopify's merchant help page describes checkout on Meta surfaces including Muse when available. A connector is permission to enter a store and help a customer. It is not evidence that Meta owns the store's till.

There is a useful limit to this strategy. Amazon has blocked Muse from shopping on its platform, citing concerns about unauthorised access. Adding willing partners improves coverage, but Meta does not get universal access simply by making Muse more capable. For a shopping agent, the missing store can be the very store a customer wanted.

The Mac update gives Muse a different route to repeat use. Meta says that, with permission, it can drive apps on a Mac and continue after a person walks away. Muse's own email address could help it coordinate tasks without a user constantly copying messages between apps. Neither was accompanied by usage or billing data. Their financial importance depends on whether people hand Muse valuable work often enough to keep paying.

Can Muse Make Money for Meta Today?

Yes, there are established ways to charge for the product. Meta says Muse is free for most everyday needs, with paid plans for heavier use. Its Power plan costs $20 a month and Maximum costs $100 a month, according to TechCrunch's reporting on the launch. Muse first rolled out in the US; a wider release, including India, should not be assumed from Connect's international glasses announcements. Meta has not disclosed the number of paying Muse users or a separate Muse revenue line.

Meta also charges developers for access to models through its Meta Model API. Standard Muse Spark pricing is $1.25 per million input tokens and $4.25 per million output tokens, with separate prices for other services and a discounted tier that permits training use. Those tariffs prove a real model business exists. API revenue is not the same as revenue from the Muse consumer agent, and Connect did not disclose how much API usage came from its new announcements.

There is a third, more uncertain path: transactions. Muse can help a person choose goods and reach checkout through partners. But neither Meta's Connect announcement nor its connector platform page states a standard commission Meta receives on a Muse-facilitated order. PayPal, Stripe's Link and Shop Pay being present does not answer which party collects a payment fee. If a shopper spends $100 with a retailer, that $100 belongs in the retailer's gross sales, not Meta's revenue. Meta would recognise only whatever contractual amount it earns, if any.

Finally, there is hardware. EssilorLuxottica said more than 7 million AI glasses were sold in 2025. That is evidence of a real device category, and Connect expands the range. It is not 7 million Muse users: sales are not active wearers, not every device has a confirmed agent rollout, and some wearers may already use Muse on a phone. Meta's Q2 Reality Labs revenue was $431 million across its broader hardware, software and content business. Multiplying every jointly sold pair by its retail price would misstate what Meta reports.

How Much Would It Take for Muse to Add $1 Billion a Year?

Here is the cleanest way to read Connect financially. Follow a task through four steps: 

• a person uses Muse, 
• Muse completes something valuable, 
• a customer or partner pays, 
• and Meta keeps more than the task costs to serve. 

The keynote made the first two steps more plausible. It gave no public conversion, fee or cost data for the final two.

These are alternative hurdles, not four revenues to add together. The subscription figure is gross before discounts, refunds, sales tax, fees and serving costs. The shopping cases show exactly how much the answer depends on a commercial term Meta has not shared. The hardware example deliberately shows why a tempting calculation fails. If an investor takes the $1,299.99 VR glasses price and multiplies it by a sales guess, the same problem remains: units are unknown, and the product is not Muse itself.

To see what Connect specifically might add, think in terms of new paid users caused by these changes, rather than everyone who opens Muse on a second device. One million additional people who become regular Muse users through glasses or connectors would produce the following gross annual subscription revenue if they paid for a full year at $20 a month:

The one million new regular users and conversion rates are illustrative assumptions, not Meta guidance. Existing Muse subscribers who simply start wearing glasses add no second subscription. A six-month average paid life would halve the table's annual amounts. Meta would also need to reveal which users arrived because of the Connect features before the lift could fairly be attributed to the event.

Shopping offers a separate sensitivity test. Suppose one million regular users complete two $75 purchases per year through Muse. That would create $150 million of merchandise value for retailers. At an assumed 1% fee retained by Meta, the agent would produce $1.5 million in annual Meta revenue. If Meta retains no transaction fee, its direct transaction revenue is zero. These amounts do not include any possible advertising or merchant-service payments, because no Connect-specific terms were disclosed. The test explains why a long list of retailers is useful product news but not, by itself, a billion-dollar earnings model.

Our conclusion from the math: a $1 billion annual direct Muse business is possible at Meta's distribution scale, but Connect did not establish a credible calendar date or enough inputs for a precise forecast. The most defensible model today is a sensitivity model that can be filled in as paid-user, repeat-task and partner-payment data emerge.

Muse AI Costs: What Could Higher Usage Cost Meta?

Revenue matters only after the service works economically. Muse is not merely returning a short answer. Meta's launch description says it gives each user a dedicated secure virtual machine, a browser and a separate safety system. Background tasks can keep running. Connect adds live conversation and, potentially, continuous generated video. Each improvement can make the product more useful while also increasing the amount of computing it consumes.

Meta's research post on Muse Realtime Avatar reported an approximately 870-millisecond delay to the first synchronised voice and video response. Its optimisations increased serving capacity eightfold versus its own stated baseline, reaching 12 concurrent video sessions on one GB200 in the described test. That is promising engineering evidence. It is not an eightfold reduction in Muse's total cost, a measured product margin or proof that millions of people will pay for video avatars. Meta also notes that its research examples do not all represent avatars available in the app.

A simple unit check helps. At $20 a month, a Power user brings in $240 a year at the listed price. If the total variable cost of that user, including computing, payment fees and support, were $5 a month, the annual contribution before fixed costs would be $180. At $10 it would be $120; at $15 it would be $60. These cost numbers are hypothetical. Meta has not reported the actual figure or enough usage detail to estimate it reliably. Heavy users on flat monthly plans can produce the most revenue and the most serving expense. A great demo could make a service more expensive to run unless efficiency improves alongside usage.

That is why it would be wrong to compare one hypothetical Muse revenue total with all of Meta's AI capital spending and declare the project profitable or unprofitable. Meta expects $130 billion to $145 billion of capital expenditure, including finance lease principal payments, for 2026. This supports ads, training, research, new products and other infrastructure, not Muse alone. In Q2, Meta generated $60.80 billion in revenue but just $784 million of free cash flow after heavy investment. Those are company-wide figures, not a Muse income statement. Meta's official results make both the opportunity and the cost of waiting visible.

Could Muse Affect Meta's Ads Business More Than Its Own Sales?

This is the investor question that a subscription-only forecast misses. Meta earned $59.36 billion in Q2 2026 advertising revenue. One percent of that quarterly amount is about $594 million, or about $2.37 billion if that same quarterly amount were repeated four times. The latter is a scale comparison, not a forecast. It is roughly equal to a full year of revenue from 9.9 million users paying $20 a month.

If an agent helps people return to Meta's apps, increases useful activity or gives businesses better ways to complete a sale, its indirect value could exceed the subscription line. There is an opposite possibility: someone who asks Muse to find an item might spend less time scrolling past ads. No public figures let us calculate either effect today. Crucially, Meta says Muse conversations and data in its secure virtual machine are not shared with Meta's ad systems. The convenient story that every Muse interaction instantly improves ad targeting contradicts Meta's stated product boundary.

Meta One adds another source of confusion. Its September subscription announcement said it had 15 million subscriptions and trials to date and had launched more than 50 features, with paid options that include more use of certain AI tools. These are not 15 million Muse subscribers. Meta One's $7.99 Core bundle and $19.99 Premium bundle also differ from Muse's $20 Power and $100 Maximum plans. It would be double counting to add Meta One members to an assumed Muse payer count without knowing how the products overlap or how Meta records them.

What Should Investors Track After Meta Connect?

The event improved distribution and capability. The following disclosures would tell us whether those improvements have reached the financial statements.

The last line needs care. Meta reported $1.007 billion of Family of Apps other revenue and $431 million of Reality Labs revenue in Q2. On the Q2 call, management said the former was driven chiefly by WhatsApp paid messaging and subscriptions, while stronger AI glasses helped the latter. 

Those totals can signal progress, but neither is a clean Muse counter. Business agents are also a different product: management said more than one million businesses used Meta Business Agents weekly in Q2, before the Connect announcements. That is a promising distribution fact, not a disclosed Muse revenue number.

Author’s View: A Stronger Product Case, an Open Revenue Case

Meta Connect made the consumer story more convincing. Glasses can meet a task when it appears. Connectors give Muse more places to act. Mac control makes it useful when work does not fit neatly inside one app. Meta already has a huge network to introduce the product, and the 7-million-plus AI glasses sold in 2025 show that wearable distribution is more than a slide in a presentation. These are real advantages, even though Muse's initial US release limits any near-term global-user arithmetic.

But a platform is only as valuable as the work users trust it to finish and the economics Meta retains. Connect did not announce a new Muse paid-user count, a merchant fee, glasses-driven conversion or a cost per successful task. Its most visually striking advances, such as real-time avatars and a pocket charm, could deepen engagement without contributing much near-term profit. Retailers can also restrict agent access, and more automatic actions increase the stakes of errors and privacy concerns.

Our assessment is deliberately firmer than the headline excitement: Connect increased the probability that Muse becomes a daily habit, but it did not validate a multibillion-dollar annual Muse revenue estimate. The first meaningful proof would be sustained paying users or reported partner income alongside acceptable service costs. Until then, Meta's advertising business remains the earnings engine, while Muse is a potentially powerful new route to customers. The product now has more doors. Investors still need to see which doors lead to Meta's cash register.