Meta Muse AI and Connect 2026: What It Means for META Stock
Meta Muse AI and Connect 2026: What It Means for META Stock
For most of 2026, the debate around Meta Platforms has been a debate about money going out. The company guided to between $130 billion and $145 billion of capital expenditure this year, and in the second quarter that spending left it with just $784 million of free cash flow. When Meta reported those numbers in late July, the shares fell sharply. Investors wanted to know what, exactly, all of that compute was for.
September gave the first real answer. On 8 September, Meta launched Muse, a personal AI agent that plans and carries out tasks rather than simply answering questions. Two weeks later, at Meta Connect 2026, Mark Zuckerberg presented nearly every new device on stage as a way into Muse: new Ray-Ban Meta glasses, an ultralight VR headset, and a keychain-sized gadget called the Muse Charm. The app went to number one on the US App Store, analysts raised their targets, and the stock rallied hard for the month.
This article works through what Meta Muse AI actually is, how Meta can make money from it, what Connect 2026 changed about the hardware strategy, and whether the current valuation already prices in the good news. The honest framing matters here: Muse is an early and promising product sitting on top of the most expensive infrastructure bet in Meta’s history, and those two facts have to be judged together.
1. Key Financial Metrics
| Metric | Value |
|---|---|
| Ticker | META (NASDAQ) |
| Share price (25 Sep 2026 close) | $751.66 |
| Market capitalisation | About $1.91 trillion |
| 52-week range | $520.26 to $779.82 |
| Revenue (TTM) | $228.25 billion |
| Q2 2026 revenue | $60.80 billion, up 28% YoY |
| Q2 2026 operating margin | 31% |
| Q2 2026 diluted EPS | $6.18 |
| Q2 2026 capex | $31.08 billion |
| Q2 2026 free cash flow | $784 million |
| 2026 capex guidance | $130 billion to $145 billion |
| 2026 total expense guidance | $165 billion to $169 billion |
| Cash and marketable securities | $90.26 billion |
| Long-term debt | $83.66 billion |
| P/E (TTM) / Forward P/E | 28.3 / 23.3 |
| Dividend | $2.10 per year (about 0.28% yield) |
| Analyst consensus | Strong Buy, average target about $761 |
Data as of 25 September 2026. Verify before publishing.
2. Company Overview: Where AI Fits at Meta
Meta runs the largest social media ecosystem in the world. Facebook, Instagram, WhatsApp, Messenger and Threads reached an average of 3.6 billion daily active people in June 2026. Almost all of that attention is monetised through advertising, which brought in about $59.4 billion in the second quarter alone.
The company reports two segments. Family of Apps is the profit engine, producing $60.37 billion of revenue and $23.39 billion of operating income in Q2 2026. Reality Labs, which builds glasses, headsets and the software behind them, generated $431 million of revenue and lost $4.62 billion in the same quarter.
AI is not a side project at Meta. It already runs the business in three places:
- Ad ranking and recommendations: deciding which post, Reel or ad each user sees
- Advertiser tools: Advantage+ automation and generative AI creative tools for businesses
- Consumer AI: the Meta AI assistant and, since September, the Muse agent
- Devices: AI glasses that act as an always-available interface for the assistant
What changed in 2026 is the fourth and fifth layer of that stack. Meta Superintelligence Labs, the unit led by chief AI officer Alexandr Wang, now builds its own frontier model family called Muse Spark. That model powers Meta AI and the new Muse agent, and it ships on Meta’s own glasses. The strategic question for shareholders is whether this consumer layer becomes a new business, or stays an expensive feature that protects the existing one.
3. What Muse Actually Is: The Technology Behind Meta Muse AI
From chatbot to agent
Most AI assistants answer a question and wait. Muse is designed to take a goal, form a plan, and carry it out. According to Meta’s launch announcement, it can send emails, book travel, fill out forms, shop and research projects. It keeps working after the user closes the app, and comes back when something changes or when it needs approval.
It also remembers preferences and makes suggestions without being asked. One example Meta gave is turning a saved recipe Reel into a shopping list. At Connect 2026 the company added voice and real-time video conversations, and a new model called Muse Realtime Avatar that lets users video chat with a personalised face for their agent.
The model underneath: Muse Spark
Muse runs on Muse Spark, a multimodal model built for agentic work. It was introduced in April 2026, and version 1.3 arrived on 2 September, days before the agent launched. The model is designed for long tasks, coding, messy inputs and real-world computer use, and supports a context window of around one million tokens.
Meta has also kept one foot in open source. A smaller open-weight sibling, Muse Glimmer, was released under the Apache 2.0 licence and is designed to run offline on a single consumer GPU. Muse Spark itself is proprietary, but it is sold to developers through the Meta Model API, which gives Meta a third route to monetise the model beyond ads and subscriptions.
How Meta is addressing trust
An agent that reads your email and holds your payment details is only useful if people trust it. That is a harder sell for Meta than for almost any other company, given its record, including a $5 billion FTC penalty in 2019 over privacy violations. Meta’s design choices are clearly built around that problem:
- Muse Secure VM: each user’s agent runs in a dedicated virtual machine with encrypted storage
- Sentinel agent: a separate agent must approve every action that goes out to the internet
- No password or card visibility: Muse does not see passwords or payment methods; checkout runs through Link by Stripe
- Ad separation: Meta states that Muse does not share conversations or VM data with its ad systems
- Roadmap: a Muse Confidential VM with end-to-end encryption is promised later this year
These are Meta’s own claims, and independent security experts have not yet tested them in depth. For investors, the ad separation promise is the most important line on that list, because it defines how Muse can and cannot be monetised.
What was added at Connect
The Connect keynote extended Muse well beyond a phone app:
- Muse on Meta’s AI glasses, activated by a wake word, arriving in the coming months
- Muse controlling Mac computers and running desktop applications on the user’s behalf
- A dedicated email address for each Muse, so users can copy it into threads
- Retail integrations with Best Buy, Gap, Sephora, Walmart, Wayfair, Expedia and Instacart
- Payment partnerships with Stripe, Shopify and PayPal
- A developer connector programme that drew more than 1,500 applications in under a week
The commerce list is the one to watch. It shows Meta positioning Muse as a buying agent, a space OpenAI entered earlier with agentic shopping inside ChatGPT.
4. Business Model: How Muse Could Make Money
The honest starting point
Muse is not yet a material revenue line. Meta has not disclosed Muse revenue or paying subscribers, and it will not show up as a separate number in the third quarter results. Any claim that Muse is already moving earnings is ahead of the evidence.
What Meta has built is a set of monetisation routes that do not depend on putting ads inside the agent. That matters because Meta has promised to keep Muse data out of its ad systems.
The revenue routes
- Consumer subscriptions: Muse is free for everyday use. Meta One Core at $7.99 per month adds more AI usage, and Meta One Premium at $19.99 per month adds expanded usage
- Creator and business plans: Meta One Essential ($14.99), Advanced ($49.99), Expert ($149) and Max ($499) per month, including a Business Agent for 24/7 customer replies on WhatsApp
- Model API: Muse Spark sold to developers through the Meta Model API
- Commerce: Muse completes purchases with major retailers, which could open transaction-based revenue over time, although Meta has not disclosed any fee terms
- Hardware: glasses and devices that ship with Muse Spark built in
Why ads still matter most
The indirect effect may be larger than any of the direct routes. Meta’s CFO Susan Li said on the Q2 call that ranking improvements drove an 8.3% increase in ad clicks and a 15.7% uplift in conversions on Facebook. Advantage+, Meta’s AI-driven campaign automation suite, reached an annual revenue run rate above $75 billion, and more than 9 million small businesses now use at least one AI creative tool.
Zuckerberg also said daily users of the Meta AI assistant rose 60% after Muse Spark was integrated. More engagement with Meta’s apps means more ad inventory. The flywheel investors are betting on is simple: better models improve the ad engine that pays for them, while the consumer products keep users inside Meta’s ecosystem.
Moat factors worth naming:
- Distribution: 3.6 billion daily users and the ability to promote Muse across Facebook and Instagram for free
- Data: years of interest and engagement signals that already power ad ranking
- Hardware lead: a volume lead in AI glasses, built with EssilorLuxottica
- Cash engine: a Family of Apps business that can fund losses elsewhere
5. Meta Connect 2026: The Hardware Strategy
Connect 2026 opened on 23 September. The clearest message from the keynote was that Meta no longer treats its devices as separate bets. Each one is a way to reach Muse.
The lineup
| Device | Price | Availability |
|---|---|---|
| Ray-Ban Meta (Gen 3) | From $449 | On sale now |
| Ray-Ban Meta Audio (no camera) | $349 | Ships 13 October |
| Meta Ray-Ban Display | $799 (US) | Expanding to UK, Canada and EU |
| Meta VR Glasses | $1,299.99 | Spring 2027 |
| Muse Charm | Not announced | Targeting the holiday season |
Prices and dates as announced at Connect 2026. Verify before publishing.
What each device is for
The Gen 3 Ray-Ban Meta glasses are the volume product, with a slimmer frame, a 12MP camera and a dedicated AI button. Ray-Ban Meta Audio is Meta’s first pair without a camera at all, a lower-cost and more privacy-friendly entry point that works as headphones plus a voice assistant.
Meta VR Glasses, reported for months as Project Phoenix, weigh about 100 grams on the face and move the battery and processing into a tethered puck. They are a premium product for 2027 rather than a near-term revenue driver.
The Muse Charm is the most revealing announcement. It is a palm-sized device with a small OLED screen, cameras, microphones and 5G, designed for people who want to talk to their agent without wearing glasses. Meta has not set a price, and Zuckerberg acknowledged that only a few units had been built so far. Its main strategic value is timing: Meta showed a dedicated AI agent device before OpenAI, which is widely expected to enter consumer hardware.
Is the hardware actually selling?
The best evidence comes from Meta’s manufacturing partner. EssilorLuxottica sold more than 7 million Ray-Ban and Oakley Meta glasses in 2025, more than triple the roughly 2 million units sold across 2023 and 2024 combined. Sales of Meta glasses nearly doubled again in the second quarter of 2026, and the two companies have discussed raising production capacity.
That is real traction. It is also small against Meta’s scale. Reality Labs revenue was $431 million in Q2 2026, up 16% year on year, against an operating loss of $4.62 billion. Glasses are growing fast from a low base, but the segment remains a heavy drain on group profits.
6. Financial Analysis: Meta Capex and the Free Cash Flow Squeeze
Q2 2026 in numbers
The core business is performing well. Revenue rose 28% to $60.80 billion, ahead of the roughly $60.2 billion analysts expected. Ad impressions grew 14% and the average price per ad rose 12%, so growth came from both more ads and more valuable ads.
Costs grew much faster than revenue. Total costs and expenses rose 55% to $42.03 billion, research and development alone reached $21.66 billion, and operating income fell 8% year on year to $18.78 billion. General and administrative expenses included $2.4 billion of legal charges. Diluted EPS of $6.18 missed consensus of about $7.14.
- Operating margin: 31% in Q2, down from the prior year
- Free cash flow: $784 million, against about $8.5 billion a year earlier
- Share buybacks: none in Q2 2026, against $10.17 billion in Q2 2025
The guidance
Management guided third quarter revenue to between $61 billion and $64 billion. Full-year expense guidance was raised to $165 billion to $169 billion, and the capex range was narrowed to $130 billion to $145 billion, from $125 billion to $145 billion previously. Meta declined to give 2027 capex guidance, describing infrastructure planning as highly dynamic.
Why the capex number dominates the debate
To put the scale in context, Meta’s entire first half of 2026 generated about $64.1 billion of operating cash flow. A capex budget at the top of the guided range would absorb more than all of it on an annualised basis. That is why buybacks stopped and why free cash flow almost disappeared in Q2.
Meta is not alone. Every major hyperscaler is spending at record levels, which is exactly why NVIDIA’s results have been so strong and why data center power has become an investment theme of its own. The difference with Meta is that it does not rent this compute to cloud customers. Every dollar has to be earned back through its own apps, devices and model API.
The balance sheet still gives Meta room. It holds $90.26 billion of cash and marketable securities against $83.66 billion of long-term debt, and the Family of Apps segment produced $23.39 billion of operating income in a single quarter. This is a company choosing to spend its cash flow, not one running out of it.
7. Competitive Landscape
Meta competes on several fronts at once, and the rival is different on each one.
| Front | Main rivals | Meta’s position |
|---|---|---|
| Digital advertising | Alphabet, Amazon, TikTok | Co-leader, with AI ranking gains showing up in conversions |
| Frontier AI models | OpenAI, Google, Anthropic | Late but moving quickly with Muse Spark |
| Personal AI agents | OpenAI (ChatGPT), Google (Gemini) | Strong early downloads, unproven retention |
| AI glasses and wearables | Apple, Google, Snap | Volume leader today |
Where Meta wins
Distribution is the obvious advantage. Muse reached the top of the US App Store within about ten days, and third-party app trackers estimated between 2.3 million and 4.3 million downloads in its first weeks. Meta also promoted Muse aggressively through its own apps, and by 22 September Muse ranked among the top ten brands by advertising spend. Few competitors can buy that much attention from themselves.
In glasses, Meta has a multi-year head start in both product and manufacturing partnership, with more than 7 million pairs sold through EssilorLuxottica in 2025 alone.
Where Meta is exposed
Meta is a newer entrant in frontier models than OpenAI, Google or Anthropic, and enterprise buyers already have established relationships elsewhere, as the Salesforce and Anthropic partnership showed. Early download numbers also say nothing yet about retention. A spike driven partly by Meta’s own house ads needs to become durable daily usage before it counts as a moat.
8. Risk Analysis
The risks here are less about whether Meta can build good products and more about whether the spending pays back on a timeline shareholders accept.
- Capex risk: $130 billion to $145 billion of 2026 spending, no 2027 guidance, and free cash flow already near zero in Q2
- Monetisation risk: Muse has no disclosed revenue, and Meta’s own ad separation promise limits the most obvious route to monetise it
- Trust and privacy risk: Muse needs deep access to email, calendars and payments, and Meta’s history makes that access harder to win
- Legal and regulatory risk: in August 2026 Meta agreed a multistate settlement of up to $18 billion over social media harm to young users, including new limits on teen usage and features
- Reality Labs risk: a $4.62 billion quarterly loss with no stated path to profitability
- Execution risk: the Muse Charm has no price and only a handful of units built, and Muse on glasses is still “coming months” away
- Competition risk: OpenAI and Google are pursuing the same personal agent and device opportunity
9. Bull vs Bear Case
🐂 Bull Case
- The ad engine is getting measurably better: Facebook conversions up 15.7% from ranking improvements and Advantage+ above a $75 billion run rate
- Muse gives Meta a direct consumer AI product with strong early demand, plus three new revenue routes in subscriptions, API sales and commerce
- AI glasses have real traction, with more than 7 million sold in 2025 and sales nearly doubling again in Q2 2026
- Revenue growth of 28% on a $228 billion trailing base shows the core business is far from mature
- A forward P/E of about 23 is not extreme for a company growing revenue at this pace
🐻 Bear Case
- Free cash flow collapsed to $784 million in Q2, and buybacks stopped entirely
- Operating income fell 8% even as revenue grew 28%, because costs grew almost twice as fast
- Muse revenue is undisclosed, and early downloads were boosted by Meta’s own heavy promotion
- Reality Labs lost $4.62 billion in one quarter, and the flagship new devices ship in 2027 or have no price
- The $18 billion settlement and a weak privacy record could slow consumer trust in an agent that needs personal data
⚖️ Base Case
- The ad business keeps funding the build-out, with revenue growth staying strong into Q3 guidance of $61 billion to $64 billion
- Muse grows usage steadily but contributes little disclosed revenue before 2027
- Free cash flow stays compressed while capex remains elevated, which keeps the stock sensitive to every capex update
10. Valuation: Is the Good News Already Priced In?
At $751.66, Meta trades at about 28 times trailing earnings and 23 times forward earnings, with a market value near $1.91 trillion. The stock sits close to the top of its 52-week range after a sharp September rally, and it fell 3.3% on 25 September as some of that momentum faded.
Analysts are overwhelmingly positive. The consensus rating is Strong Buy, and JPMorgan’s Doug Anmuth raised his target to $920 from $820 after Connect, citing Muse’s strong start. Yet the average price target of about $761 sits only slightly above the current price, which suggests much of the Muse optimism has already been absorbed.
Three ways to look at it:
- Earnings multiple view: a forward P/E in the low 20s is reasonable for 28% revenue growth, if margins stabilise
- Free cash flow view: on current capex the free cash flow yield is very thin, so the stock is priced on future earnings power, not current cash returns
- Optionality view: investors are paying something for Muse, glasses and the model API, none of which has disclosed revenue yet
The valuation is justified if the capex produces the ad and product gains management expects. It is harder to justify if 2027 capex rises again without a matching lift in revenue.
11. Catalysts to Watch in the Next 12 to 24 Months
- Q3 2026 results: revenue against the $61 billion to $64 billion range, and any first commentary on Muse usage or subscriptions
- 2027 capex guidance: likely the single most important number for the stock over the next six months
- Muse on glasses: the promised wake-word launch in the coming months
- Muse Charm: final price and whether it ships on time for the holidays
- Muse Confidential VM: delivery of end-to-end encryption would strengthen the trust argument
- Meta Ray-Ban Display: sales in the new UK, Canada and EU markets
- Meta VR Glasses: launch in spring 2027 at $1,299.99
- Regulation: implementation of the multistate settlement and any new AI or privacy rules affecting agents
12. Final Thoughts: Is Meta a Real AI Business?
Yes. Meta Muse AI is not a narrative bolted onto a social media company. Meta builds its own frontier models, runs them across 3.6 billion daily users, and already earns measurable returns from AI in its ad engine. Muse is the first time that capability has been packaged as a consumer product people can choose to use, and the early response has been strong.
The biggest opportunity is that Muse becomes the everyday interface for a large share of those users, on phones, glasses and new devices, and opens subscription, API and commerce revenue on top of the ad business. The biggest risk is that the infrastructure bill arrives years before that revenue does, while free cash flow stays near zero and trust questions slow adoption.
For long-term investors comfortable with high capital intensity, Meta offers one of the clearest ways to own a company turning AI spending into products. For investors who prioritise free cash flow and buybacks, the next few quarters are likely to be uncomfortable. Either way, the 2027 capex guidance will tell you more about this investment than any product launch. Meta is also exploring other new rails for its platforms, including its stablecoin strategy, which is worth following alongside Muse’s commerce ambitions.
FAQ
What is Meta Muse AI?
Muse is Meta’s personal AI agent, launched in the United States on 8 September 2026. It is powered by the Muse Spark model and is designed to carry out tasks such as sending emails, booking travel, filling out forms and shopping, rather than only answering questions.
Is Muse Spark open source?
Muse Spark is a proprietary model, sold to developers through the Meta Model API. Meta has released a smaller open-weight model, Muse Glimmer, under the Apache 2.0 licence, which is designed to run on a single consumer GPU.
How does Meta make money from Muse?
Muse is free for everyday use. Meta earns from paid Meta One plans that add AI usage, from selling Muse Spark through its model API, and potentially from commerce as Muse completes purchases. Meta has not disclosed any Muse revenue yet.
Does Muse use my data for ads?
Meta states that Muse does not share conversations or data in a user’s secure virtual machine with its ad systems. These are Meta’s own claims, and they have not yet been independently audited in depth.
What did Meta announce at Connect 2026?
Meta introduced Ray-Ban Meta Gen 3 glasses from $449, the camera-free Ray-Ban Meta Audio at $349, the $1,299.99 Meta VR Glasses for spring 2027, and the Muse Charm pocket device. It also expanded the Meta Ray-Ban Display to more countries and added voice, video and computer control to Muse.
How much is Meta spending on AI infrastructure?
Meta guided 2026 capital expenditure to between $130 billion and $145 billion. In Q2 2026 alone, capex was $31.08 billion, which left free cash flow of just $784 million for the quarter.
Is META stock a buy after Meta Connect?
That depends on your view of the capex cycle. The core ad business is growing 28% and the forward P/E is about 23, but free cash flow is compressed and the average analyst target sits close to the current price. It suits long-term investors who accept heavy spending today in exchange for AI optionality.
Stay With the Analysis
If this breakdown was useful, read our related coverage on AI and big tech, subscribe for future updates, and visit the One More Stock YouTube channel for video versions of these deep dives. We cover AI stocks weekly on the OneMoreStock YouTube channel.
Financial Disclaimer:
This article is for educational and informational purposes only and should not be considered financial or investment advice. Investing involves risk, including the possible loss of capital. Always conduct your own research and consult a qualified financial adviser before making investment decisions.