Meta 的 AI 战略正凭借 Muse 的推出获得进展，并得到强劲广告收入支撑；尽管高额资本开支拖累自由现金流，但相对于未来盈利增长，该股仍被视为估值偏低。

2026 was a rough start for Meta Stock and as of last month it was down over 10% year to date. That was significantly lagging the overall market. However, something has changed.

Over the past month, the stock is up 31% and earlier this week on Monday, it went up 11% in 1 day.

Today, I'm going to explain what saved Meta Stock. There were two big things that happened over the past month. And the one that caused the recent 11% increase in one day is from the launch of their new personal AI agent called Muse that hit number one in the app store ahead of Chat GPT.

That's both in productivity and overall in free apps.

And as of right now, it looks like this product is a hit. It already has 35,000 reviews with a 4.9 star rating. I'm starting to see positive word of mouth online. The reviews are quite good.

So, what is Metam Muse? Well, it's basically a personal AI agent. If you've been following AI earlier this year, you know about Open Claw and its massive popularity, but it was incredibly hard for the average person to use.

This is basically taking that exact same concept of the AI harness where you give it its own computer and the ability to edit local files. You can connect to all of your different applications that you use.

And it's basically turning AI into a useful assistant that can do anything you can do on a computer. But what Muse finally did right is that they had the UI and user experience right for the average person.

It's a simple, easy to download app. You don't need to be technical. You don't need to know what model you're using. All of that is abstracted behind the user.

And this is honestly from an application perspective Meta's first big home run hit in the AI space. And honestly, it's one of their best internally created apps since Facebook.

I mean, they bought Instagram, they bought WhatsApp, and they did a good job adding things like reels into Facebook and Instagram, but this really is a whole new product, and it's starting to get investors excited about Meta's AI strategy.

An 11% jump on a random Monday is a big move. So, today I'm going to give my update analysis on Meta Stock after the announcement of Muse. I'll explain what the second big factor is that caused the stock to move up 31% over the past month.

I'll dive into the cash flow machine that is Meta's advertising business which has allowed them to invest so heavily in AI and also the risks around their AI strategy which has kind of pulled the stock down in early 2026.

Also explain what this Muse announcement and the positive stock performance means for the broader market including the entire AI hardware supply chain companies like Nvidia and more.

There you can get a full picture look at the financials and valuation of Meta Stock and we will definitely be diving into that. But first I want to focus on the story here. Why was Meta Stock doing so poorly as of just August 30th?

It was down 13% year to date when the overall market was up. Most AI stocks were up. Most big tech stocks were starting to rebound. You had Microsoft rebounding. You had Amazon rebounding. Why was Meta lagging?

Well, part of it you had a lawsuit overhang on the stock. You may have heard about this. Meadow is being sued over their social networks causing harm among the youth. I'm not going to get into all the details around that, but they just settled one of the largest cases, and it started to give some certainty around the downside.

In this case, Meta has to pay $18 billion over 10 years.

Now, this sounds like a lot, and it's actually one of the largest settlements in the entire history of the US. I believe it rivals the settlements of the cigarette companies back in the day.

But the reality is that's not a large amount of money for a company like Meta.

Over the trailing 12 months, so the past year, they've generated $130 billion of operating cash flow. That's up 27% year-over-year. From 2024 to 2025, they grew their operating cash flow by more than $18 billion.

And that settlements over the span of 10 years. And Meta is not done growing. Their cash machine continues to increase every single year. And it's due to their dominant social networks that they own and the highly profitable digital advertising business built on top of it.

Instagram and Facebook ads, meta ads, they crush. They print money.

And they have been using that cash machine to make sure they're one of the top AI players. And the strategy is starting to work. Over the past year, they've spent $89 billion in capital expenditures.

That was up 71% from the year prior. And this is still ramping up. There are no signs that it's going down. In fact, basically every single thing they're saying is that capex is going to keep growing.

In the latest quarter, they spent $30 billion in capex. This is all for the most part going to the AI data center buildout.

And yes, Meta is able to spend this from a position of strength and it has primarily been funded with their cash flow. But it's led some public market investors to wonder whether or not they're going to get a return on this investment. And it's understandable to think that.

Unlike Microsoft, Amazon, Google, Meta doesn't have a cloud business where they're renting compute to external customers. Now, many people might not know this. Meta has operated data centers for many, many years.

So, they do have a lot of experience in that area, but it's always been for internal use. And as of right now, they still don't have a cloud business. These AI data centers they're building are being built out for internal use and training of their models.

And the other big thing is that they don't have a model API business. that's generating significant or really any practical revenue at their scale. Not anywhere close to Enthropic, Open AAI, or even Grock.

But Meta's models are starting to actually get much better. And I'll dive into this later as I get into the Muse announcement.

But focusing on the story of why the stock is down this year. It has to do with this large capex investment, questions about return on investment, and it really all comes down to this metric, free cash flow.

Ultimately, a company is worth all of the free cash flow they'll generate in the long term discount back to the present value of today. And Meta's operating cash flow has been growing great, but they're investing all of it in capex.

Their free cash flow has barely grown over the past 3 years. It's up 9%. And if you look here, it's actually been trending downward. In fact, in the latest quarter, they only generated $1.75 billion.

Over the trailing 12 months, it's $40.9 billion, which is down 18% year-over-year.

And some investors that are not as bullish on AI think of this as a concern. It was even impacting companies like Amazon, who have a much clearer return on investment with Amazon Web Services.

Amazon has been even more aggressive, taking their free cash flow negative, while they hit record highs in operating cash flow.

So with all of this in mind, it starts to make sense why Meta Stock was underperforming the overall market and it wasn't rallying like we are starting to see with the hyperscalers Microsoft, Amazon, Google.

The monetization path for their AI strategy was unclear, but it's starting to get much more clear and this Muse announcement was a big part of clarity in the app side of that strategy.

But even on the data center and cloud side of the business, I think some people are underestimating the potential monetization that could happen if they start renting out that compute. And I'll dive into that in a second.

But first, let's focus on Muse. If you've heard Nvidia CEO Jensen talk about AI, he describes it as a five layer cake. You have energy, you have chips, you have models, you have data, and you have applications.

And the announcement and early success of Muse is a big statement in that application layer for AI.

In that layer you have chatgbt, you have claude and really over the past year and a half to year the big innovations have been around computer use, tool use and agentic capability.

And so far there's been different applications of that. Coding agents has been the best monetized version of this so far in 2026. Claude code has basically been a huge hit. You have similar things like cursor here.

You have coding agents being able to do tasks. There's the claw code harness around it. They have access to computers, files. It can happen locally. It can happen remote in the cloud.

And it's wild to think that this happened in 2026 as well, but you also had OpenClaw come out. And this basically was an agent harness that you could connect to anything and give full access to a computer.

You can also host it in the cloud. However, from a consumer standpoint, this was incredibly hard to set up. You have to know how to use a terminal. So, if you're not a tech person and if you're not a software engineer, you're probably not going to be someone that's actually going to use a tool like this.

But this is really the foundation of what Muse is built on. It's this concept and innovation.

Both OpenAI and Enthropic, they've been very focused on that coding agent use case since that has been one with the fastest takeoff. It also has the largest capabilities, but they've been aware for most of this year that it can be broadened outside of just coding.

So throughout this year, you saw Claude launch Claude Co-Work. You saw Chat GBT recently launched ChatgBT work. And this is basically that same concept of computer use, but it's natively within chat GPT.

But the interesting thing is that OpenAI, they aqua hired the guy who made OpenClaw, but they never actually made their own version of OpenClaw.

And over the past few months, you've been seeing some of the other AI labs start to take that agentic idea and the foundations of OpenClaw and they've been taking it seriously and they have been releasing some hit products.

You had XAI and Grock. They released a product called Grockbot which is a huge hit. It's the same idea of Open Claw and Hermes Agent, but it's put into a very refined product that's closer to something like Slack or Teams and you can easily set up your bots.

They all have access to their own computer. You can connect to all the different services you use and they frankly just made it a good UI, a good user experience. They abstracted most of the technical details behind it like what model it's using.

You don't have to install it in your terminal. It doesn't constantly break all the time. It's a real high quality application. And Grockbot is a hit. It's one of the best products that XAI has released and it's making huge waves in the enterprise, too.

And to be honest, I think all of the big AI companies, especially the ones focused in enterprise, they're all going to end up copying Grockbot. I think Microsoft should create a clone of Grockbot.

I think Open AI and Enthropic should consider doing it. But that consumer side of the market was still left empty for that AI personal assistant. And there you have Muse entering the chat and it is taking off and it's making people wonder why didn't Open AI do this earlier this year.

Is it because they're compute constrained and they just want to focus on chatbt and codecs? I think that's probably the answer. This honestly created a big gap in that consumer AI agent market and Meta is really good at apps.

Plus they have the distribution of Facebook, Instagram and WhatsApp. And if you open your Instagram or Facebook, you will see a little prompt telling you to download Muse.

And I've been seeing a lot of positive word of mouth about Muse recently. It's only been out a week, and I think it's going to be here to stay. And based on that 11% stock price increase on Monday after the weekend where Muse is out and you're starting to get some positive word of mouth, I think that means that the market is thinking that Muse is a real product for Meta and they're starting to wrap their head more around what Meta's AI strategy is going to look like.

And what did they do? They made an easy to use consumerfriendly app and they're plugging it directly into the Facebook, Instagram, WhatsApp distribution. And here's the big kicker.

They're making it free, which is actually a pretty differentiated thing in the AI agent space. You're up to 100 million tokens a week. Also, with the reality of how this works, that means you're getting like a virtual private server yourself that you're not paying for because every single one of these Muse agents has their own computer.

So, as of right now, Meta is being aggressive in subsidizing this to try and get Muse to be very popular and make it a hit product.

And from what I'm hearing, the free tier basically is good for most use cases. However, they do have a power tier of $20 a month and a max tier of $100 a month and lets you use more tokens.

So, more AI inference. So, right now where we sit as of like a week after the product came out, it's looking like it's going to be a hit and it will grow a lot, but there's some questions around monetization that I don't think the market is necessarily thinking about right now.

And that makes sense. It's very early. But, as I'm looking at it, all I see is cost right now with Muse. And I'm wondering how they're going to make money.

Now you have the subscriptions for the power users, but I listened to an interview earlier this week with Mark Zuckerberg and it was kind of interesting in that they were talking about it as they would have a fee take rate on agent transactions and they're working with Stripe to kind of build this out.

And I'm going to be honest, my early thoughts on that, I'm not sure how that's going to work out. This is basically a big bet on people making transactions with their agent, shopping becoming very popular, but also people making money with their agents.

Zuckerberg was emphasizing that as well, and I do think that will be something that can happen, but I'm not sure that will always be routed through something Muse can monetize.

And on the shopping side, Amazon recently blocked Muse after it came out. And that's because Amazon wants people natively shopping on Amazon.com and the Amazon app because one of Amazon's most profitable businesses is their advertising business.

So, they need people on the platform, not agents going in and buying stuff without people looking at the ads. Amazon also has their own AI on Amazon.com. It's called Rufus. I haven't used it before, but apparently it's actually been generating billions of dollars for Amazon on Amazon.com, but all of that needs Amazon on their actual platform.

So, Amazon has been pretty anti- AI agents buying stuff because it hurts their advertising business.

So, I'm going to be honest, I don't know how I feel about this AI agent take rate monetization. I'm not sure that's going to be what it is long term, especially for a consumer application.

I think a lot of people that are in consumer, they're not as interested in the productivity angle of it. Now, what could be interesting, and Meta has said that they're not doing this so far, but I think they'll reconsider that longer term, is the idea of having ads within Muse.

I mean, if you think about it, Meta's entire business is digital advertising pretty much. If you've been following the monetization of free AI and people talking about putting ads in chatbt, which is an early roll out for free users, why wouldn't Meta also do that?

Now, it makes sense not to do that yet because they're subsidizing it. They're trying to get as mass adoption as possible. But long-term, they already have that killer advertising business.

I think it makes sense to put Muse in ads. And if I was a betting man, if we look back 3 to 5 years from now, Muse is going to have ads.

Now, I mentioned earlier that this is a big thing for the broader AI trade, for AI hardware, for the semiconductors industry, and that's because if the market starts to respect Meta's AI strategy more, then it gives increased confidence that Meta will continue being a large data center builder going forward.

And if you're following it, basically that's what they've been saying anyways. But now, they're actually having some product success. Their AI models are getting much better. So this $89 billion of capex that's going to go up and up and up and up for the years to come.

Also related to muse and the idea of agents, you need a lot more than just inference. You need CPUs, you need storage, you need memory because every single one of these has access to computers or at least some provisioned part of a virtual server.

And this is not necessarily unique to meta. This is happening broadly across the industry. And that's why you've been seeing Nvidia move into the CPU market as well and the entire Vera Rubin system.

It's built for the idea of agents. So, they've been planning for this, but really we're in such a demand environment right now that it's going to benefit everyone.

So, Meta doing well. It's good for companies like Nvidia. It's good for companies like Broadcom. It's even good for companies like Dell. It's good for that AI hardware buildout.

And a lot of that capex spending is going to these semiconductors like Nvidia and it's their free cash flow.

So Meta's story around their AI investment. It's getting better. You have their first big win in the application layer. It's early stages, but you can see how this could snowball into being a huge product for the company.

But the part that I don't think people are talking about is the infrastructure side of the company. Yes, they don't have a cloud business, but we've seen one prime example of a company that didn't have a cloud business and then all of a sudden they're getting tens of billions of dollars of annual revenue and that is SpaceX, the owner of XAI.

And if push comes to shove, Meta could end up doing the same thing. And then all of a sudden, what do you know? Meta now has a hypers scale cloud business.

That said, they might not have to go that route because Meta has something that SpaceX doesn't have, which is a digital ad business which prints money. So if they want to, they could be aggressive.

They could subsidize Muse and their AI products, try and get as much market share as possible. They can go that route, but also with the capex they're spending and their plans for the data center buildout, they might be at a point where they also have excess compute.

And I think a lot of players in the market would like to see Meta start to rent out that compute going forward.

Let's wrap it up by talking about the valuation of the company even after the stock price has gone up around 30% over the past month. And here you can see the price to the estimated fair value of the company based on its historic multiple that it trades at.

So here you can see with earnings per share the median multiple over the past 5 years has been 26.49. It currently trades at 27.74p ratio. This is trailing 12 months adjusted earnings per share.

So the company is currently trading slightly above that historic median multiple.

It was trading at a slight discount earlier this year before the recent increase in stock price. And that's why a lot of valueoriented investors were starting to buy meta stock.

After all, it's one of the most dominant big tech companies in the world and they were trading at a somewhat cheap multiple.

But here's the thing. based on free cash flow, which I've said is one of the most important long-term financial metrics. Their capex has been ramping up far more than their operating cash flow growth.

So, their free cash flow has been declining. So, you can actually see that fair value of the company has gone down in 2026 if we're basing it off of free cash flow.

Median multiple is 28.47 of price to free cash flow ratio over the past 5 years. The company's currently trading at 45 price to free cash flow. So if you're basing it based on free cash flow, that metric, the stock looks expensive.

And right now, Meta is looking like it's going to keep growing in the coming years. Currently, analysts are expecting double-digit earnings per share growth basically every single year going forward.

Current 2030 analyst estimates is $54.20 for earnings per share. At today's price, that would be a forward-looking P ratio of 13.6. So, it's trading at 13.6* 2030 earnings.

So, even if the stock trades at its current 23.6 6 P ratio multiple. That's the forward looking P ratio in 2026. Even if they keep trading at that, if you assume they hit those earnings per share estimates in the coming years, that would give a 2030 projected price of $1,279.

That'd be if they're earning $54.20 of earnings per share and they trade at 23.6p ratio. That implies 73.7% upside from here, which would be 13.8% annualized.

So, even after going up 31% a month, Meta Stock, it seems like it's still poised to do well long term. But that's just my opinion.