Meta Investors May Be Sitting on a Goldmine

Meta is shaping up to look like it's going to be one of the greatest comeback stories in big tech history. Here is why I believe that Meta is going to head to a thousand dollars within the next 12 months. Let's go ahead and have a look. First up, semi analysis, which is one of the greatest gold standards for reports on artificial intelligence, reported on Meta Super Intelligence Labs or MSL. They spoke about Muspark 1.1, which is one of Meta's newest models and it's actually for coding and agentic use cases. They put it on par with GLM 5.2, which is another Chinese open- source model, and then an older enthropic model at about Opus 4.6. The whole reason for this anyway is that they said although we're not going to be using Muse Spark 1.1, the amount of change and how good it looks versus previous models and how fast that they've grown. That's the part that you need to pay attention to the trend. And as you go through, what semi analysis ends up talking about is this data, talent, and compute. This is an important point because in order to get the best models possible and have the potential to even service all of the potential tokens that customers might want to use, you need an unbelievable amount of compute. And Meta has been spending as if they are a frontier model provider. Yet, they do not end up selling tokens other than just the recent Muse Spark 1.1. So, they're getting into a new revenue stream that only started about a week ago. To put this into perspective, we'll use OpenAI's own report. As they talked about at the end of 2025 having roughly 1.9 gawatts of total compute, they have identified 8 gawatt that they want to build into and they want to get to around 30 gawatt by 2030. The cost per gawatt of a data center is roughly 50 billion and soon to be a hundred billion using Nvidia's new Vera Rubin technology. So the total buildout is anywhere between $1.5 to $3 trillion. In order to have that amount of compute, you need to be making a lot of money. And right now, OpenAI is hopeful that they'll see a lot of that money. But Meta is a different story. They are generating a ton of new capital and we'll go through that in the actual analysis of the company. But if there is somebody that could have models as good as OpenAI, a company like Meta could actually fund this buildout. And Semi analysis talks about this. They say at the simplest level there are three things that you need to build a true frontier model. Data, talent, and compute. We believe Meta is the only hyperscaler/neoud on track to be world class at all three of these and therefore has the best chance of catching up with enthropic/ OpenAI. They did also talk about Google as well. They said that Google's focus on selling all of this compute is taking away from their ability to use it internally. You're not focused on training your own models for external use. whenever you have all of that ready to go at very high margins for Google Cloud. Let's just talk really quickly about how much that compute that we ended up showing Meta and OpenAI and Enthropic sort of fighting out for number one. How much revenue that's actually bringing in these companies. OpenAI is bringing in anywhere roughly about 25 to30 billion for which they've not updated this number in a long time. Enthropic updated this number to 47 billion and now we're having some yippit estimates that are ended up saying that it could be as high as $69 billion which would be on pace to do approximately $550 million per day. And unlike OpenAI or Enthropic that has to raise a lot of money, do a lot of series funding, potentially go public, Meta can just fund this through the businesses that they already have. Facebook ads, Instagram ads, their advertising business is huge. So they're one of the few companies that can actually afford to continue to spend more and this is what their advantages come down to. They have distribution. They have cash generation, proprietary data, infrastructure and monetization. Let's go through this list one by one and talk about why Meta has an amazing chance at becoming a real AI player. First up, we have distribution. This is through apps like Facebook, Instagram, WhatsApp, and Messenger. These amount of overall apps essentially translate to 3.5 nearly 3.6 billion daily active users. That's essentially 3.6 billion daily active people who are going to Facebook and Instagram not for artificial intelligence purposes that will be potentially using those apps the same way that they would use chatbt or claude. No need to pay for advertising for your apps whenever you own all of the platforms as well. But then we have cash generation. See, the thing is is that Meta already makes a ton of money. And you can see the split here between advertising revenue and then reality labs. The most amount of their overall money comes from advertisements. This is not from AI purposes or charging for highMP compute versions of their apps. And this growth is actually accelerating now to over 33% year-over-year. That alone should be bringing up Meta's stock. Yet, we have an opportunity to buy this stock at a cheaper price. put up against the other big tech names that also have cloud services and AI ambitions, Meta is by far the fastest growing name out of all four of these names, including Google, Amazon, and Meta. And we're not just talking about revenue here. The overall amount of margin that they're bringing on on top of this revenue is at the highest that the company has ever seen, 47.5%, nearly 27 billion of net income. And whenever you compare this margin, aside from Google, which last quarter had a very big benefit because of SpaceX and anthropic shares that they own, Meta doesn't really do a lot of that venture funding type of things, a lot of their actual net income comes from income from operations. It's not non-operating income. So, this is a really good sign how high the margin is overall. But next, we have proprietary data. Think about the amount of data that's coming in from 3.6 billion daily people. Whether that's a social graph, knowing who talks to who and and how that ends up working, knowing how people end up communicating with each other, or all those images and videos of Instagram and WhatsApp and all of these things all flowing in to Meta or the real-time engagement feedback, what do people like? What do people not like? Or how that actually advertises and converts into real products, which is a big way that OpenAI wants to make money is through advertising. Meta is already there. let alone all the other products where they could have even more visual learning or other ways of understanding their customers whether that's through meta forms or threads or new meta glasses or VR opportunities. Meta has the data and not accessible data. We're talking proprietary data, data that is not accessible by Anthropic or OpenAI or Google because they have their own platforms which allows them to end up expanding into all these new products that they are using trained on their own proprietary data which allows them to create their own apps like Muse image which looks very good comparatively to where OpenAI, Gemini and Enthropic are at. This data advantage is already giving them a big upper hand whenever it comes to their own business. And we're not talking about just selling tokens in terms of artificial intelligence. We're talking about increased amounts of ad impressions or the actual click-through rates and how much money that they're charging for those advertisements. This is making the entire business much stronger. But that brings us to number four, infrastructure. Meta is no longer a super free cash flow positive capital light business. They're building some of the most ambitious data centers that people could have imagined like Prometheus or Hyperion. Capital expenditures have gone from a couple billion dollars a quarter to 20 plus billion dollars a quarter. And that pales in comparison to where they want to be next. But even looking ahead, July 8th, this was very recent. Meta just broke ground on their first data center in Canada, likely to end up getting towards 1 gawatt in scale. or just today, June 13th, Meta is now targeting a $250 billion data center, which is likely to be upwards of 5 gawatt. That's more than their entire 2026 budget for artificial intelligence in one data center. In fact, that's more than any of these companies are spending in one data center. And yet, although they're not nearly as capital-like as they were before, they still maintain the highest free cash flow margin across any of the other cloud providers in AI names in big tech. Looking at a free cash flow margin of 23.5%, which is definitely lower than where we were years ago, but also that's enabled us to accelerate revenue and accelerate their net income. In a Reuters article titled Meta to put AI chip production in September as it looks to double compute capacity memo shows. This was an internal memo sent from Mark Zuckerberg to the rest of the Facebook team essentially saying that they're targeting 7 GW of compute in 2026. The majority of that 5.5 gawatt is coming on the back half of 2026, but they want to double this capacity to 14 gawatt in 2027. So if you thought capex was going to slow down, you're essentially looking at them not only spending way more in the back half of 2026, but that they want to essentially increase that by another 100%. So potentially targeting $300 billion worth of capital expenditures next year alone. This capital raise is getting so fierce that Meta is actually looking to raise tens of billions of dollars in stock offerings as it seeks new sources of capital following what Google did raising $85 billion in equity the past week. So they want to boost their infrastructure because they believe in what's happening in artificial intelligence and potentially even more. And let's talk about this. That comes up in number five, monetization. We already know about the ads. We know about the commerce. We know about what they just just launched. I mean last week Metamuse Spark APIs, the token revenue that OpenAI and Enthropic is making. But now let's talk about MetaMP compute cloud. So we already know how big and sizable this company is just currently doing advertising revenue. You can see that breakdown here. $56 billion and almost all of that is coming from advertising. Free cash flow is coming in at $13 billion. But this is not enough for them. They want to open up to new products. And as Mark Zuckerberg said, we are building a business agent focused on helping entrepreneurs and businesses across the world use our tools and others to grow their efforts, reach new customers, and serve existing customers better. These agents will work together to form an ecosystem. Then Mark Zuckerberg says, "I think people are going to want to be willing to pay a lot of money to have premium or highMP compute versions of it. Muse, Spark, their actual AI tools." And you can see that this is the case whenever you look at cloud revenue backlogs across the big five. You're looking at Google, Microsoft, Amazon, Oracle, and Corewave. All these backlogs essentially flow down to the model providers. If you're finding Enthropic useful or your own models very useful, then you're going to need GPUs. And so this is why the backlog is climbing for all of the biggest names in the cloud space. On top of that, you're also seeing the cost per hour of GPUs climb across many of the Nvidia chips, and that continues to go up. But this is the interesting part. Mark Zuckerberg is now exploring an AI cloud business that would make sense as they're getting offers that are so high that it might end up making more sense to rent it out instead of using it internally. Some of the plans include selling access to hosted AI models, but also raw computing capacity. So, let's talk about a company that just recently did this, SpaceX. They bought a lot of compute for their own internal use, but then ended up selling it off whenever Grock was not the biggest model and actually wasn't seeing a need for the 1.3 gawatts worth of data centers that they had from Colossus 1 and Colossus 2. So, they ended up signing a deal with Google and ended up signing a deal with Enthropic to end up selling this compute externally. As of today, there's currently three deals that have actually been made. One with Enthropic, Google, and then Reflection AI, a much smaller one, but still of decent size. and for a pretty long period of time, 36 months, 33 months, 42 months for reflection. Now, all of these deals are able to be cancelled within a 90-day period from either customer. So, they are short-term in nature, but if that was to actually go through, you're looking at $15 billion of annual spend from Enthropic, 11 billion in annual spend from Google, and about 1.8 billion from Reflection. So SpaceX, which had no cloud business, is essentially staring down $27.84 billion in cloud revenue annually. That's just from the customers that they've sold today without slowing down their ambitions of Grock 4.5, which just recently launched. And maybe they could do a lot more internally, but right now they're saying Anthropic needs it a lot more than we do. Google needs it a lot more than we do. And the price that they're offering is just way too high to ignore. If you look back at some of these blended hourly rates, these are way higher than what we're end up seeing at market levels. They paid an extreme premium. Enthropic Google and Reflection did in order to get access to compute today. That's the main part is data centers that are already created is of the highest priority because all of the deals that are being signed with Oracle and Microsoft and Google, these take years and years to build out. So customers are signing deals with Cororeweave and Nebius and all of these companies for 2027, 2028, and many years into the future. All of these deals that were just signed were for today. Let's do the math on this. SpaceX has roughly 1.3 gawatt today, which could expand to 2.3 gawatt as they want to continue to expand Colossus 2. Meta has seven gigawatts in 2026 with actually hopes of them getting to 14 gigawatts by 2027, but I'm not even including that. We're going to look at the bottom half of this. So essentially 3x, not the 5x of what they have on 1.3 gawatt. We're going to assume they're fully built out today, which is not even true. This is the amount of Meta Platform's total revenue that they bring in and what is expected for 2026, 252 billion. Even though that they've beat earnings almost every single quarter, we're likely to just see what Wall Street is saying right now. I'm being super conservative. If you were to tack on the potential cloud revenue that SpaceX did on 1.3 gawatt, remember Meta has much more than this. They essentially signed $27.8 billion worth of total revenue that had nothing to do with Meta's current existing business model, which would not be cannibalistic or anything like this. If you were to then put them to size, which is to say they have essentially 5.5 gawatts coming online at the back half of this year and then potentially upwards of 14 gawatt. We're not even talking about that, but the just ability for them to 3x SpaceX over the next year, you're looking at potentially $83.5 billion of revenue. Remember, this is the money that SpaceX made while also still putting out newer models like Gro 4.5 and also serving existing customers. Meta uses a lot of their compute internally and I don't even expect for them to make $80 billion in revenue. Let's imagine they make 10 billion. What is that going to do to the stock as we end up seeing a way higher than expected overall growth rate and we do see a brand new line of revenue that other institutional investors and analysts were not expecting. But that brings me to valuation because although they have a lot of opportunity, so much compute, amazing models, and unbelievable talent across their entire corporation, the valuation does not line up to the opportunity here. So, let's just talk about not including cloud here. If we take a look at Meta's PE ratio, which has actually climbed up within the last month or so, this is because stories like the overall cloud business and Metamuse Spark 1.1 being way stronger than Wall Street analysts had expected. Now we're starting to see some appreciation in the stock. Same thing with the forward PE being at 20.5 times roughly. Even whenever you put this up against the rest of big tech, Meta comes in at a PE ratio, not a forward PE ratio of essentially the second lowest on the list, just above Microsoft, which is another name that I'm very bullish on. But on the forward basis, they actually come in lower than Microsoft, only by a little bit at 20.5 times. whenever Google and Amazon are much much higher and Microsoft is low just like Meta. All of the other companies on this list have explored clouds have explored selling other external AI tools for profit. Meta is only starting to do this. They don't even have a quarter where you see any dollars of them selling Metamuse Spark to external customers or to developers through APIs. And although they're not likely to see a lot of new revenue from Metamu Spark 1.1, like Semi analysis said at the beginning of the video, it's not about this model. It's about their trend and their ability to do this based on compute. They have the compute, they have the customers, they have the data, they have the monetization, they have the cash generation, everything that you could possibly need for them to be a frontier model. Mark Zuckerberg has those tools. If you were to just look at Meta versus its all-time high versus today, all the company has done is make more advertising revenue, bring on more customers, and expand the business just internally on the businesses that we already have. But we are at an inflection point. We should already be above all-time highs. Now, we have an opportunity to really send this stock much higher if Mark Zuckerberg can actually do well and sell this compute either to other customers or use it internally for their own API business. compete with OpenAI and Enthropic or compete with Google Cloud and AWS. Either way, this is brand new revenue for the company and they are ready to do this. They have all the tools necessary. Let me know what you think down in the comments down below, but make sure you subscribe. Until next time. Bye for