Meta shares are still cheap and worth buying. Here's why
A strong new model, an impressive app for personal agents, and an inviting chart are just some of the reasons why it’s a good time to start building a position in Meta Platforms . Meta’s latest artificial intelligence releases — the new Muse Spark 1.3 model out last week and the Muse agentic app unveiled Tuesday night — are casting a different light on the stock’s undemanding valuation. Below the surface, the stock looks even cheaper because Wall Street doesn’t appear to be giving Meta enough credit for what it’s building. Of course, shares have looked cheap versus the market and peers for a while and that hasn’t been enough to lead to a sustained and fruitful rally. But these new AI updates, combined with other potential levers to pull to make money, should help change the narrative. Investors were buying it Wednesday, with the stock up 5%. Keep in mind though, shares down 2% year to date and needs to gain another 22% to get back to record highs around $790 last September. Add in a strong technical setup, and investors will be rewarded for building a position now, if they can stomach any volatility between now and the midterms — even more so if they take advantage. We consider ourselves to be fundamental analysts rather than technicians. Nevertheless, we can appreciate that fundamental analysis is all about being unemotional, objective and rational. Technical analysis, on the other hand, doesn’t discern between what’s rational and what’s irrational; it provides information that is the result of past supply-and-demand dynamics for a given stock, regardless of why investors are buying or selling. When the technical and fundamental setups do not align, investors must proceed with a bit more caution and play a careful balancing act between the two. When the fundamentals and the technical setup are in agreement, then you may be looking at a very good opportunity to make some money. That’s now the case with Meta Platforms. META 1Y mountain Meta’s stock performance over the past 12 months. Shares of Meta currently trade at about 19 times earnings estimates for 2027, according to FactSet data. That already looks attractive given it matches the market multiple one might ascribe to your “average” company, but even more so considering Meta’s highly positive operating cash flow core business — known as the Family of Apps — is growing revenue at 20% per year. However, when you start to consider what Meta may be worth from a sum-of-the-parts perspective, things start to get very interesting. The valuation looks even better when taking a sum-of-the-parts view, which allows us to more closely consider the value of its artificial intelligence efforts. We’ll bracket those AI efforts into two main buckets: the model and the compute. In both areas, the market may not be giving Meta the respect it deserves. Frontier lab in our midst On Sept. 2, Meta released its Muse Spark 1.3 large language model (LLM). It quickly garnered praise not only because of how cheaply it is priced versus other models, but also because of its capabilities. In fact, the model proved so capable that, for the first time, we started to hear hype about how Meta’s AI efforts have resulted in an AI lab that can potentially rival the work taking place at leading pure-play labs like Anthropic and OpenAI. For example, Muse Spark 1.3 scored a 48 on the Artificial Analysis Intelligence Index, which bills itself as an independent industry benchmark. That score leapfrogged over GPT-5.6 Sol, which at the time had been OpenAI’s leading model. As of Wednesday, OpenAI’s new Astra model has overtaken Muse Spark, according to this benchmark, while Anthropic’s Fable remains in the lead. No matter how you slice it, Meta’s AI lab is on the top three podium — and that should have some value to investors. That Meta Platforms has managed to stand up a frontier-level AI lab — and one with access to more owned compute than either Anthropic or OpenAI — is simply not being appreciated by the market. Consider that Elon Musk’s SpaceX , which owns xAI and its Grok chatbot, has a current market capitalization of about $2 trillion. Anthropic was valued at $965 billion in its latest private fundraising round — and it’s reportedly seeking a valuation twice that amount in its planned initial public offering. OpenAI’s latest round valued it at $852 billion. IPOs from Anthropic and OpenAI are expected this year or in early 2027. Anthropic and OpenAI are not believed to be profitable on a Generally Accepted Accounting Principles (GAAP) basis. For its part, money-losing SpaceX is expected to flip to profitability in the September and December quarters, thanks primarily to lucrative compute-lease agreements signed this year with Anthropic and Alphabet’s Google (more on these in a bit). Of course, SpaceX, with its rocket business, is much more than a pure-play AI lab. And while SpaceX’s extraterrestrial ambitions have historically most excited investors, the company said itself in its IPO filing that 93% of its total addressable market is the type of AI products that LLMs provide exposure to. Now, how about these companies’ valuations? Based on media reports that Anthropic has a $65 billion annual revenue run rate, we’re looking at a price-to-sales valuation between 15 to 30 times (using a market cap of $1 trillion to $2 trillion; we know, that’s a big range). With an annualized run rate topping $40 billion, OpenAI is valued at roughly 21 times sales. For SpaceX, based on fiscal 2027 sales estimates of about $97 billion, it has a price-to-sales multiple of about 20.6, according to FactSet. Meta, meanwhile, has a market cap of about $1.65 trillion, with 2027 sales estimates of about $305 billion and earnings of about $33.87 per share, according to FactSet. That means its price-to-sales valuation stands at roughly 5.4 — clearly, much lower than those other three. It also has a user base of over 3 billion people around the globe and multiple apps through which to leverage and monetize AI offerings. We’re using price-to-sales instead of the more standard price-to-earnings so we can compare apples to apples since the aforementioned companies are not profitable. Seeing as Meta doesn’t really have a dedicated AI business and most of the benefit is simply showing up via the company’s core Family of Apps suite, it appears the Street is ascribing almost no value to the company’s AI lab, despite it having now demonstrated to the world that it deserves to be in the conversation when it comes to frontier models and may even lead the pack for certain use cases given its low price point. Sure, Anthropic remains king of the hill and OpenAI has reclaimed its ground on the back of Astra 6, but the race has tightened. And that’s even before we’ve seen Meta’s Watermelon model, which is on the near-term horizon. To be sure, we don’t have an AI-specific revenue figure for Meta, but that’s not the point. Anthropic and OpenAI are believed to be as valuable as they are not because of the revenue they have today, but because of the expected growth. Put another way, the valuations are based on what Wall Street thinks these companies can generate in revenue in the years to come, thanks to the quality of their most important asset — the frontier model. Multiples give us a way to quantify how the Street is thinking about value, but the more important point is that, regardless of today’s sales, Wall Street believes a frontier AI lab —meaning one that can pump out frontier-level models on a consistent basis — is worth in the area of $1 trillion because of the revenue potential (think total addressable market) that a frontier model brings for multiple years in the future. Whether Meta should be valued the same as these other labs is open to debate. However, the more important point is that an asset worth hundreds of billions of dollars, maybe even a trillion, appears to be getting no credit at all. Surely, a model now proven to be in the running with those at the bleeding edge deserves some valuation greater than $0. Even if it’s not $1 trillion, the discrepancy between Meta’s valuation and that of other AI labs suggests there is a much larger margin of safety priced into Meta shares at current levels than meets the eye.This remains true, even factoring in Wednesday’s advance on excitement around the new Muse agent app. Released Tuesday night on a limited basis, the Muse app is Meta’s official foray into personal agents, powered by Muse Spark. It’s getting rave reviews online, and having used it a bit ourselves, we concur that it’s pretty incredible. Jim described the Muse agent as a “souped-up Alexa,” referring to the Amazon digital assistant. There are others like it, including xAI’s GrokBot, which beat Meta to the market this summer. However, given Meta’s over 3 billion global users and Muse’s close integration with Facebook, Instagram and WhatsApp, Meta has a much larger base to which it can immediately start marketing. It is notable that the release came the day before Apple’s iPhone launch event, where the forthcoming AI-enhanced Siri is set to be featured prominently. Muse, GrokBot and Siri, among others, are vying to be the personal assistant of consumers everywhere. It’s still early in this regard, but Muse is a great start for Meta. Cloud business next? Meta’s optionality with its computing resources also appears to have little value assigned to it. To date, Meta has been viewed as the least attractive of the four big hyperscalers — Microsoft, Amazon and Alphabet are the other three — because it doesn’t have a cloud-computing business to help offset the risk of its aggressive data center buildout. This has been a reason why Meta’s price-to-earnings ratio has trailed its hyperscaler peers in the mid-to-low 20s. We understand this worry, and Jim has been pushing Meta to develop a public cloud of its own to help soothe the market’s concern. We were disappointed when Meta didn’t have more to say on this on its late July earnings call, but the good news is that this option isn’t off the table. Meta has built up a vast war chest of Nvidia’s coveted graphics processing units (GPUs), while also developing its own custom silicon alongside Club name Broadcom. The value of those Nvidia chips hasn’t declined in the past month and a half; in fact, we’ve only learned from the likes of CoreWeave and others that the compute has a longer useful life than we thought. If Meta wanted to follow in SpaceX’s footsteps and lease some of the GPUs to external customers, we think it could flip the switch and do just that. That makes a lot of sense for SpaceX based on its current situation; the revenue injection means a lot more for SpaceX than an equivalent amount of dollars flowing to Meta would. What Meta has demonstrated with its technical AI chops since the July earnings call — and with Watermelon’s arrival looming — helps us understand why management hasn’t launched a cloud venture yet. Management seems to believe they will generate more long-term return from using this compute itself than allocating it to, say, a rival like Anthropic. For that reason, we’re honestly not sure whether Meta will ever go forward with a public cloud offering. But at the very least, investors should be willing to ascribe some value to that optionality today. We don’t think its present high-teens price-to-earnings multiple contains any right now. Meta becoming an advertising juggernaut with a cloud business would make it look more like Alphabet, which has Google Search and YouTube ads paired with Google Cloud. Alphabet’s forward P/E is about 22. At 22 times 2027 estimates, Meta would trade at about $745 apiece — and that is arguably a very conservative price objective, given that the cloud business would likely boost the earnings estimates used as the basis to arrive at $745. That would put Meta’s market cap at about $1.89 trillion, and that arguably still does not account for what could be a $1 trillion AI lab in the making, compared with Alphabet’s current market cap of about $4.1 trillion. Putting it all together, Meta shares already looked attractive at a market multiple simply based on the growth of its core social-media business and the cash it spits off. But that’s not been a great reason to buy in recent months. The fact that we are now seeing that Meta may have a frontier-level AI lab inside the company that is getting little credit, along with compute that is proving more valuable than previously due to extended useful lives and a high level of optionality, is why this is a stock to own going forward. With that, let’s think about how one might go about building that position from a technical perspective. Moving averages and trend lines The first thing to note is that shares have now reclaimed both the 50-day and 200-day moving averages, which currently stand at about the $599 and $623 levels, respectively. That means that we are now looking to these levels as support. The more cautious investor will want to wait for shares to pullback to the 200-day MA to “retest” the support and be sure that it holds. However, if you choose to be a bit more aggressive and step in here, those are the two next buy levels to keep in mind as they represent opportunities to reduce your overall cost basis. Above us, shares are trapped by downwardly trending overhead resistance, dating back to August 2025, that currently places resistance at around the $660 level. That does support being a bit more conservative with your buy until that downtrend is overcome. But we would argue the fundamental argument provides the justification needed to step in now and worry less about that potential resistance. Below the 50-day MA, and we would hit a bit of an air pocket on the chart. That said, at around the $550 level, we find a very strong, slightly upward trending level of support going all the way back to April, when shares sold off on AI spending fears. That line has proven to be a buying opportunity, including in early 2025 when shares were hit by President Donald Trump’s “Liberation Day” tariff announcement, and earlier this year during the broad-based market decline spurred by the start of the Iran war. If the stock’s recent momentum were to fade and shares drift back toward the $550 level, we would need to consider if there was a change in the story that prompted the reversal. For now, we think those are the four levels for investors looking to build a position to keep in mind: $650 where it’s trading Wednesday on an intraday basis $622, which is the 200-day MA $599, which is the 50-day MA $550, which has proven to be a key support level for more than a year. Momentum Meta’s Moving Average Convergence Divergence, known as the MACD, is also flashing a buy signal. The MACD tracks several exponential moving averages, in hopes of revealing information about future momentum/direction. The blue line is the actual MACD. It’s calculated by subtracting the 26-period exponential moving average (EMA) from the 12-period EMA. The red line, called the “signal” line, is the 9-period EMA. The other component to note is the 0-line, which indicates whether the MACD is positive or negative, and by how much. The early indication comes when the blue MACD line crosses through the red signal line. The MACD crossing above the signal line, while occurring below the 0-line, as we saw in late August, is bullish. The confirmation occurs when the MACD crosses through the 0-line. A cross of the MACD from below the 0-line to above, as we saw in the days following the Muse Spark 1.3 release, is viewed as confirmation that momentum is now to the upside (again, bullish). The Relative Strength Indicator, which also provides insight into momentum, stands at about 70, right at the threshold at which technicians consider a stock to be overbought (at least in the near-term). So, further strength from here in the near-term should be approached with caution, but given the fundamental updates, we don’t think the current reading should scare anyone off the shares. Just don’t be too aggressive and understand that some consolidation to work off nearly overbought conditions would be a longer-term positive. (Jim Cramer’s Charitable Trust is long META, GOOGL. See here for a full list of the stocks.) As a subscriber to the CNBC Investing Club with Jim Cramer, you will receive a trade alert before Jim makes a trade. Jim waits 45 minutes after sending a trade alert before buying or selling a stock in his charitable trust’s portfolio. If Jim has talked about a stock on CNBC TV, he waits 72 hours after issuing the trade alert before executing the trade. THE ABOVE INVESTING CLUB INFORMATION IS SUBJECT TO OUR TERMS AND CONDITIONS AND PRIVACY POLICY , TOGETHER WITH OUR DISCLAIMER . NO FIDUCIARY OBLIGATION OR DUTY EXISTS, OR IS CREATED, BY VIRTUE OF YOUR RECEIPT OF ANY INFORMATION PROVIDED IN CONNECTION WITH THE INVESTING CLUB. NO SPECIFIC OUTCOME OR PROFIT IS GUARANTEED.
