3 AI Stocks to Buy Before the Next Wave of Growth

Hey everyone and welcome back to another episode for you today. In today's episode, we are going to talk about some AI names, some very popular names actually in the AI space. I'm joined here with Rachel. We're going to cover three AI stocks I think are worth to put on your watch list. Maybe they're already on your buy list. Without further ado, Rachel, what's your first pick? Well, my first pick for today is a name that I think everyone that is following AI stocks is talking about right now. And that of course is Alphabet, Google's parent company. You can't go through a video about AI stocks without talking about Alphabet. And you know, for all of the talk around AI, so much of what Alphabet is doing with AI is really monetizing its core businesses. Of course, the advertising business and the cloud business. It still runs primarily on search, YouTube, its broader ad network together make up the large majority of its revenue. And that's generated every time someone clicks a search ad or watches a monetized YouTube video. And then of course there's Google Cloud. This is the third largest cloud provider in the world alongside Amazon's Amazon Web Services, Microsoft's Azure, and it's developed its own family of AI models, Gemini. And that now powers both a consumer chatbot app, which I'll talk about in a little bit, and then of course the AI layer that's built directly into Google search itself. And you know, it's interesting because this is far from being a pure AI infrastructure company. Some of those businesses could, you know, live or die based on whether their specific technology takes off. What Alphabet has done is sort of run the AI experiment from multiple angles at once. A product feature inside search, a standalone consumer app. of course its enterprise cloud offering and all of this is funded entirely by a balance sheet that is deeply profitable cash flow producing and based on its uh decades old advertising business that doesn't really need any of these bets to work in order to keep growing. Now for a few years there has sort of been this dominant bare case on Alphabet. this idea that AI chat bots like chat GPT would just eat into Google search the way that search once cannibalized print advertising, right? This idea that oh, Alphabet's reign is over. Well, that hasn't happened so far. Search revenue has continued to grow. And that's even after Alphabet rolled out AI overviews. There's a full AI modes uh inside search itself. And essentially what I think we're seeing is that Alphabet has been able to layer AI onto its existing business rather than cannibalize it. And that's also really accelerated the growth story for Google Cloud. And just to put a few numbers for that, in the second quarter of 2026, Alphabet's overall revenue grew 24% year-over-year. Search and other advertising revenue grew 17%. Google Cloud revenue grew 82% year-over-year. That's an acceleration from already, you know, generous prior growth. The cloud backlog, which is contracted future revenue, reached $514 billion. The cloud business is seeing incredible incremental progress from the incorporation of AI. The Gemini app is reaching 950 million monthly active users and daily active users actually tripled over the past year. Then there's Gemini Enterprise. That's Alphabet's AI product for businesses. It's now used by nearly 90% of the Fortune 100. And what's interesting is if you remember how the market reacted after Alphabet reported earnings a few months ago, the stock actually fell. And I think this is because investors were really focusing on how much Alphabet is spending on AI infrastructure relative to expectations. Of course, like many of the other players in this space, Alphabet has continued to expand its expected capital expenditures. It's investing heavily in AI infrastructure in its TPUs in data centers in growing its business eventually looking at that as going to be cash and profits that flow back into the business and that's why we are seeing some pressure on the growth story on the profitability in the near term but in the long term I think that's going to continue to give it a lot of optionality across its consumer uh and enterprise businesses. Now, obviously, there's always the regulatory risk in the background for Alphabet. This is something that Alphabet has overcome time and time again, but it's always something to bake into your thesis there. For me, the bottom line is Alphabet has spent the last few years essentially just proving this idea that AI would kill search. And in fact, it's optimizing its search business with AI. It has its own proprietary chip business that it's building out. it's uh tensor processing units or TPUs which is helping to reduce its reliance on outside suppliers and even though most of the growth story is driven by search advertising by Google cloud there's that other bet segment there's some of these key investments that Alphabet has of course Whimo is the one that's probably one of the most uh prominent with its expansion into autonomous vehicles who is by far leading the autonomy race in the US at this point it's rapidly expanding outside of the United States as well. But then there's other areas that Alphabet's investing in like isomorphic labs, AIdriven drug discovery. There's a lot of different ways in which you can gain exposure to a lot of very exciting um industry trends by investing in this one business. But importantly, this is also a growth story that's built around real profitability, cash flows, and steady revenue growth. And again, you know, the dominant leader in search advertising and one of the key leaders in cloud. So, I think, and I say this as a long-term shareholder of Alphabet, I think Alphabet remains a continued uh compelling buy in the space if you're looking at AI stocks, and it's one with a valuation that has trended towards being more favorable uh than a lot of the competition. So, one to consider if you're looking to add an AI stock to your portfolio. >> I would agree with what you said. I would agree. I think uh Amazon and Google I think is is the two names I uh I spoke about last week and yeah Google is to its core a very very profitable business. Now it is true that maybe in the future maybe in the future search does get disrupted but search as a percent of the overall revenue has already come down and is now closer to 50% of it just because the other side of the business especially Google cloud is just growing much much faster and becoming extremely extremely profitable uh for them. So Google I think is always one where I'd put it on on any list out there. Um, mine is a company that is much smaller. It's Nebus. Um, some of you have have been following me for a while. You know, Nebus is a name that that we like a lot. Been covering the for for for a while. It's Nebus, the team behind it is the same team that built Yandex, which is sort of a Google of Russia. Since we talked about Google, we will will say this. um and they had to spin off spin off their business or or just move away from it. So they built Nebus headquarters actually in Amsterdam. So you can call it actually a a European a European success story and it might even be the first big European uh publicly traded company that could reach $und00 billion in market cap for the past uh decade plus. Now Nebus the business early on is quite simple. They rent out AI compute power. They sold out the years ahead, but they've been making a couple of acquisitions. They're working on making sure that the business is not just selling bare metal to the likes of Microsoft and Meta. We'll talk about that in a bit, but it's to make sure that they do have a very sticky software platform that of course makes more money from every GPU it owns and runs and then rents out as well. Now if you look at the numbers Q2 by the way Nebus is a company worth around $60 billion. So when you look at the numbers Q2 revenue $582 million yes it is trading at a very high premium but revenue is up 454% year-over-year and up 46% from uh Q1. The AI cloud business alone grew 514% year-over-year. ARR and they and many others they calculated basically the revenue from the last month. You multiply that by 12 that hit $3 billion up from $1.9 billion just 3 months before. It is profitable on an adjusted basis. Group adjusted EIDA was $236 million with roughly 41% margin. Now going forward of course they project AR R to grow also quite rapidly between 7 to9 billion. Of course all of that growth how do you reach it? You have to spend a lot of money. Capex is expected to be also quite high 20 to$25 billion. My early predictions for 2027 is that we're going to see capex be much much uh bigger than that in 27 but ARR towards the end of the year for me could also reach 30 to40 billion which again means the last month multiplying that by 12 which you could say is maybe just next year's revenue or in this case 2028 revenue but it's not really it's just a simplified way to say this because the last month. Let's say you're generating $100 million, but two months afterwards you might be generating either 70 or maybe 150. So multiplying one month by 12 and making that the whole story to me doesn't make that much sense. But in the meantime to accelerate the growth to make sure that they do have money to well spent on all of this buildout, they have a 40 around a $40 billion contracted backlog. Microsoft being a big part of it, Meta being another big part of it. Now, we don't fully know at 100% all of the details of those deals, but those deals are definitely not high margin uh deals for them, which is okay because they use those deals to again solidify the core business, making sure that they can spend to grow the business faster than others. The future of this business is not to have huge contracts with the likes of Microsoft, Meta, maybe potentially Google or whatever for the foreseeable future at super low margins. No, the future of this business is to try and create another AWS, another GCP where you do attract more enterprise customers where the margins are much higher. Yes, the headline might not say, "Oh, they got a $20 billion contract for Microsoft or for Meta, but the headline should be they're generating tens of billions of dollars in revenue at high margins." That's that should be uh the headline here. Now, we know that there is way more demand than supply right now across the board. The co said that Nebus could sell its entire plan 2027 capacity today on basic terms, but it is holding back some of it because prices are higher. They actually did a capacity auction and that cleared 15% above the highest price it has ever charged for the Nvidia Blackwell chips which and I think since then they've also increased prices across the board. So if you can generate previously I think 10 to 15 million per megawatt and that goes to 20 to 25 and some of it you can sell for 40 million then it makes sense that you do not want to sell out all of your capacity because if you can charge a premium for x amount of time again it helps you just accelerate the business because the business is capital intensive. So you need that cash early on to make sure that the future of the business is secured. Now Debbius is also a company that has made quite a lot of acquisitions especially more recently and all of those acquisitions are made for one key product which is called the token factory. It's a managed inference platform meaning companies pay to run AI models in production instead of just renting out the the raw chips. and they've bought Tavili earlier this year which gives AI agents live web search an agent that books travel or does research needs fresh information not just what the model learned in training now developer gets the compute and the search in just one place so they have one less vendor and one more reason to stay the next one is Agon AI they make the model itself run faster on the same hardware so faster models means more answers per GPU per hour which also means more money uh that can be made by Nebus. The next one is Clarify. They optimize the whole system around the model how jobs get scheduled across thousands of GPUs. They tune the engine and clarify which was the other acquisition tunes uh the amount of traffic. And then the more recent one which is literally happened this week is called Infrise and that's a company that didn't even exist a year ago. So they acquired this company in I think around $150 million. The people that I think founded the company were previously I granulate which is a company that Intel acquired but then closed down its business in the whole cost cost cut um environment that they were in. So what does Infarise do? Again, it comes back to the whole making sure that token factory is the one product that gets clients in but make sure that these clients and customers they stay inside. So here a GPU sitting unused between workloads is of course losing money. There is this thing called cold start just like with vehicles. I'll be honest, I didn't even know this thing was a the same term that could be used in in GPUs and data centers, but it is a thing. So if you can squeeze out the cold start, if you make sure that the cold start is not 20 minutes but let's say 1 minute, you are here as well saving quite a lot of time for the customer and then in turn saves a lot of time for Nebus which means again Nebus can make sure that the utilization rate per GPU goes higher and higher. If it goes higher and higher, they'll be generating a lot more revenue than most of their competitors. They can then rent it out also at better prices because the utilization rate is just so much better, which makes their competitive advantage bigger and bigger. Now, of course, all of that is great, but here as well, execution is key. permits to build data centers. We've we've heard a lot of issues with delays, permit delays, lawsuits, etc., etc. So, all of that of course needs to be considered. It's not just a Nebus issue. It's an issue maybe across the board, but I do think that Nebus is very well diversified where and how many data centers they have. United States is quite big, Middle East is getting bigger and Europe is getting bigger as well. So I do truly think and I think their chief revenue officer said more recently that they do have a a path to reach $und00 billion in revenue in the not too distant future. Of course, execution is key. But when you look at the past execution, the trajectory, the fact that the majority of the engineers there already built a huge successful company before in Yandex is is a very very big competitive advantage, especially in in this space where yes, moving quickly, having the capital, but especially having all of the engineers that are moving in sync is is huge to me. And that's why I put Nebus here on on this list. >> Yeah, it's been a certainly an interesting company uh to watch and you know they've been Nvidia has been a huge investment partner. They've invested I think at this point about $2 billion into the company. Their commercial backlog is almost $40 billion in customer commitments and obviously stretched out over several years. So it's uh definitely one of the moonshots in this space so to speak. uh but it's become an increasingly key player to watch I think. Um the final stock in today's video that I wanted to talk about is Arista Networks. If you're not familiar with Arista Networks, this company makes the networking switches and software that actually move data between the servers inside of data centers. So that kind of high-speed networking that connects thousands of individual servers and of course increasingly AI accelerator chips into a single computing system. So Arista networks, they don't make chips or run cloud infrastructure. It makes the networking layer that everyone else's compute and storage hardware plugs into and that's built on top of its own software platform that runs across its entire product line. So that means that customers have a single consistent operating environment really regardless of what switch model they're using. Now we we talk a lot about AI training clusters. This isn't, you know, just a single uh computer. This is thousands of individual chips that all need to act like one. So if the network that's connecting them is too slow or it's unreliable, those expensive chips can sit idle waiting for data instead of actually computing. And that means the networking layer directly determines how much value a company actually gets out of the GPUs that it already paid for. And that's what of course makes Arista Networks such a relevant player in this space. Now training a large AI model, it doesn't just require powerful chips. it requires thousands of those chips to communicate with each other near instantly. Uh and the networking technology connecting them has become one of the major bottlenecks that we're seeing and this is as AI clusters have scaled up. So Arista is aiming to be one of the key players to undo that bottleneck. It has positioned its Ethernet-based networking as a more open and flexible alternative to other proprietary interconnect technologies. uh Nvidia's Infiniband as an example and that positioning has been paying off. Management has been saying in recent earnings reports that AI networking demand has moved to an outright supply bottleneck and that Arista now has supply constraint across chips, optics, components. And what's interesting is when you look at competitors like Nvidia's Infiniband networking, they lock customers into Nvidia's own ecosystem end to end. So Arista's pitch to hyperscalers is essentially, hey, you can build an AI cluster that performs just as well using open Ethernet standards. You're not going to be locked into a single vendor's proprietary networking tool. And that is a big part of why Arista has been able to land deals with companies like Meta Platforms and Microsoft that could have technically gone all in on a single vendor's full stack. But as these bottlenecks are becoming apparent, it's become increasingly clear that the hyperscalers also want to diversify the vendors that they're working with. And that's of course been a really important continued part of the growth story for Arista Networks. So in the second quarter of their fiscal 2026, they reported that back in August, revenue topped $3 billion for the quarter for the first time ever. That was up about 38% year-over-year. Management raised their guidance for 2026 to 12.6 6 billion. That would be growth of around 40% compared to fiscal 2025. They raised their AI specific revenue target to about $3.6 billion for the year. And the company has now assigned more than 100 cumulative AI fabric customers using its Ethernet switch platform. And that's just compared to a few years ago. You go back to 2024, if you've been following this company for a while, there was just a handful of early adopters. So they've really really ramped up the growth story there over the last couple years. They estimate that Arista Network's total addressable market with its Ether Link switch platform and all of these AI customers could be upwards of $105 billion. So now Arista is working directly with a lot of the major AI hyperscalers. We're seeing a shift towards Ethernetbased AI networking that's really benefiting the company. And Arista has also introduced its next generation 1.6 terabit per second AI fabric platforms and liquid cooled switching options. Those are built directly for training for the largest AI training clusters. That's positioning them well ahead of the next wave of cluster scaling, which means they're not just serving the current generation deployments. They're looking towards the future and the growth story. One of the things I like about Arista Networks is the fact that in addition to obviously sitting at a genuine bottleneck in building these largecale AI infrastructure, they've also built a hardware and software reputation over a period of years. They have a net promoter score of 89. That is among the highest of any enterprise technology company. So that means that as a lot of these key players are trying to address these bottlenecks in the AI buildout, Arista is one of the trusted default choices for that job. Um, and of course a net promoter score figure is also a really strong signal for an enterprise hardware company specifically. You know, brands chase uh net promoter scores like that. And so this has been something that has been a continued part of Arista's story for a while now. Obviously the AI related revenue piece there's a customer concentration risk which of course I think is important to point out you have a relatively small number of the hyperscaler customers that are making up a large and growing share of that AI specific business. It is also the case that these hypersklls hypers scale customers are the key players that are driving most of the AI spending that we're seeing right now. uh but it's something to note and of course there's always supply chain constraints that can pause the growth story from quarter to quarter but Arista has become one of the more essential if not under the radar companies and AI infrastructure that networking layer that has to work for a training cluster to function at all and I think the risk you know it's not whether the AI networking bottleneck is real it's how concentrated that opportunity remains among a small number of giant customers I think for the years ahead this is going to continue to be a key bottleneck that companies like Arista Networks can help address. It's a well-run business and I think it's one to at least add to your watch list if you are building out your position in AI stocks. >> Yeah. And if we actually look at the last 12 months revenue generated, it's just over 10 billion. And if we look at what the market expects this company to do two two and a half years down the line, expect to double expect to double that revenue. So there's definitely a lot of demand there. We've seen that with Broadcom. We've seen that with Nvidia. Um, so another crucial area, crucial part in this whole AI story with uh with Arista Networks. Um, with that said, let us know what you think about these three [music] names down in the comments section below. If you have any other AI names you would like us to cover, let us know as well. As always, like, subscribe, and we shall see each other in the next one. Bye-bye.