4 AI Stocks to Buy After the Market Sell-Off!

Well, we're in a time where it seems like the market is repricing a lot of AI stocks. And while that may or may not be a longerterm trend, it's certainly the case that not all AI stocks are created equal. In a previous video, we talked about the ways in which the market is repricing, some of the businesses that are most reliant on corporate debt that are under pressure as bond yields and interest rates are rising. But there are still a treasure trove of well- capitalized, profitable, cash flow producing AI businesses that might actually be trading at a little bit more of a discount right now after the recent route in the market. So Neil and I are going to be talking about four AI stocks that we think are quality buys in this era. >> Neil, good to see you. How are you doing today? I'm doing much better than my portfolio. I'll tell you that much. So, uh although that's not that difficult, I guess after after the last couple of days in the stock market, but uh yeah, we'll talk about some good businesses. Although my first pick does not generate any free cash flow as of right now. Um but I still view this as a as a quality business. It's a name I've been covering for quite a while. I own it and that's Nebus. So, I know I know what people will say. Oh, high risk, right? Cuz it's it's in the Neocloud space. Although Nebus themselves do not view the company as a Neocloud company. They view it as a AI cloud or nextG hyperscaler. But okay, everybody can come out with their own definition. But Nebus alongside all the other new clouds out there have been hit quite a lot. We've talked about it yesterday actually. Cororeef, Oracle, uh they've been dragging down actually uh the space. So, of course, Nebulus being in the exact same industry gets dragged down as well because of the fears of or still the same fears of maybe overspending in AI, what's the return on investment? And second of all, all of these companies that still need to grow quite a lot, but already take on quite a lot of debt, although core with way more than Nebus, but they're definitely not profitable as of right now. And so, that's why you're seeing these companies take a hit. Nebus in my opinion is still bestin-class at what they do. For those that don't know Nebus, the same engineering team, management team was previously in charge of Yandex, which was the so-called Google of Russia that to divest and they created that company a couple of years ago, headquarters in Amsterdam. So, they have a huge presence in Europe, which in my opinion is a very very big advantage to them. But they also have a big a big exposure in the United States in the Middle East. And I think there are some rumors that they're going to grow even more in Asia. But NEIS as of right now on paper it seems like it's super expensive. Why is this company worth close to $50 billion barely generating any revenue as of right now? Which is true. They're going to go from last 12 months revenue $1.3 billion to over $3.3 billion the end of this fiscal year to then close to 12.2 billion next fiscal year. Although I do think that number can be much bigger and then over $23 billion in fiscal 28. So it's definitely a company that is still expected to grow super quickly. Now that's great. But what about profitability? Because again the buildout costs a lot of money. Well, the thing that we know right now in the market is that the price of GPUs, the price of megawatts keeps going up. And Nebus has done something very smart, which is something that a lot of their competitors have not done is that they've taken yes some big contracts with Microsoft with Meta. Those are usually very low margin contracts, but this allowed them to move much quicker and raise capital at at a let's say cheaper price. This also means that they are ready for what is coming right now. They can rent out megawatts at a higher price, which also means that they don't have to be sold out. A lot of companies that say we're sold out for 27. Nebu said we could have been sold out for 27, but we decided not to because we know that we can price at a premium. Prices have been increasing if certain companies want to have only 3 months of allocation. Okay, we'll charge them a little bit more, which means that our margin profile suddenly improves quite a lot, which of course helps the company build faster and not need doesn't need to raise more capital, especially not in this uh rate environment. So, we're looking here at a company that in my opinion is best-in-class, continues to grow quite quickly. You'll see the margin profile of the company actually improve over time. The fears here across the board are a bit overblown, but I understand I understand the fears as of right now because we're talking about super big numbers, high growth rates for companies that are just let's say just starting off. But I do want to put Nebis here back back on the list because we are looking at a company that could and that's according to management, not myself, could potentially generate hundred billion in revenue in the not too distant future. That's not I'm I'm not saying this. Well, although I think it's going to happen, it is part of the executive team that have said it I think recently in uh in an interview. And look, if we're going to go and generate maybe in 2030 $100 billion in revenue, I think that's a big big jump up. Of course, here execution is key and a lot of things can happen between now and 2030, but that's that's with every other company out there. Well, it it's a good balance to some of the other stocks we're going to be talking about today in terms of uh risk profile. You know, for my first stock, I'm really want to talk about a trend that we've been seeing. So, the stock I'm going to talk about is momentum holdings. The first I think you have to understand that as we're looking at the AI buildout and there are many bottlenecks. One of them is something that the industry has essentially coined the term for called the copper wall. So for decades data inside uh clusters, compute clusters was moved between microprocessors using traditional copper wires. Now we're seeing AI model sizes scale into, you know, trillions of parameters. These huge arrays of thousands of high-end GPUs have to process and they have to synchronize data at the same time. And when you push that volume of information through standard copper cabling, that electrical resistance generates huge amounts of heat. It slows down processing speeds. It also consumes prohibitive amounts of power. Essentially, copper is physically incapable of keeping pace with the internal data demands of the cutting edge silicon that's required in today's AI race. The reason that's important is that solutions have emerged and the solution has been primarily the total transition to optical interconnects. These are systems that convert electrical signals into light and then they route that data via fiber optic transceivers at lightning speeds. Now moving data at the speed of light that allows the data centers to scale up their computing power but it also means that they can lower their power consumption and thermal strength. So this is a mandatory requirement for the AI architecture buildout and this is where a company like Lumenum comes into play. So Lumenum's business model, they're a B2B industrial hardware manufacturer. They design, build and sell uh photonics, specialized laser laser chips. Um the lasers and optical transceivers that turn electrical data from computer servers into beams of light. That's the core of Lmentum's business, allowing that data to travel through fiber optic cables at these high speeds. They also have an industrial tech part to their business and that's where uh they have applications like 3D sensing chip uh chips uh high precision industrial lasers uh for cutting and welding sheet metal or manufacturing electric vehicle batteries. But the cloud and networking business, that's the primary driver of Lumenum's growth right now. And Lumenum essentially sits at the middle of the kind of global tech supply chain. They have their customers essentially fall into three key areas. Number one is the cloud hyperscalers. So of course the tech giants that are building the AI data centers like Alphabet, Microsoft, Amazon. They need Lummen's optical systems to connect thousands of AI chips together in the way that I explained earlier. They've also have clients like traditional telecom giants and then even uh original equipment manufacturers that embed Lummen's lasers into retail products. So just to kind of understand what Lumen's business is and how it works, you know, they are seeing obviously tremendous growth in their business. Uh just their Q4 results. Net revenue was up over 100% year-over-year. they continue to kind of essentially blow past Wall Street's expectations. There was an interview that the CEO of Lumenum, Michael Hurston, conducted with Bloomberg television recently. Essentially, he stated that Lumenum is unable to meet about 70% of customer demand for its core AI products. This essentially means that also they can't address about 30% of their order log for alternative components all the way through 2028, which means their soldout product capacity is essentially already pushed out into early 2029. It takes about 3 years to build and qualify a brand new semiconductor laser fabrication facility. So they are very much insulated in many ways from a lot of their fast following competitors. They have a $2 billion strategic purchase commitment from Nvidia to secure components for the Blackwell chip rollout. And of course, they've also been fortifying their long-term market position in a few different ways. They acquired a company called Cloud Light that has really accelerated their high-speed transceiver packaging abilities. There was an event that was held recently called the OCP Global Summit. While there, Lumenum de demonstrated essentially their pioneering optics technology alongside industry giants like Corning and Astera Labs. And this technology essentially is aiming to replace the standard motherboard copic uh copper traces with lightweave conduits. What this means is that Blumenum is looking to expand its addressable market to the internal chipto-chip architecture. So, it's a really interesting business. We've seen a little bit of a pullback in the price just kind of broadly that's happening across the AI trade, but I do think that this is a really really um fascinating play within the broader AI buildout. they continue to see some fluctuations in terms of their bottom line growth just kind of given the mechanisms of this business but you know their backlog the purchase commitments they're seeing the inability to meet so much of their customer demand because the demand is so high and then just where they sit in the value chain in the AI buildout I think Lmentum Holdings is an interesting one to watch >> yeah and similarly to Nebus we have Nvidia backing them and investing in the company as well so you in pretty good position to be in. Now my second pick is actually some call it the anti- Nvidia which is which is Broadcom. Now why is it called the anti- Nvidia? Because well every the big customers that go to Broadcom are of course huge customers of Nvidia but they want to diversify away uh from Nvidia and not rely too much on Nvidia chips and so they went to Broadcom. When you look at Entropic, OpenAI, Meta, Google TPUs, they go to Broadcom. They design, they co-design those chips with Broadcom to make sure that yes, Nvidia cannot, let's say, squeeze every dime out of these companies with their chips and they rely less on the Nvidia architecture and more on on their own, which in turn can make their internal workloads of course more efficient. The return on investment there can make more sense. Does that mean that Nvidia will not continue to grow? Well, I still believe that Nvidia will stay the number one player in the field for many, many years to come. But Broadcom, again, based on what the management has told us, they're looking at $30 per share of EPS, $30 of earnings in 2028. As of right now, this is a company that's trading at around $340 per share or so. So if you want to put a small multiple on that 2028 target, it's really not that expensive for a company that is expected to grow very quickly. They told us that AI revenue only AI revenue, not that's just a segment which is the largest segment of the business, but it's not the whole business is expected to double next year to $115 billion and then they see doubling again to $230 billion the year after that. To me, when management tells me this, of course, management needs to have a good track record, otherwise they can make up any numbers and then they won't deliver on it. But if they tell me already today, and it fits the whole story of the whole industry that there's just so much demand, they cannot supply enough, then we're looking at a company that's going to generate two over $230 billion two years from now while still growing across the board. Margins of course are very good. This is a very unlike my first pick, this is a company that does generate billions and billions of dollars, tens of millions of dollars in a free cash flow. Yes, this is also a company that is participating maybe in the circular financing uh game, but we've spoken about that I think in in past videos as well. I just see this as them viewing an opportunity to accelerate the industry. They have the means to do it and so this is what is happening right now. But Broadcom, yes, it has risks to it because it does have big clients like Antropic and OpenAI and it's something that we spoke about in in Friday's video is what happens if those companies do not grow as fast as expected. How does that impact a company like Broadcom? It is a risk to know, right? But it also has customers like Meta, like Google where I don't think there are issues and others as well where I don't think there are issues of not meeting their commitments there. But purely focusing on Broadcom's business, I do think it is is very very well positioned because I do think that there are going to be these big companies and maybe more in the future where yes, they would like to develop more and more of their own custom silicon. So Broadcom, Marvel being another name, but Broadcom being the biggest one there is definitely going to benefit from it. And then the question becomes can OpenAI and Entropic really grow as fast as expected. They don't have to make sense valuation wise per se. They just have to make sense business. The business has to make sense. like if they go public at the $2 trillion valuation, but they're only going to generate, let's say, $100 billion, then $200 billion in revenue, it's still $200 billion. So that's still good to meet the commitments and grow, but it might not be good enough for the stock to be trading at $2 trillion. But again, that's not my problem as a as a Broadcom shareholder. That's probably a problem for Entropic and and OpenAI shareholders. But yeah, I do want to put a broadcom in in this discussion because it's also trading at I think forward PE is closer to 20 times 21 times right now for a company that's expected to grow this quickly and is already quite profitable in this market. I I view this uh as a favorable riskreward pick. >> Yeah, I really like Broadcom. I think they're a quality business and the market's certainly been a bit hard on the stock recently. The final company we're going to talk about today, my stock is a business called Coherent. And uh I talked earlier about the AI value chain. This spans everything from semiconductors to networking and physical infrastructure to cloud compute to large language models to software applications. And so you can find a lot of companies within each of those individual buckets. Coherent sits at that critical networking interconnect layer. Again, similarly to Lmentum, operating within the continuous bottleneck that physical AI hardware infrastructure is facing. You know, when Nvidia, for example, designs GPUs, the physical challenge of modern AI cluster scaling is data movement. And if it can't move between thousands of GPUs fast enough data, the entire system stalls. So coherent is another one of those key businesses acting as a link that translates the you know computer processing into hyperfast low latency light signals. So their business model is again built on kind of controlling the entire physical stack from growing raw compound semiconductor crystals in a lab to building the final high speed optical transceivers that power a lot of modern infrastructure. So, Coherent owns specialized fabs that produce what are known as engineered substrates. This includes indium phosphide, silicon carbide, and gallium arsenide. And what are these is? They transform these raw wafers into laser engines essentially. So, they generate revenue through volume-based hardware sales. Their components have to be customtailored and qualified inside customer systems over months or years. So this creates a very predictable recurring revenue as well as just you know repeat purchase dynamic. Coherent has especially over the last couple years been really aggressively divesting its non-core assets to double down on its data center and communication segment. This is something we've seen broadly within the space. Coherent has been no exception. So its data center and communications segment accounts for about 75 to 80% of its total revenue. Um that's most of the business at this point. The remainder of its business serves uh industrial manufacturing lasers like precision cutting and EV batteries and consumer tech. Uh but most of it is the data center business. And coherent has also shifted its strategy towards selling integrated optical platform architectures. It has its photon link platform for example and this bundles lasers, optics and fibers into a single plugandplay assembly for its clients. And one of the things that's also interesting as well is as AI clusters are kind of outgrowing traditional copper cables, they're pivoting towards what's known as co-ackage optics and near package optics where these laser fiber optics, they're mounted directly onto the same silicon substrate as the compute switch. And of course, you have the backing of the big companies like Nvidia and others that are key to coherence growth story. Now, management has said in recent earning calls that the primary ceiling on their data center revenue, which as I noted now makes up the bulk of the bulk of the business, the the ceiling isn't assembly, it's not testing capacity, it's raw internal laser output. So, Coherent produced is producing 80% more indium phosphide lasers than it was a year ago. And it's currently on track to double its internal output uh within the next few months. And so we're continuing to see also that it plans to double that again to meet the continued hyperrowth cycle. This is a company with a very diversified enterprise uh footprint. You know they they generate revenue across a global portfolio of you know industrial manufacturing lasers advanced engineering materials. Their revenue base is about $7 billion as of fiscal 2026. So they have a nice capital cushion there. And Coherent is also poised to be one of the first movers in the upcoming co-packaged optics architectural shift. Management has said that they plan to start re uh recognizing meaningful revenue from this business uh within the next you know quarters that would place them broadly ahead of the industry curve. It's a really really interesting business. You know, the other thing I will note is that the AI industry is looking at an engineering transition from 800 gigabit optical networks to the next generation 1.6 terabit transceivers. And at that 1.6 terabit networking speed, the power consumption of a standard data center uh module threatens to really overheat clusters unless operators are utilizing coherence indium phosphide uh modulators. So, we are continuing to see that growth from this business. You know, revenue growth is significant. Their backlog is significant. I do think that they're broadly well insulated from a lot of the broader credit fallout uh that we've been seeing in the space. And of course, they're working with some of the most well- capitalized companies on Earth as their clients. So coherent another stock to consider within the AI infrastructure buildout and looking specifically within that infrastructure bottleneck. Well, four very different AI stocks for today. Nebus, Momentum, Broadcom, and Coherent. Let us know in the comments. Are you a buyer of any of these stocks? Do you already own these businesses? What other stocks do you think we should be talking about as we continue to look at the AI revolution and buildout? We'd love to hear your comments. Please like and subscribe and we will see you in the next video.