AI Stocks Are on Sale. It Won’t Last

0:00 Inflation is hot in America. The US 10-year yield hits its highest since 2002. The Fed is going to have to raise rates and it could come this month. >> With all of that going on, how can the stock market still be within single digits of an all-time high? The short answer is that profits are growing a lot faster than the higher rates can pull prices down. But under the surface though, this past week was a lot rougher than the index actually shows. The S&P 500 only slipped about 1% but chip stocks fell about 9% and every one of the five stocks in this video fell more than 9%. I know that right now everything feels like it's on sale and that's exactly what I'm going to be using a filter for today's five stocks.

0:39 To make this list, a company's profits have to be expected to grow at least 15% next year and analysts have to have raised their forecast since July and the stock price hasn't kept up with those higher forecasts. So if you're new here, I'm Brian and I retired at the age of 46 after a career of making billions of dollars for companies like Target and Amazon and I put those same skills to work for my own investments.

1:01 >> Interest rates are to asset prices, you know, sort of like gravity is to the apple. When there are very low interest rates, there's a very small gravitational pull. Interest rates power everything in the economic universe. >> When a safe Treasury bond pays you about 5.3% a year for literally doing nothing, a stock has to work a heck of a lot harder to actually be worth the risk. That's why higher rates pull stock prices down and the pull just got a lot stronger. The biggest banker in the country just got asked on his earnings call whether the 10-year could actually hit 6%.

1:39 >> And so of course things like that can happen. We would never take it off the table. >> Thankfully profits are fighting back though. Since the end of June, expected profits for the S&P 500 are up about 10% while prices are only up about 3 and 1/2%. That means that the market actually got cheaper compared to its profits while it was still climbing. It's the same idea as my coiled spring video just a week ago with the rates pressing down and the profits are pushing up. Now going into this earning season, 72 companies in the S&P 500 told Wall Street to expect more profit than it was forecasting, the most since FactSet started tracking it back in 2006. Even the bubble crowd sees this.

2:19 So Steve Eisman, the investor that's made famous from The Big Short, has been warning that this AI boom could end badly. But listen to what he owns. >> Nvidia, God bless them, and I own the stock. >> He's the loudest one warning us about a bubble and yet he still owns the company that's sitting right in the middle of everything. So here's a simple question we should all be asking. What happens when those profits stop growing? So the slowdown is already baked into the forecast with profit growth expected to fall to around 3% by the middle of next year and history tells us that the market can usually handle that. So in eight of the last 10 times that profit growth had peaked, the S&P 500 was higher a year later and the real damage only really came when profits began to fall. And fair warning that most of these swing a heck of a lot harder than the market. So on the days this past year when the 10-year rate jumped the hardest, the S&P 500 fell about 1%.

3:12 Three of the AI hardware stocks in this video fell about 2.4 to 3%. So that's why I buy stocks like this in stages and keep the riskiest ones extremely small. Even after a week like this, the filter stays the same. And what I'll be showing you is where my system has each of those stocks on the signal ladder right now. The rungs of the ladder themselves barely move, but when prices swing this hard, a stock can fall right through one of them in a single day. So these readings won't settle down until the market does. My expectation is that a lot of these start moving back up over the next few weeks and I'll post the updated levels on Patreon as things settle. Let's go ahead and get started with the biggest company on the list.

3:51 Now I think we all know that Nvidia makes the motor of an AI data center, but a motor is not going to run without the radiator, the fuel line, and definitely not without a battery. And this company makes all three, the power and the cooling gear that keeps those chips from melting down. That company is Vertiv, and its CEO says the demand is not slowing down. >> Our global pipeline momentum remains very strong. And we expect another year of robust orders growth.

4:16 >> Vertiv's profit per share jumped about 60% last quarter, and it raised its outlook for the whole year across the board. The stock went the other way, down about 31% since the middle of July and about 45% off of its high. Part of that drop is on Vertiv itself. Sales came in light last quarter because a few giant projects took a heck of a lot longer than planned. When it reported this, the stock dropped more than 13% that day. And Vertiv has also been buying companies left and right, and the market doesn't exactly love the price tags that it's paying. The biggest one was a $1.4 billion deal for a company that makes power right next to the data center because a lot of those projects can't get power from the grid fast enough. So, if that works, Vertiv goes from just cooling the building to actually helping to turn it on. And the rest of it is just gravity because Vertiv takes the biggest hit of anything on this list when rates begin to jump.

5:07 Meanwhile, analysts have kept raising their numbers, and profits are expected to grow about 36% next year. And that forecast has gone up since July even while the stock just kept going down. Nvidia is also changing how power gets into the rack starting in 2027. And Vertiv's gear for that is due out by the end of this year, which lines it up for a whole new upgrade cycle. Before I cover some of their fundamentals, I need to point out that I get asked about the PE and the PEG ratios quite a bit. And you could go to 10 different websites and get 10 different numbers because some sites use the profit that a company made over the last 12 months, and others use what analysts expect it to earn this year or next year. The fundamentals that I reference lean forward looking because what a business is about to earn tells you a heck of a lot more than what it's already earned. To help everyone out, I put a cheat sheet in the link below that shows exactly how I calculate them. So, you'll know why my numbers might not match the site that you typically use.

6:03 For Vertiv, the PEG ratio is right around a one, which means you're going to be paying about fair price for that growth. On my signal ladder, Vertiv's price has dropped into buy a lot. And my investment system agrees with that call. The stock is also below its average price over both the last 50 days and the last 200 days. And the further that a price sits under those two lines, the tighter that the coiled spring is going to get wound up. Even with it sitting at buy a lot, the system's advice is to keep the position small because the trend has not yet turned. Vertiv is going to be reporting on October 28th, and I try not to buy or sell anything within a week or two of that report.

6:38 This is one where I'd consider just waiting until those numbers are actually out. And once they are, I would dollar cost average in just a little bit at a time. Now, here's an interesting part about the AI boom. While everyone is trying to figure out which AI stocks are going to win, there is another way to benefit from this shift. And that is learning to actually use these tools. Companies are investing billions into AI, and they're going to need people who know how to use it and how to work with it faster and solve all the problems more efficiently. So, it's not just about which companies win the AI race, but whether you're equipped to actually take advantage of it. And that's where tools like GPT-6 Astra come into play, which brings me to something worth actually me introduce you to the GPT-6 Astra Crash Course. It's a free 3-hour workshop at 10:00 a.m. Eastern, where you're going to learn to build your very own Astra workflows hands-on. As an official OpenAI partner, the Outskill team has exclusive access to Astra before anyone else, which means that they have tested its limits, and they've turned those learnings into a crash course. The link to sign up is in the description. Or, you can just scan the QR code on the screen. In the workshop, they're going to show you everything from writing prompts that get Astra to nail that output the very first try to getting better outputs while using fewer tokens. You'll also learn job hunting where Astra finds jobs that truly match you and tailors your resume for each one. And also investing where you can tell Astra the stocks, mutual funds, and IPOs that you care about and it alerts you the moment that they drop or they jump and it even explains why. These models are becoming so powerful and everyone should learn to leverage them with their investments. Seats are limited so I'd recommend grabbing they fill up. When you sign up, you'll also get the Astra power playbook with prompt formulas and workflows ready for you to plug and play. Let's go ahead and get into the next one and it's a company that keeps power flowing and the chips cool inside these data centers. Think of the breaker box and the fuse panel in your basement but just scale that up to a building just full of servers. Then add the plumbing that pumps the liquid right up to the hottest chips to try to cool them down. This is Invent and like a lot of other companies, they can't build fast enough.

8:37 >> Given the strong orders, backlog, and visibility we have with our customers on with the cooling demand, this expansion is not going to be enough. >> Invent sales jumped about 53% last quarter and it now expects to sell more than $2 billion of data center gear just this year, which is more than double what it did last year. When it first became its own company, things like data centers were only about 12% of its business and now they're close to 60%.

9:03 Since July, analysts have raised next year's profit forecast about 14% and the stock is down about 9% over that same stretch. All of it really happening in this last week. The profits are expected to grow about 27% next year. Now, on the downside, the biggest risk here happens to be competition. So, back in July, Schneider Electric got its cooling built into the design of AMD's new AI rack. So, Invent has some very large companies fighting for the same business and this is exactly where gravity shows up in a way that you can actually begin to see.

9:35 Invent just bought a company that makes data center power equipment for about $1.75 billion, and it borrowed a big chunk of that money at just over 6% interest. And I think it goes without saying that higher rates make that debt a heck of a lot more expensive, which is part of why the stock has really been just stuck in the mud. The Signal Ladder has Invent's price at buy a lot, but my investment system steps that down to a buy because even after this drop, the stock is still on the expensive end compared to its own history. The price now sits about 9% below its 50-day average and just above its 200-day. So, the spring is getting more compressed. Invent reports at the end of the month, so I'm just going to sit tight until after that, and then maybe I would start a small position afterwards. Hey, if you're getting any value out of my research or my YouTube videos, please, I would love it if you would consider pressing that like button so I can continue to grow. Let's go ahead and move on to a company that is all about quality control. Now, every AI chip and every stack of memory has to be checked before it ships. That's because one bad layer can ruin a chip that costs tens of thousands of dollars. So, picture a home inspector who checks every floor while the house is going up instead of waiting for it to be finished. That is where Onto Innovation comes in because it builds those inspectors. Its tools are in high demand for the way AI chips and memory get stacked and packaged together. Onto set a sales record last quarter up about 35% and its backlog of orders topped $1 billion for the very first time. It now expects the part of its business that inspects that stacking and packaging to grow about 80% this year. One packaging customer alone ordered more than $200 million of these inspection tools, and most of them don't even ship until 2027.

11:17 Onto is going to lay out its new long-term targets on December 17th. So, for me, that is the date that I am very interested in. Since July, analysts have raised next year's profit forecast by about 18%, and over the last month, 10 of them have raised it with not a single one cutting. Over that same stretch since July, the stock is down about 26% and profits are expected to grow about 44% next year, which puts its PEG ratio at about a 0.6, which is obviously below one. In other words, on paper, you're going to be getting that growth at a discount. And there is a reason for that discount. A big chunk of Onto's business sells to memory makers and a Chinese memory maker just announced a new chip design that knocked memory and chip equipment stocks way down. And sadly, that $200 million order that I mentioned rests on a single customer and Onto swings about three times as hard as the overall market. Onto's price has now fallen just below the bottom rung of my signal ladder. And when that happens, I don't buy more until I've gone back through the entire business. Now, it does still look solid with analysts raising their numbers and the biggest backlog of orders it's ever had. The price is a very different story because it's sitting well below both its 50-day and its 200-day averages. So, this spring is about as compressed as it's going to get. Onto is going to report in early November, so I'd let that report come out first. Then after that, I would keep it small and I would dollar cost average on the way down because, let's be honest, we never know where the bottom is. And with all of this memory news still shaking out, I don't think anybody knows either. Next up is the crew that actually builds the data centers that these chips end up in.

12:50 Think about a new restaurant going up where the walls go up fast, but nobody's eating there until the wiring and the heating and the cooling are all complete. This company is that crew for data centers, hospitals, and factories. And it even builds whole sections of buildings in its own plants and then it ships them to the job site already to hook up. That crew is Comfort Systems USA and its CEO is already talking about how good things look well into 2027.

13:14 Last quarter, Comfort Systems grew sales about 50% and its profit per share nearly doubled. Its backlog, the work that customers have already signed up for, is now about $14 billion, dollars, up about 73% from a year ago. And a big reason is data centers because technology customers went from about 43% of its sales just a year ago to almost 60% now. Analysts have raised next year's profit forecast about 11% since July, and everyone who's changed that number in the past month, seven of them, moved it up. The stock still fell about 16% in that time, and it lost about 15% in just the last five trading days alone. But analysts expect profits to grow about 23% next year, which works out to about a PEG ratio of around a 1.3. So, you are paying a bit of a premium here, but this is also a business that's beaten Wall Street's estimates four quarters in a row. Now, where this could go wrong is that same data center bet. If the big tech companies slow down in any way and how fast they're going to build, that $14 billion backlog is what's really going to get tested. And most of its jobs are also fixed price. So, if labor or materials cost more than it had planned for, that comes straight out of its profit. Comfort Systems price is sitting in a buy on my signal ladder, and my investment system dials that back to just buy a little. Compared to its own past, the stock is still priced near the top of its range, and the system's estimates for the business are spread pretty wide. So, it knocks the call down two full rungs. The price is about 11% below its 50-day, and it's about 8% below its 200-day. So, this spread is also getting tighter. With its earning report only a few days out, I would obviously wait for those numbers and then begin to build my case afterwards.

14:52 Now, we'll move on to my long distance play, which is the smallest company on the list, and it's the most volatile stock in the video. Every single AI model has to be taught, and then it has to be tested. So, think of this company as the tutor and the exam board for AI. It writes the practice problems and grades the answers for some of the biggest tech companies in the world. That company is Innodata, and its last quarter was a record. Innodata's sales jumped about 58% last quarter, and profits are expected to grow about 27% next year. But the stock is still down about 15% since July and it's trading at about half of its June peak. It's also building a lab to capture motion data for robots and management sees an opportunity testing AI models for the government. Neither of those are going to show up in the numbers quite yet, but that is where the next leg of growth could start to kick in. Now, the biggest risk is how concentrated the business is because just two big tech customers make up about 71% of Innodata sales and if either one of them pulls back, well, the whole story begins to crumble. And I believe that being independent is what really protects it here. Scale AI is one of Innodata's biggest competitors and last year Meta paid about $14 billion for roughly half of it and it hired Scale's CEO to run Meta's AI push.

16:07 Within days, Google, Microsoft and XAI were all reported to be backing away from Scale and it's no surprise because nobody wants to be training their models on data from a company that's half-owned from a rival. Innodata, on the other hand, isn't owned by any of these companies, which makes it completely neutral. However, if Innodata ever got too wrapped up with any one of these giants, it would have the exact same problem as Scale. Now, Innodata also changed CEOs at the end of September and it set up a plan to sell up to $300 of new stock, which really does put a lid on rallies. This particular stock has moved 10% or more in a single day about 15 times this year. That is exactly why it's not going to be money that you're going to want to touch anytime soon.

16:51 Innodata's price has now slipped below the bottom rung of my signal ladder. So, I went back through the business and it still checks out, but there is no clear setup right now. The stock is about 10% below both its 50-day and 200-day averages. So, the spring is already very tight and sadly it could slide a little bit more before it's ready to pop. They'll be reporting in early November, so I'd obviously hold off until then. But for me, this would only go in the riskiest slice of my money. So, if you do want to start, I would keep it very small and add just a little over time.

17:22 This whole video started with gravity, and I think Buffett is absolutely right that interest rates are gravity. If the Fed raises rates again at the end of the month, then gravity is going to pull a heck of a lot harder on the companies in this video. But, gravity only really wins when profits actually start falling. And right now, analysts are raising their numbers on all five of these companies, and every one of them reports over the next few weeks. That's where we find out if the engine is even still running. So, my plan is to wait out the reports that are right on top of us. I'm going to buy in stages, and I'm going to keep the riskiest ones extremely small. Hey, I hope you were able to get some value from today's video. As always, thanks so much for watching.