2 Hidden Businesses to Watch as AI Keeps Heating Up

0:01Matt Frankel:Growth is accelerating for the world's largest chip maker. You're listening to Motley Fool Hidden Gems Investing.

0:11Matt Frankel:Welcome to Motley Fool Hidden Gems Investing. I'm Jon Quast and I'm joined today by Fool contributors Matt Frankel and Rachel Warren. We're going to take a question from the mailbag later in the show about selling a stock for personal reasons. But first, I wanted to hit this news from Taiwan Semiconductor. That is ticker symbol TSM. Now, the reason that I want to talk about this is not because it's a hidden company. I know I'm going to get skewered for this, but it's a$2 trillion company. Okay, it's large. I get it. It's not hidden. But I think that some people still underestimate how important it is.

0:43Matt Frankel:The company fabricates over 90 % of the world's most advanced computer chips, and it's an estimated 60 % of all computer chips. So it's a very important company. And when we see things with its business, it's really a leading indicator of global AI demand. And so I did want to highlight TSMC today because AI stocks have pulled back in recent weeks. I think some investors are saying, listen, the trends have peaked. The downtrend is coming. But this company just reported monthly numbers and it throws cold water on the fear narrative. Rachel, walk us through some of the important things we need to know.

1:17Yeah, I mean, you can look at trends in terms of looking at how share prices are performing, but the numbers are what really tell you what's happening behind the curtain, so to speak. And TSMC is, if you want to think of it this way, essentially the whole factory floor for the entire AI boom. And that's why their monthly sales report, their first half sales reports are really important reality checks for the market. They are not seeing a slowdown. TSMC just announced that June revenue jumped nearly 68 % compared to last year. And why is that so important? Well, June sales usually dip because of normal summer seasonality, but this year they actually rose from May and it shows that demand isn't just steady, it's actively accelerating.

1:57It very much pushes back against the narrative that the AI buildout is losing steam. And I think that's something that's really, really important for investors to pay attention to.

2:05Matt Frankel:For sure. And this is one of those companies that does give us kind of these periodic monthly updates on what's happening. And you look in April, 18 % year-over-year growth, May, 30 % growth, and now that 68 % growth that you just highlighted. So clearly an acceleration trend. But I'm going to zoom out just a little bit and ask you just to walk us through what does that look like? What's the context here now for the first half of the year? And what can we expect maybe for the back half? So for the first half of 2026, TSMC's total revenue reached 2.4 trillion New Taiwan dollars. In USD, that's about 75 billion dollars.

2:42That was up 36 percent compared to the same period in 2025. You know, I noted that 68 percent year over year increase in June revenue. The company also reported it was a six percent sequential increase. So up from the prior month. So TSMC, you know, they manufacture semiconductors for a wide range of applications spanning smartphones to high-performance AI computing systems. Their key clients are all the big US tech leaders, including the likes of NVIDIA and Apple and advanced micro devices. So TSMC is on track to generate over$40 billion just from AI chip revenue in 2026. That's about 25 % of its total revenue overall.

3:19And the reality is that the companies from Microsoft to Alphabet to Meta are locked into massive capital expenditure budgets. They're essentially waiting in line because TSMC's advanced manufacturing lines are booked solid. Now, TSMC reports their full earnings this Thursday. A lot of the focus is going to be on their capacity bottlenecks, whether they can raise prices for their core products. But essentially, when the primary supplier of the world's most critical tech shows this kind of momentum, it would indicate that the hardware cycle still has a lot of runway left.

3:47Matt Frankel:For sure. And to be clear, it is not just Taiwan Semiconductor. I also want to highlight some news here from Meta Platforms. And again, I realize that it is not a hidden company, but I think that we're going to hit here just a moment. If you hang with us, there are some underlying trends that we want to get to. But Meta Platform's making another announcement that just kind of throws cold water on the AI trend is peaking narrative. You look at its Hyperion data center in Louisiana. This is a very important data center for the company that is currently being built out. Originally, a$10 billion facility announced late 2024, then late last year, upping that to$27 billion.

4:24Matt Frankel:But the company coming out today, and it is increasing this project yet again, Matt.

4:29Jon Quast:Yeah, so there's a lot to unpack here. So Meta announced today, you're right, the Hyperion project. It's getting even bigger after two big increases that already happened. Even after that jump to$27 billion in estimated costs last year, the company bought an additional 1 ,400 acres of adjacent land. So we kind of saw this coming. So now we're expecting at least$50 billion in direct investment to one facility in a site that's now expected to be a 5 gigawatt project originally scheduled for two. And that's more than 3 ,200 acres in size. Just for context, that's roughly four times the size of Central Park in New York.

5:04Jon Quast:So not only that, but reports have indicated, this is from Bloomberg, that the total expected investment for this project, including things like the power infrastructure that has to go with it, could eventually be more than$250 billion. So the details really matter here. So the full buildout isn't expected to be completed until about 2036, so over a decade. So it's not like they're spending all of this money tomorrow from their balance sheet. And Meta, to handle the power problem, then you can't just connect the facility this big to the grid. It would collapse. Meta is paying for 10 new natural gas plants, 10 to support this one site and is contributing over a billion dollars toward local infrastructure like roads and to make sure the water and sewage infrastructure can handle this increased capacity.

5:44Matt Frankel:One of the things I want to ask here, Matt, is that we have seen so many what we're calling bottlenecks in the industry. And that is really essentially there's so much demand for certain things in these data center build outs that the demand is far outpacing supply. That is boosting the cost of things. I think we've seen this in memory in particular. The cost of certain things are going up because the price per component is going up. So it's not necessarily that the project is getting bigger. But is this a issue of the project is getting bigger or everything's just getting more expensive?

6:20Jon Quast:Well, it's both. It's not just a bigger scope. The cost of building data centers is soaring, and this is kind of a good tee up for our next section. The original$10 billion estimate for a 2 gigawatt facility, it was going to be a build cost of about$5 billion per gigawatt, just dividing one by the other. So now at$50 billion for 5 gigawatts, the cost per gigawatt has doubled to$10 billion. the power infrastructure for data centers, the specialized cooling they need, the other costs of building a data center, the chips that go inside, have risen along with the massive demand. So it's worth noting, however, that last year's$27 billion figure that you mentioned earlier kind of took this into consideration because that was still based on the original two gigawatt projection.

7:00Jon Quast:So the per gigawatt costs have actually come down since then for this round. It could be maybe they're expecting more efficiencies to come with the scale of a larger data center. Maybe their original estimates just weren't that great because they were based in 2024 when AI was like back at a previous generation. So there's a lot of moving parts here when it comes to the cost.

7:17Matt Frankel:So we've talked about Taiwan Semiconductor. We've talked about Meta Platforms. These are two of the biggest, most important companies in the world. But after the break, we're going to go under the radar. You're listening to Motley Fool, Hidden Gems Investing.

7:30Jon Quast:Whether you're trying to analyze market or business trends or dig through hundreds of pages of quarterly reports, the sheer volume of data can feel overwhelming. Lately, I've been using Claude's deep research feature as my ultimate thinking partner. I hooked it up to my Google workspace tools and it was able to run a reliable, comprehensive analysis across dozens of different financial sources in minutes, helping me spot long-term connections that on my own, it may have taken me hours or even days to see. Claude is the AI for minds that don't stop at good enough. It's the collaborator that actually understands your entire workflow and thinks with you.

8:05Jon Quast:Whether you're debugging code at midnight or strategizing your next business move, Claude extends your thinking to tackle the problems that matter. For problems worth solving, get started with Claude at claude.ai.fool. That's claude.ai.fool. And check out Claude Pro, which includes access to all of the features mentioned in today's episode. Claude.ai.fool.

8:28Matt Frankel:Welcome back to Motley Fool Hidden Gems Investing. We've just highlighted some trends here. Meta expanding the data center expenses that it is putting out. We're looking at Taiwan Semi showing the accelerating growth rate. In light of that, in light of these things, we want to highlight some hidden companies. And so this isn't necessarily a list of stocks to buy because we don't give personalized investing advice on this show, but we are Motley Fool Hidden Gems Investing. So we do want to go under the radar. So I've asked Rachel and Matt to give us a company that a lot of investors don't know about, but that they should.

9:04Matt Frankel:I've asked them to explain what they do to make money and what could go right in light of the trends that we just looked at. Matt, you're up first this time. What is a company that many people don't know about, but they should?

9:16Jon Quast:If you listen to the podcast frequently, you've probably heard me talk about the AI picks and shovels as my favorite way to play the trend. So one that I'd like to bring to the table is Comfort Systems. The ticker symbol is FIX, FIX. It's one of the largest heating, ventilation, air conditioning, plumbing, electrical contractors in the United States. And not for like your house, they specialize in commercial properties. They've historically served health care systems like large hospitals, manufacturing plants, and other really large scale commercial property types. Over the past few years, the company has smartly and through necessity pivoted to AI data centers.

9:49Jon Quast:And to say it has paid off would be a real understatement here. I mean, their revenue grew 56 % year over year in the most recent quarter, and it wasn't hospitals that did it. Their earnings per share more than doubled. The backlog grew by 80%. Data centers need enormous amounts of cooling and precise cooling and electrical work. And that's really becoming true the more sophisticated that data centers come. Data centers now being built are generally liquid cooled. They used to be air cooled just because of how sophisticated and how much heat the chips inside of them put off. Comfort Systems, they're one of the few companies that can really handle the scale and the scope of these projects.

10:22Jon Quast:So related to today's stories in particular, Taiwan Semi's CapEx is now guiding for the high end of its previous range. That's a positive item for companies like Comfort Systems that profits on this buildout. The company's sold out language they're using and the discussion of lead times of well over a year on new orders. It indicates a more durable demand curve than we previously expected with AI. In projects like Meta's massive build-out, it's exactly the kind of complex situation that companies like Comfort Systems can handle that most others can't.

10:53Matt Frankel:Yeah, Comfort Systems USA is a great company in my opinion. I wish I had found it several years ago. All right, Rachel, it's your turn. I'm going to ask you the same question. What is a company that you want to highlight here for our listeners? What does it do, and what could go right in the coming years? We were talking about TSMC. We look at that company, they make the silicone chips. We look at Meta, they buy those chips to power their AI models. But you can't just plug a loose chip into a wall outlet at a Meta data center. That is where the company I'm bringing to the table enters the picture.

11:21And that is Celestica, ticker CLS. So Celestica is an electronics manufacturing services powerhouse, but they're also very much one of the companies that I would count as an architect for AI infrastructure. So Celestica essentially takes the advanced chips from companies like NVIDIA and AMD, mounts them onto custom circuit boards and assembles them into these massive liquid-cooled AI server racks. They also build the high-speed networking switches that allow tens of thousands of these chips to actually talk to each other. They essentially sit in the middle of the supply chain. So they take the raw tech components, they turn them into the really functional supercomputers and systems that hyperscalers need.

11:59And if TSMC's massive 68 % revenue spike is to be our guide, a tidal wave of new silicon is about to hit the market. And the big tech companies like Meta are moving away from that off-the-shelf hardware. They want the custom-designed proprietary AI server racks to help them save on power and maximize efficiency. Celestica is one of the very few companies globally with engineering expertise to build these complex, bespoke systems at scale. And because they work hand-in-hand with the major chip designers, think of companies like Broadcom, Celestica gets locked into these multi-year data center buildouts early.

12:32So as we continue to see these expansions of CapEx from the big tech companies, the build out of next-gen AI clusters, Celestica just needs to keep assembling the digital bricks that it continues to use to fund the AI software race. And if this hardware cycle has a multi-year runway, which I personally believe that it does, I think Celestica is very well positioned to capture a nice chunk of that growth.

12:53Matt Frankel:Yeah, I think one of the reasons that this company does continue to sit beneath the radar is it's not exactly an easy business to understand, but certainly very important to the whole underlying infrastructure that we're talking about. So thank you both for highlighting these. I'm curious, Matt, do you own either of these?

13:09Jon Quast:I don't own either of them yet, but Comfort Systems is very high up on my watch list. How about you, Rachel? These are also both on my watch list right now.

13:17Matt Frankel:All right. Well, we hope they go on the listeners' watch list as well. We're going to head to the break. After the break, we're going to dip into the mailbag. You're listening to Motley Fool Hidden Gems Investing.

13:27Jon Quast:They say leadership isn't just about where you're going. It's about the conviction it takes to get there. For those who demand the world and possess the drive to claim it, there's a vehicle of equal distinction. Dynamic by design and engineered for pure impact, the Range Rover Sport rises to meet you the moment you take the lead. This is the most advanced Range Rover Sport yet, a master class in uncompromised performance and unbridled agility. Inside, the innovation is seamless. You'll find an elegant 13.1-inch touchscreen that puts total control of the vehicle's systems right at your fingertips, but it's the refinement that sets it apart.

14:03Jon Quast:Sculpted 22-way heated seating with built-in massage function ensures every journey is defined by peerless comfort. Whether it's through unique interior finishes or custom wheel options, the ways to personalize your Range Rover Sport are nearly unlimited. Command attention and experience ultimate luxury in motion. Exclusive offers are available now. Explore further at RangeRover.com. Your team just added its 67th AI tool and also your 67th security blind spot. The good news? The Vanta agent works like a GRC engineer in the background, finding every app your team uses, scoring the risk, and drafting fixes for you.

14:39Jon Quast:Vanta is the platform used by over 16 ,000 fast-moving companies like Ramp, Cursor, and Harvey, who are shaping the future with AI and staying ahead of AI risk. Get started at Vanta.com.

14:53Matt Frankel:Welcome back to Motley Fool Hidden Gems Investing. One quick note, we want to make you a part of the conversation. As I said earlier, we can't give personalized investing advice because we're not certified financial assistants or anything like that. But we do like to answer questions that you have for us about stocks or general investing questions. So you can send those in at podcastatfool.com. We'd love for you to keep them foolish. We'd love for you to keep them short enough to read on air. But we do enjoy taking those questions. So get those into us again at podcastatfool.com. And we have a great question teed up for you here.

15:28Matt Frankel:And I'll just read it. It says, hi, fools. I'm going to skip a little bit of the intro. But he says, I'm considering going back to grad school and could cover all the tuition by selling about a third of my portfolio. I've been debating this versus taking out student loans. And so essentially the question from Ben in Sacramento is, what should I do here? Should I take out a loan or should I sell a third of my portfolio? Now, I will reiterate, we can't speak to Ben directly because we don't know his full financial situation. But we are going to just kind of explore this question. when I have a need in my life and I would be able to cover it with an investment, what should I do?

16:07Matt Frankel:And I'll just throw out an anecdote here. This was something that I came up with recently in my life. Last year, I had a sudden need for a vehicle. I had an investment that was roughly a 10X that could cover the cost. And so I did sell that investment to fund a real need in my life. So I can relate to this question, but I'm going to throw it here to Matt, who probably has faced similar things in his life. Matt, what do you think about this question from Ben?

16:30Jon Quast:What a great question. And for me, it's going to be the next time I see this happening is when my kids go to college and need some money. So there's a lot to unpack here. So for starters, you're exactly right that investing is a means to an end. It's not just that something that has no end point, you're saving for something. Paying for planned life expenses is a completely valid reason to sell stocks, especially if it keeps you out of debt. It's also true that psychologically, it can be difficult to sell large volumes of stock, especially if you've spent years building that buy and hold mentality.

17:00Jon Quast:So here's kind of a good mental framing for this. Your investment portfolio is a representation of the purchasing power that you've stored away over your lifetime. By selling stocks to pay for grad school, assuming that you're getting a degree that will either boost your income or boost your job security, you're not exactly withdrawing or spending your money. You're transferring that capital from one productive asset, stocks, to another, which is your own earnings power. And depending on the earnings power of degree you're thinking of getting, it could certainly be the better return option of the two.

17:30Jon Quast:There are some negatives though. So avoiding debt, and I know I'm going to catch flack for saying this, avoiding debt isn't always as black and white as Dave Ramsey's going to have you believe. As one example, what kind of capital gains tax bill would you be looking at by selling one third of your stock portfolio? Because if you do that all at once, it could certainly move the needle here. What are the interest rates on the student loans available to you? Are you talking about private loans that if you have really good credit you can get for 4 %? Are you talking about loans that you can get for 10 %?

17:57Jon Quast:Those also move the needle based on your expected returns in the market by just staying invested. The question of which third of your portfolio is an important one, meaning which individual stocks are you going to decide to sell, which are you going to decide to keep, is an important one. And that's the tougher piece of the puzzle to solve. And without knowing your holdings, that's a whole other conversation for another episode.