The AI Infrastructure Opportunity

Motley Fool analysts discuss their views of the AI spending race and three stocks poised to profit regardless. They also make three predictions for the AI industry.

To catch full episodes of all The Motley Fool's free podcasts, check out our podcast center. When you're ready to invest, check out this top 10 list of stocks to buy.

A full transcript is below.

This podcast was recorded on Oct. 06, 2025.

Tim Beyers: The AI infrastructure opportunity. Just how big is it? You're listening to Motley Fool Money. Big up Fools, I'm your host, Tim Beyers, and with me are two of my most AI immersed teammates Yasser El-Shimy and Asit Sharma. Thank you both for being here. How are you feeling this morning? Are you feeling suitably enthused about AI?

Asit Sharma: Well, Tim, after not saying that we're your two favorite analyst, but maybe two of your most AI ford analysts, I guess I'm feeling OK.

Tim Beyers: This is a compliment. It's not a backhanded compliment, Asit.

Asit Sharma: I hear you, my friend.

Tim Beyers: Yasser, how about yourself?

Yasser El-Shimy: Yeah, well, I'm going to steal your expression here and say caffeinated and ready to go.

Tim Beyers: Perfect.

Yasser El-Shimy: Excited about talking AI with you this morning.

Tim Beyers: Very good. Let's talk about this because there's a lot of AI infrastructure spend. I'll give you a quick stat here because I find this to be absolutely bonkers. The New York Times is reporting that Amazon, Microsoft, Alphabet, Meta, and OpenAI combined are projecting 325 billion. That's with a B, 325 billion in spend by year end. We're in October, this was published in mid September. We're in October now, 325 billion by year end in pursuit of AI. That seems absolutely outrageous to me. Let's talk about other outrageous things that AI does before we get into the stocks. I just want to go around the horn because we're spending so much money. What is the last thing you asked AI to do? Could be for work? Could be for something utterly ridiculous? Yasser, you are laughing. I want to hear the ridiculous thing that you asked AI to do.

Yasser El-Shimy: Definitely not a ridiculous thing, at least not lately, although I would say that the last thing I ask AI to do is to help me out prepare for the podcast.

Tim Beyers: Perfect.

Yasser El-Shimy: This is just how tightly integrated into my workflow it has become. I find it actually, and this is incredible just the short amount of time since we started using AI here in any capacity. Then now we are at a point where I find it pretty hard to imagine doing most aspects of my job without the help of some generative AI model or the other.

Tim Beyers: Yeah. I mean, that's a paradigm shift. That's what we call a paradigm shift when habits change. Asit, how about you?

Asit Sharma: Same Tim, I used it to help me prepare for this podcast, but skipping that, the next most recent use, explain to me the phrase [FOREIGN]. This is a Turkish phrase that you didn't expect at all. Enough said, this shows how much this has taken over my learning curve among a number of disciplines.

Tim Beyers: You know what? Fair enough. I mean, I'll let that sit there, and then listeners can look that up for themselves. I will say I do this, I have started to use AI more as like a search. It's not replaced search for me, but now I ask utterly ridiculous questions all the time. I don't even do keyword searches anymore. I ask full questions. But let's talk about the opportunity, we have three views, one for each of us and three stocks which we're going to talk about. I'm going to again reiterate the amount of spend here. Microsoft just over the trailing 12 months, 64.6 billion in CapEx. Amazon, 107.7 billion, Alphabet, 66.9 billion, Meta, 52.2 billion. Of the big names, the only one that looks like the sober responsible one is Apple at 12.4 billion. I'm not sure they are or they aren't. But here's what I want to know. Yasser, what is your view of this spend because at some point, that spends got to pay off. What's your view of this and what's a stock that you think should be on everybody's radar in the midst of all this spend?

Yasser El-Shimy: Yeah, so I'm going to start off here by maybe quoting Satya Nadella, when he said every company is becoming an AI factory. We're witnessing these just absolutely incredible numbers of investments that you just outlined Tim, that all of the big large cap, hyperscalar technology companies here in the US and abroad are making in order to catch up with this AI revolution, let's put it that way. I would say when you said Apple is being responsible by spending only $12 billion. In fact, I might argue that could be a potential yellow flag for Apple. I think that a lot of investors have actually treated Apple accordingly because Apple has not shown us so far, at least, that they are able to come up with any major innovation when it comes to AI, they have had to rely on ChatGPT for the IOS systems here sold in the US, and they are relying on the Alibaba Qwen model for phones sold in China. There have been constant changes and staff and so on. Anyway, I could go on, but the point is, we're witnessing a wholesale paradigm shift in how we approach computing and what we use computing for. I was just talking earlier about how tightly integrated AI has become into my workflow, but that's not just me. That's a lot of people out there, and a lot of companies are spending a lot of time thinking about how to invest in AI.

Tim Beyers: Give me the company then that profits from this.

Yasser El-Shimy: All right, so the company, in my view, so on a risk adjusted basis, I would argue that Alibaba probably offers investors the best investment opportunity in AI over the next decade. Predictably or logically, a lot of investors here, when they think about AI investing, they think only about US companies, and that's understandably so. However I think given just a combination of very attractive valuation for Alibaba, as well as the massive investments it's making in its AI capabilities, and that includes a false tech. It's not only building the large language models, I mentioned Qwen earlier, which is integrated into iPhones sold in China, but it's also building data centers. It's building custom chips for AI. That it is even starting to sell to other Chinese companies that are in need of those chips. That is not necessarily replacing NVIDIA GPUs, but they're working on the inferring side to supplement and complement, but they have a major cost advantage compared to, let's say, US made chips or US designed chips. We are starting to see the early signs of green shoots of that massive jump in AI deployment, usage, consumption from Alibaba's Cloud Intelligence unit, which has had its revenue grow in triple digits over the past several quarters, where I think eventually Cloud revenue is going to become a meaningful mix of its overall revenue and margins are likely going to improve as a result of that change. All in all, I would say Alibaba offers a pretty good opportunity for investors here.

Tim Beyers: All right. Ticker B-A-B-A, BABA. Asit, what do you got for me here? You agree, disagree, refine? Give me something.

Asit Sharma: Yeah, I'm generally aligned with Yasser's view on AI capacity and the buildout that we're all witnessing. Several years ago, right before everything exploded with ChatGPT, and our lives changed without any chance of looking back, Jensen Huang, CEO of NVIDIA, said that the gains out of Moore's law were coming to an end, meaning thereby this linear ability to pack more and more information on a chip, had run out. At the time, lots of folks thought that he was simply saying that it's going to be harder for us to solve this mathematical type problem. But it turns out he was saying something else. You all are going to pay more to use this stuff. Lo and behold, a few years later, he's saying now that the reason all of this capacity needs to be built out. Why we're going to keep using this stuff more and more is because we're demanding more inference. We demand that reasoning that comes from this complex multi step task that GPU performs. Hence really to go back to what Yasser was just mentioning, we're going to as consumers, keep using these models. There is no turning back. Now, you're probably going to ask me, OK, if that's the case, what stock? Should we be looking at?

Tim Beyers: Read my mind. Tell me.

Asit Sharma: Here it is. It is NVIDIA's chief rival AMD. There's news out today that AMD, Advanced Micro Devices has signed this multi year deal with OpenAI. AMD is going to supply an enormous amount of gigawatt compute capacity, something like 10 gigawatts over the next several years. Tim, that deal is estimated to be in the 10 to billion dollars of revenue. In turn, OpenAI is going to take an ownership stake in AMD that's going to vest over time. Hello, circular economy. Let's slide by that for just a second. What I think this deal is is really a validation of a decision that CEO Lisa Su made more than a year ago, to buy a small company called ZT Systems. This is an AI infrastructure company, it allowed AMD to provide rack scale solutions. That is connecting tons and tons of GPUs together within a data center to provide, guess what? Better inference. This also is a validation of AMD's next generation of accelerators due out next year, which is going to be the first time this company will be able to really compete head on with NVIDIA. As we tape, this stock is up some 28% on news of this deal with OpenAI. But I think AMD is going to find many more purchasers for its systems over the next few years. Oracle already has come out as a very big purchaser of its accelerators, and I think it vindicates the narrative that AMD doesn't have to beat NVIDIA in the marketplace, it just needs to show that it is a worthy competitor, and it will shave off market share, which is going to be very good for a stock that is really yet to see the love that NVIDIA has experienced for the last couple of years.

Tim Beyers: It's interesting. You both make interesting cases here. Ticker, AMD is, as it sounds, Advanced Micro Devices, AMD. I'm going to go with Cloudflare. I'm going to zag while you guys are zigging. You're going into the silicon and the big Cloud providers. I think it's interesting we have not yet reached the era of AI efficiency. I think that history says that is coming. That is one million percent coming, in my opinion. At it's recent birthday week, Cloudflare said, essentially, AI cart at scale in order to serve as many users as possible. What we mean by this is every product, everything that they produce, any company can have it at full enterprise license for anything. Every product's got to be great on its face. They're not going to sell you a big bundle. They're not going to sell you a big tier, and then everything is jammed into that tier. They're not doing that anymore. Essentially, what they're doing is instead of looking at trying to win a customer at the biggest enterprise scale account they can they want to win problems. They want to solve as many problems as they can and then roll them up hopefully across customers. I think this is super interesting because at some point in a market where AI is just absolutely vacuumed up, money, attention, time, people resources. It's just hoovered it all up in ways that I think are probably somewhat damaging to some companies. There is a way to fix that is to make it a little bit more efficient. Instead of applying huge amounts of resource to solve the problem, we do better with less. I'm going to come back to that we've seen that cycle before, guys. But before we do that, we're going to move on to the next section here in a minute. We're going to forecast the future and make some reckless predictions. You're listening to Motley Fool Money.

All right, we're back. We're going to have some reckless predictions now. I'm going to tee this up. When I said Cloudflare and again, I failed to mention the ticker last time. That ticker is NET for Cloudflare. I'm going to start with you, Asit, by teeing you up this way. You mentioned whether you meant to mention it or not, the circular economy in AI, and boy, is it circular? It's not even oval shaped. It is perfectly circular. We're not just talking about the hyperscalar here. Let me give you some stories here. NVIDIA committed $100 billion to invest in OpenAI. OpenAI has committed $300 billion to Oracle. Intel and AMD, are talking about a foundry deal. Everybody is doing business with each other. Asit I have seen this before. I saw it during the dot com days when all of the cable companies and all the infrastructure companies, all the telecom companies were doing business with each other and selling fiber agreements and booking revenue that wasn't really revenue. Give me your reckless prediction here. You've got so much experience both of you with AI but where are we heading here? Does any of what I just told you give you pause?

Asit Sharma: Of course, it gives me pause, Tim, and I hope our listeners today, some of you all of you are jugglers or can juggle, I can juggle long term bullishness with this sector with my reckless prediction teed up so perfectly by Tim Beyers. I predict a mini crash in the AI infrastructure investment theme and that economy that's associated with it within the next three years. Now, at that point, I think we're going to see major hyperscalers suddenly reevaluating their spends. We've seen some previews of this already. The entire ecosystem is going to suffer as a result. This will be a time of significant sell offs across the semiconductor, the hyperscalar, and the energy sectors. We're going to see lots of pundits come out and claim they were right to point out this house of cards in the infrastructure buildout. I think we'll also see a flurry of news articles about the illusory value of AI or the infinitesimally small value of AI. But at that point in time, I think you're also going to see the major hyperscalars will be rejiggering their spend, so will sovereign governments who are getting into this game. They're going to complete their first realistic calibration of their investment time horizon versus their expected results. The industry is going to get back to work. It's going to continue to grow. That point in time, when this mini crash happens, I think it'll be an excellent time to buy your favorite high conviction names.

Tim Beyers: I like it. Yasser? What you got for me? You want to disagree with that?

Yasser El-Shimy: Well, I'll disagree with that, but maybe I'll offer a separate prediction later. I would say that we are going to be all surprised at how durable the AI infrastructure CapEx spend is going to be. Right now, we've seen all the forecasting agencies, and that includes Gartner and others continuously raising their forecasts of AI spend, not because they are just so extra bullish, but rather because they have been reacting to what CEOs and CFOs have been saying and have been indicating. The last I heard was Gartner saying that by 2026, we're going to have $2 trillion of investment next year. Alibaba's CEO, Eddie Wu, he's predicting now that by 2030, we're going to see $4 trillion of CapEx spend. I think that the buildup is real. I think there is a race here in order to create that capacity, compute capacity, be that the favor provider, if you will. Having said that, of course, I assume that the music is going to stop at some point. Now, when is that point? That's when it gets very testy. I would say that just as long as you avoid investing in highly speculative, extremely valued companies, you're probably going to be OK, in a long term basis. Just as the dot com bubble did burst in 1999, that didn't mean that the Internet was not useful and did not create trillion dollar companies as we look back in retrospect. We just have to be careful on how we deploy our capital and how we invest.

Tim Beyers: Yeah. I mean, and portfolio management is one way. We can do that, we can take small positions and build them up over time. I think I'm going to be in between the two of you here. I do think the dot com cycle efficiency interrupt and capital buildout is absolutely going to happen. That is going to happen. Having said that, I think the way that the efficiency manifests itself will unleash some real potential value here. I've said this before. If you've listened to me on Fool 24 or other places, I've said this before, I think that specialist AI models or embedded AI models are the thing that is most likely to gain huge traction over the next five years. In other words, inside of a distinct application, you have an AI model that does something that does something limited, specific and valuable and creates a workflow, instead of just something giant and monolithic, where all this spend is going, I see less of that and more of this. That is a pretty interesting way to spend money and get a return on AI quickly. For an example of what this is, we'll see whether or not he is right, but Aman Narang, who is the CEO and co-founder of Toast, has said, I want to make the greatest GPT in the world for restaurants. They're working on something that they very cheekily call Sous Chef. You're going to have an AI that they call Sous Chef and it'll be a specialist AI. That's what I'm talking about. Is it going to add a whole bunch of value to Toast? I have no idea, but I do know that is something we're likely to see a whole lot more of. Here's what we want to know Fools. Make your own reckless predictions. Comment on this podcast. Tell us which reckless prediction you think is most likely to come true. Your answer can be none of them, and you have a totally distinct reckless prediction. But let us know what you think. Up next, we're going to talk about some of those companies that are trying to seed new ground, and we're going to name some fakers or some breakers. The IPO edition, stay tuned. You're listening to Motley Fool Money.

We're back. We got three companies and I want each of you to weigh in on this. I've played this game with Rick Benares before. This is called faker or breaker. This is the IPO edition. For those who don't know, a rule breaker is a company that can grow at an accelerated pace for a much longer period of time than expected. A faker is not the opposite of that, it is the Fools gold of the rule breakers universe where there's outrageous growth for a period of time, but it lacks some of those other rule-breaker traits. Just as a selfish plug here, you can read more about the six traits of a rule-breaker in David Gardner's new book, Rule Breaker Investing, where he lays out the traits, the habits of a rule-breaker investor, and the characteristics of portfolio management when you are trying to build a portfolio of rule breakers. Great book, you should check it out. But we want to distinguish here. A faker is one that has a lot of growth but can't sustain it 'cause it just doesn't have the traits to back it up. Three companies, and I'm going to get you each to weigh in. Yasser I'm going to start with you. Are you ready?

Yasser El-Shimy: I am.

Tim Beyers: All right. Ticker, KLAR. That's Klarna. Recent IPO Klarna, I'm going to give you some facts on this one and then you tell me faker or breaker, 20% year over year revenue growth reaching 823 million in Q2. Fairly strong commercial penetration. They have a deal with Walmart. They displaced a firm there. It looks pretty good. However, they are relying on people spending because it's a buy now pay later platform. They may be relying on consumer spending unsustainably here. We have heard reports about this. People get their delivery, their meal delivery, and then they buy it on buy now pay later. Not great. That doesn't seem like a really great habit. Yasser for you, faker or breaker, Klarna?

Yasser El-Shimy: I'm going to go with breaker here. This is a company that has been one of the very earliest companies to pioneer the buy now pay later industry, if you call it that. It has been one of the biggest financial engineering introductions in the last decade or so. It has been able to sustain its revenue growth at a decent clip over the past few years, even though it has only recently IPOed, so we don't really have a long track record of share price appreciation, for example, or other traits of rule breakers. But I'm going to go on a limb and call it a breaker.

Tim Beyers: I'm going to keep moving because we're running short on time here, Asit, and move to the next one. That's going to be StubHub. You've probably heard of this one if you ever tried to buy tickets for live events in the secondary market here. Dominant 34, 30-40% of the market here, Asit, Ticker STUB. Seems to have a flywheel that they can rely on here, $8.7 billion in gross merchandise sales in 2024. That was up 27% here. But I have to believe this is one that is maybe, if not hated, might be a little bit suspicious and might have regulators looking over their shoulder. For you, Asit, faker or breaker, StubHub?

Asit Sharma: Faker, Tim.

Tim Beyers: Okay.

Asit Sharma: What you mentioned is part of it. Now, StubHub doesn't have quite the consumer hate that Ticketmaster seems to elicit.

Tim Beyers: Yeah, sure.

Asit Sharma: It certainly struggles with margins. It's a platform, which I think is very upendable. I think challengers can come in the future. It is one that enjoys a lot of brand presence, so we can't discount that, but it's ability to stay as this first mover top dog a type of company, I think is very much open to question. I think in the long run, this is a business that runs toward commodity and a business that extracts your irate response on some days, it brings hate instead of love. At some point, when you're trying to buy a ticket or get to a basketball game that you want to get to, I think that can't be a company that is rule breaker over the long term.

Tim Beyers: I got to say, I agree with that. I'm going to give you both the quick skinny on Fermi. This is Ticker, FRMI. This is brand spanking new, just born. This is a company that is building out data center scale electric power, doing it with nuclear, maybe some solar, maybe some natural gas. It's Texas based, it just went public, guys. Dual listing on the NASDAQ and the London Stock Exchange. They call this project Matador, is their signature product. I don't know, that sounds ridiculous on its face, but this is 5,236 acre site. This is in Marion, Texas. It's going to deliver up to 11 gigawatts of low carbon behind the meter power. It seems like the theme of this show is AI infrastructure. This is playing right into it, Yasser. Give me the quick answer here. It's very early, they're building out. Former Texas Governor Rick Perry is leading this. What do you think? Faker or breaker?

Yasser El-Shimy: Faker, faker, faker.

Tim Beyers: Faker, Faker, faker. Wow.

Yasser El-Shimy: Yeah. I would say it's a business plan, it's not a company.

Tim Beyers: It's a business.

Yasser El-Shimy: It's basically there are a lot of promises about what this is going to be. The fact that they IPOed before pre-revenue, pre anything happening, I would be extremely skeptical. If investors are interested in AI infrastructure plays, there are definitely a lot of companies already publicly traded that provide that as opposed to just a promise of one in the future.

Tim Beyers: Asit can you top that?

Asit Sharma: Yes, faker, faker, faker faker.

Tim Beyers: Nice.

Asit Sharma: It's got one rule breaker trait in spades, overvalued. As Yasser rightly points out [LAUGHTER]. This is contravening famous advice that master investor Peter Lynch gave, which is, don't invest in pre-revenue companies. Those are way long shot, so I'd be very careful here.

Tim Beyers: Yeah. Thanks to both Yasser El-Shimy and Asit Sharma for joining me today. As always, Fools, people on the program may have interest in the stocks they talk about, and the Motley Fool may have formal recommendations for or against, so don't buy or sell stocks based solely on what you hear. All personal finance content follows Motley Fool editorial standards and is not approved by advertisers, advertisements or sponsored content and provided for informational purposes only. To see our full advertising disclosure, please check out our show notes. Fools, thank you for tuning in to Motley Fool Money. For Yasser El-Shimy, Asit Sharma, our Engineer is Dan Boyd, our producer is Anand Chokkavelu, I'm Tim Beyers. We'll see you again tomorrow, Fools. Fool on, everyone.

Asit Sharma has positions in Advanced Micro Devices, Amazon, Cloudflare, Intel, Microsoft, Nvidia, and Oracle. Tim Beyers has positions in Alphabet, Amazon, Apple, and Toast. Yasser El-Shimy has positions in Alibaba Group, Amazon, Cloudflare, Microsoft, and Nvidia. The Motley Fool has positions in and recommends Advanced Micro Devices, Alphabet, Amazon, Apple, Cloudflare, Intel, Klarna Group, Meta Platforms, Microsoft, Nvidia, Oracle, Toast, and Walmart. The Motley Fool recommends Alibaba Group and recommends the following options: long January 2026 $395 calls on Microsoft, short January 2026 $405 calls on Microsoft, and short November 2025 $21 puts on Intel. The Motley Fool has a disclosure policy.