Episode 9: John Rotonti on AI Infrastructure Plays And Where We Go From Here

Hey guys and welcome to episode nine of shooting the bull pod with drowsy and bear bear we've been looking forward to this episode for several months now and I've just been counting down the days we've been texting back and forth almost every day that we would just be giddy to have this conversation and have our first return guest on today so without further ado John Ratanti is our first return guest backed by popular demand overwhelmingly dozens and dozens of people have hit us up saying when can you have Jro back Jro is a portfolio manager at Bastion fiduciary he is a doctor of stock investing and he is one of our best friends so Jro thank you for coming back on and welcome back to the show goes wild thank you for having me on I did not know that I was your first repeat investor on the pod but thank you and that's an honorary doctor title that you are you are giving me but I appreciate it I love that he forces you to disclose that exactly several months drowsy I think we just had him on a few months ago the first time but hey we were we were looking forward to have you on again before we ever had you on the first time so thank you I I was speaking at an investing conference a few months ago and the I was on a panel and the host of the panel introduced everyone on the panel as a CFA charter holder and had to wait until he was done with his beautiful intro to tell him and the audience there's like 150 people in the audience that I am not a CFA charter holder I don't know I don't know how he got that information but yeah that's funny if this is the second we're gonna do to have you back on the show just wait for the third one when I come up with more honorific titles for you and the first 15 minutes is just backtracking the intro so as as we joked about and I do want to start the pod this way so there's lots of topics we got to get to I want to hit AI we had a lot of questions from listeners about the AI infrastructure angle that I think you are perfectly well tailored to answer I wanted to hit though first with this hypothetical baseball question that I've been teasing out for weeks with you now so game seven world series is on the line you have to select one starting pitcher who has pitched since the year 2000 who is your pitcher and why since the year 2000 correct I don't want to like Dr. Johnson's or we're not going back to Sandy Kofax here or I'm sorry young despite the inexperience in the league and in a world series I'm gonna go with what's his name scenes from Pittsburgh Paul Skeens yeah I like it okay he played a second year player this year but he's been completely dominant second-year player he was what rookie of the year his first year he was Cy Young his second year earned run I think his career earned run averages low twos maybe below I mean it's like insane numbers he played at LSU I'm from Louisiana I just I you know I think his body and mind were built for base ball and I mean I great answer and Bear and I had a separate Kelsey bet going on on who you would select and he was not on the list we thought for sure you would go with the actual doctor and that's Greg Maddox so we were a little bit surprised and Bear and I selected individuals pretty close so I said you have to go Pedro Martinez I think for one game to win he's just so dominant when did Maddox retire you know I guess it was like oh five was that too late I mean I the only reason I didn't say matter because it's because I thought it would be only 2000 I wasn't sure when he retired but he's my favorite pitcher of all time great wait this is he retired in 2008 okay so he went he went from much much further than I thought so he will take our credit I grew up loving the Braves because we don't have a team here in Louisiana and so the closest teams are pretty much the Houston Astros which were very good at the time when I when I say growing up I mean you know middle school high school when I was watching a lot of ball you know the Astros had Craig Vigio they had a lot of other good players at the time but the Braves were the other close team that I sort of adopted at the time and yeah they had you know maybe the greatest starting rotation ever yeah I think we can do a whole separate show on early 2000s baseball but I want to so I want to transition a little bit more to investing so right now we're in the great pullback of 2025 and the AI trade we have the Vic spiking we have a lot of AI winners and darlings sort of pulling back and the big question obviously is are we in an AI bubble and just to timestamp this and to make us look extra silly we're taping on Nvidia day so Nvidia reports earnings later this afternoon this is Wednesday the 19th so what we say now may be quite different at 5 p.m. today we don't have to wait long to be wrong on this to be wrong wow oh my gosh there there's the episode title but Jero where do you think we are in this kind of AI narrative it seems to be pulling back where are you on all of this you know you mentioned narrative and price you know drives narrative a lot and if you would have asked me this two and a half or three weeks ago Jensen was speaking at you know the conference in DC and he said that Nvidia had you know a line of sight to 500 billion additional revenue over the next five months and Nvidia's market cap hit five trillion and you know everyone was like this is super early innings in the AI infrastructure build out and then you know fast forward two weeks and now AI stocks seem to be selling off so you know here's here's the lens that I'm using Jensen Wong he he says there's gonna be his estimate is for three or four trillion will be spent on the AI infrastructure build out through 2030 and that's the lens I'm using right now because I think Jensen is the leader of arguably the most important AI company in the world I think that Jensen is the architect and almost the you can think of it like the general contractor of this global AI infrastructure build out so Nvidia's at the forefront of not only inventing the AI intelligence infrastructure they could fundamentally change the way the world works and lives it's so it's not only designing the most advanced leading edge AI chips but they're setting the pace of the upgrade cycle so Jensen is literally setting the pace of the AI race and then as the architect and general contractor he's building an ecosystem of key named partners much like a general contractor selects the subcontractors to work with and yes you know as part of this ecosystem build out Nvidia is investing in AI startups and they are coming up with these circular financing deals to help finance this generational infrastructure build I think that Jensen is hyper aware of the position of power that he's in and I think he's architecting this build out in a way that he tries to bring along as many key partners as he can building this ecosystem obviously benefits Nvidia it creates a lot of lock in customer captivity but it also creates wealth and value across the ecosystem of partners and across that AI supply chain and so that's the lens I'm using three or four trillion spent on compute and AI data centers over the next five years and then on top of that maybe another trillion on electrical grid upgrades and build out to not only support the AI build out but to support a larger reindustrialization of America so there's a few different angles I want to I want to hit that sort of opening foray into the conversation with the first is this narrative that's been percolating especially on Twitter and among CNBC host recently is the narrative that Nvidia chips are under fire that Google and Amazon have created chips that are leading to trained LLMs that can compete with the best and I think Google announcing that Gemini 3 was trained exclusively on its own chips I think really cemented that sort of narrative so I wanted to get your take on the chip threat to Nvidia and then we'll go from there have a few other thoughts on it too I don't think the chip threat is that concerning to me yet so Google's TPUs they've been developing I think since 2014 and their purpose built specifically for AI specifically you know for that purpose and they have a much larger lead in custom-silicon ASICs than either Amazon or Microsoft so Amazon's got Tranium and Inferensia or Inferenium whatever they call it Microsoft's got Maya I think it's called there you know significantly further behind then I in my opinion then either Google's TPUs or or Amazon's custom-silicon but the fact that I think the fact that Gemini 3 and 3 Pro are getting such rave reviews and the fact that Gemini is taking generative AI market share consistently pretty much quarter-in-quarter out from Chatchapiti. Chatchapiti is still the far and away leader I think Gemini's got like 14% now but it's a consistent eating away at that market share I think that almost forces the hand of the other model builders to get their hands on as many GPUs as they can sorry as many AI chips as they can and you know there's some estimates out there that the supply demand for Nvidia GPUs specifically is 12 to 1 I've heard Dan I've say 15 to 1 but basically for every 12 requests that Nvidia gets they can only only have capacity to fill one of those requests and so if you know I think if Gemini is doing so well and if Gemini is pulling ahead as much as it seems to be I think every other model company has to just get their hands on whatever compute they can and the leader in compute is Nvidia GPUs and let's not pretend like they're not the most advanced chip on the planet the TPUs from Google are designed specifically you know to run certain types of AI but Nvidia chips I think are still the most advanced and the most powerful and I think I actually think that Gemini success is going to sort of counter-intuitively drive demand for Nvidia GPUs if I definitely don't I definitely do not think that this AI infrastructure buildout is dead on the vine by any stretch of the imagination so we talked about that three or four trillion estimate from from Jensen at Nvidia which is a company that we own in the portfolio that I manage at Bastion let's sort of like triangulate that three or four children and like sanity check it to see if it makes sense against you know other data points and project announcements out there so data point number one US hyperscaler guidance suggests combined capex of over 500 billion in 2026 just 2026 that's over a half a trillion in capex in 2026 just from the US hyperscalers data point number two Ben Bajaran that creative strategies he's he's on X he's tracking he does very good work in my opinion he's tracking 170 gigawatts of planned AI data centers through 2030 so these are these are announcements that has been made around the world and he's tracking these 170 gigawatts multiplied by 40 billion per gigawatt that's 6.8 trillion that's above Jensen's three to four trillion proclamation Morgan Stanley estimates 106 gigawatts of data centers will be built in the US by 2028 and there I think their numbers at like five to six trillion for the buildout so so first data point was US hyperscaler guidance for just 2026 second data point was 170 gigawatts being tracked by creative strategies third data point Eaton which is another stock I own in the Bastion portfolio is tracking 2.6 trillion in announced mega projects in the US and that's that's almost 30% higher from the year before so 2.6 trillion that's very close to that three trillion dollar number from Jensen data point number four the White House the Trump White House actually has a list I could send you the link or you put in show notes if you want a list of all the quote-unquote announced investments in the US from both domestic and foreign sources that number is 9 trillion so if you think about these and now now I'm not saying all of these come to fruition right but if you think about these announcements that's Apple 600 billion commitment that stargates 500 billion that Taiwan semis 100 billion that Saudi or so on the list as of this morning when I checked at Saudi Arabia was still at 600 billion but as of yesterday Trump and the Crown Prince I believe of Saudi Arabia announced are gonna up that to a trillion and so that list on the White House is 9 trillion which is above Jensen 3 or 4 trillion I mean I could literally go on and on one one word data point number 5 so Jeff Bezos when he was speaking at this like conference in Italy maybe a month ago he said AI will affect every company in the world and then he literally clarified himself and he said I'd literally mean every company well if we look at like how much is spent on tech around the world because he's talking about every company will Gartner forecast that global IT spending in 2026 will hit 6 trillion dollars for the for the first time and so you know we can sort of sanity check Jensen's three or four trillion number and that's below pretty much all of these other sort of data points that I just shared with you so my naive grenade question was just going to be who the customer is and I think you might have just answered it it's it's it's every other company in the world that's going to be spending part of their IT budget or maybe most of it on AI maybe you can confirm that that's sort of what you're thinking but my my grenade question was going to be you know if you give drowsy a dollar for his time here and he gives me a dollar for my time here and I give you a dollar for your time here revenue is being generated but nobody's profiting and so if there's not really an end customer for AI the buildouts great and it's a lot of revenue and it's a lot of like real commerce happening but is it a creative to the profits of the world that was going to be my question you may have just answered you may have more thoughts on that but I'm just curious what you would say to that I think it has the potential to be a creative to the profits of the world in two ways one and I think the earlier possibility the the way it could contribute to profits more soon is through productivity improvements cost cuts basically and then the other way is through generating revenue whether that's from new products whatever it may be so just to clarify if you're sort of thinking about this from a hyperscalers perspective when you're saying this like Amazon or Azure sorry Amazon or Microsoft as organizations could benefit from saving costs or from generating revenue from their AI investments well I definitely think the hyperscalers are going to deploy AI internally across their firms I mean I think you know even last year or much earlier this year maybe Q4 of last year Alphabet said that 25% of all the code of the company was generated by AI so they're definitely deploying it internally but no I mean every company on earth I think there's at least the possibility for that now where we stand now is that most companies haven't adopted AI yet at least not across the enterprise and enterprise-wide so we could look at survey data you guys to determine where we are in the build out it's one way we could determine also how far along you know how inflated this bubble is because I do think and a bubble is inflating I just think that maybe we're not that far along yet of course I could be wrong about that but so you so we so we look at at survey data so Goldman Sachs put out a note recently saying that corporate adoption by AI by large firms in the US I think they mean S&P 500 okay they say large firms in the US is 13% according to the most recent Census Bureau survey so 13% that's not a lot now Goldman also surveyed their investment banking team to see how much their Goldman Sachs corporate clients are using AI and through that survey it was 37% are using AI and so you know I don't I don't know I don't know where the number lies Morgan Stanley did a survey recently and they said that 15 percent of S&P 500 companies are reporting quote-unquote quantifiable benefits from AI McKinsey did a state of the AI report in November this month 2025 number 2025 and they said that two-thirds of companies have yet to adopt AI across the enterprise and so you know I don't know where we are in the in the build out and in the adoption phase but from from the surveys that I'm reading it's not broadly adopted and those companies that are using AI it's usually like one or two departments experimenting with it rather than a tactical deployment of AI across the company and once that happens if that happens I should say and if these companies start using agents I do think there's the potential that every company could cut costs and boost profits I'm sorry boost top-line revenue in some way at least I think that's the vision so I want to I want to stick with the build out theme and you mentioned in our podcast I think it was episode four our first time with you the idea of the power generation needed to supply for all these data center buildouts and you mentioned something on the order of multiple Manhattan's would be needed to power forward I cannot tell you how many people have texted me random things about Manhattan that I had no idea about either substantively I'm like I'm not a New Yorker so I have no idea out of context completely about powering Manhattan as if I'm an electrician or something I have no idea what you're talking about and they're like J-Row mentioned I'm like oh now it all makes sense people were absolutely enamored by that fact so with that kind of context I want to sort of flip the build-out conversation on its head from what we did in the previous episode to let's kind of put the chips and what's powering it GPUs and that part of the build-out aside and start with that infrastructure angle that I know you've keyed in a lot and I know plenty of investors that I talked to plenty of professionals that I've talked to on this too aren't thinking about it through a very particular lens from that build-out angle they may know Vertif or some of these other names that we've all heard and they're like yep this is powering the future or you know you're gonna need more coolant so here's one company on coolant and nothing wrong with that but not this kind of framing not this holistic understanding of the environment and kind of just token companies to hit at it so I want to sort of start with the basic question of walk us through kind of the infrastructure build-out and what is needed and how an investor can sort of frame the necessity of understanding the topic and of playing into multiple angles to it so that Manhattan thing I forgot that I mentioned that Manhattan runs on average on I think six gigawatts a year and at the peak season they're 10 gigawatts a year that's Manhattan for a year 10 gigawatts at peak and so we're talking about building a hundred a hundred gigawatts right so we're talking about adding more than 10 Manhattan's or 10 Manhattan's at their peak by 2030 that's a lot that's a lot of power and nothing like that has ever been done like in that short of a span this is like I don't know literally like the Manhattan project-esque of science to put together something that actually is powering a nation in such a short time frame right yeah yeah you know the Wall Street Wall Street Journal published an article recently said that this will be as least as consequential as the Cold War and and yeah you know I read I read something from the Department of Energy actually so one one gigawatt which is the size of a lot of these AI factory announcements is one gigawatt that's over a million horses a million horses yeah so it's a lot yeah so you know I think that Jensen Jensen says that it costs 50 billion to build a gigawatt AI factory I've heard heard estimates as low as 35 billion so Bernstein research put out a good one saying 35 billion for one gigawatt like I said Jensen says 50 billion most of the estimates I read are 40 billion to build one gigawatt AI data center and to put that you know into perspective for a reference point open AI has announced commitments to spend 1.4 trillion building 30 gigawatts of AI data center so that's 46 billion per gigawatt now once again that's if those commitments come to fruition but based on open AI's math today or at least when they mentioned this a few weeks ago that's 46 billion per gigawatt yeah and and you know I just I don't you could you could literally start you know with the land right and then and then this land is in the middle of nowhere in Texas for example right and it's muddy and wet literally and you can't build on mud and wet and so the first thing you do is you bring in like a company like an equipment rental company like a united rentals which we own in the portfolio and the first thing they do is they put up a fence around that so that it's private right so that people know don't come in here do not enter construction area you could die so they you know you put up the fence united rental provides the fence well you can't work on this like uneven wet muddy quick sandy ground so they have outdoor matting specialty outdoor mats that you literally lay down so people can walk and trucks can drive in this area that is just all messed up they need what did they need office they need they need mobile office space you've seen these mobile offices they look like they're in like you know mobile homes are almost like containers like metal containers they provide them they need mobile bathrooms they provide those then they're going to need lift equipment bulldozers you know aerial equipment cranes they provide all of that initial stuff to even start building right so the first thing you need to do is you need to buy the land they need to get that ready but if you know if you break it down 40 billion I think I've seen estimates that any anywhere from 40% to 60% of that is GPUs okay so that's where you get these huge numbers from Nvidia of out of 40 billion 40% to 60 percent is going to Nvidia the next largest line item of the build is is networking basically networking and optics and so you know that's companies like Arista like Celestica like Amphenol and TE connectivity there's there's a third company called Molex it's owned by Coke Industries but if they're private we own Amphenol and we own Arista and then you know the the next largest line item below that is electrical and mechanical and under mechanical it's it's HVAC and cooling and under cooling you have air cooling and you have liquid cooling and so these massive data centers they're the size of dozens of football fields put together and it's often not only one data center but like a complex it's almost like a city of data centers multiple multiple multiple football fields well each one of those has these massive custom they're called applied but highly customized each VA system air cooling systems that send in cool air and extract of warm air out they're massive they're custom-made they're made by by companies like train which we own the portfolio and Johnson controls which we don't own but another really good industrial company carrier makes them as well carrier makes them as well Lennox is an HVAC company that doesn't participate in the applied space they just mainly do residential and small commercial but you know as as rack densities increased Nvidia's racks are going from 8 GPUs to 72 GPUs and then after that to 140 GPUs as rack densities increase these things run hotter and hotter and hotter and so the only way to provide the cooling is with liquid direct-to-chip cooling and there's a variety of ways to do that and you know a handful of companies that it can do it at scale or less or less you mentioned Vertive is one of them which we own and there's others and vent which we don't own Eaton has gotten into this through acquisitions which we do own and you know you just keep going down the list of things if yeah you know if you there's there's companies that actually just build the racks and sell the racks companies that we talked about the networking but all the copper all the fiber optics all the little interconnects that's all in the you know what's called white space in the compute space of the data center but then you have the gray space which is what's in the mechanical room so we talked we talked about some of the HVAC stuff but then you also have electrical and power equipment everything from uninterruptible power supplies to small low-voltage transformers to bus bars to switch gear and you have a variety of not a large variety but you have a handful of companies that can do this at scale and that's once again Eaton Snyder electric and Vertive are the main ones there it's there's not as many companies as you think because two reasons one they can't do it at scale they don't have the global service force that can get out there to service these things in an hour literally in an hour Vertive for example has over four thousand service technicians around the world Eaton has a policy where they try to get a service tech in a truck out to a data center within an hour of the call so they just they they don't have the service capabilities they don't have the manufacturing capabilities but also because as we talked about earlier NVIDIA has a list of named ecosystem partners and these ecosystem partners and the hyperscalers have their own ecosystem and open AI has its own ecosystem but what happens is you are co-designing from the jump like literally from day fucking one you were at the table designing with NVIDIA at the table co-inventing with NVIDIA and so by doing that you create a lot of value you lock yourself into that ecosystem but you also get a line of sight into what's coming down the pipe where the where innovation is going in the next one two three years because Jensen will say here's what I need to do next year go develop it go design that that moat is just I guess one way I've thought about the moat in the past is just their technology is so advanced that you know to cool these you know very hot temperatures takes a lot of technology to design and implement but it's interesting to hear the mode is actually being in the room where it happens to quote Hamilton hashtag Hamilton but also the idea that you have the global service ability to continue service to these things I mean that moat is super strong I mean it's more I guess resource intensive and personnel related than just technology but that's an almost even stronger mode and harder to overcome in some respects it is the mode and but but being at the table reinforces their technological lead because like I said they are co-designing with NVIDIA with the hyperscalers and so they are they are they are getting very early insight into where the innovation roadmap has going and then they just send that back to their R&D team to the engineering team and say get it done and so it's a it's a self-reinforcing mode if you're not at that table it's really hard if you don't have that service organization it's really hard by the way liquid cooling requires a lot more servicing than traditional air cooling because it's plumbing plumbing is involved and there's as much more opportunity for leakage and leakage can destroy the whole thing and the whole the whole the whole rack and so yeah it's you know I think there's a mode there I mean you mentioned you know verdict of just just thinking about where the where the company has has has been so it was founded in 1965 this is a six-year-old 60-year-old business not a lot of people think about it that way because it was came to the market through you know David Cody Cody in 2020 through a spec but it's a six-year-old business it was founded as the first manufacturer of precision computer room air conditioning wow does that sound familiar computer room it's just yeah massive massive massive computing room they were literally founded so this is what they've been doing for their life for 60 years and then you know over time they have largely through acquisitions and by the way forvert of these acquisitions have been very timely almost prescient and many of them made years before chat GBT which was three years ago you know they expanded into making these custom air handling systems like we talked about into liquid cooling into electrical power equipment and now into manufacturing racks for the data center and so you know but they've been doing computer room cooling since since day one 60 years ago since we're all stock pickers and it sounds like verdict is the one that you've picked in liquid cooling correct me if that's wrong I mean I'm curious about other liquid cooling companies I know a lot of people who are very interested in and invested in at some point super micro I'm not sure if you have any thoughts on them or if they're where they fit in the competitive landscape or if there are others it sounds like one of the things about verdict is they have a seat at the table with Nvidia for some reason I remember that being said about super micro I don't know if it's true I don't I haven't spent a lot of time on super micro but I never I never really thought about them as as a liquid cooling company I'm not saying they're not when I think of liquid cooling I think of Vertive I think of invent I think of Maudine I think of the company that Ethan just acquired it starts with a void void okay those are the leaders in liquid cooling is from the research that I've done I thought super micro was doing a lot of this like server racks and stuff like that but I could I could be wrong it's not a company I spent a whole lot of time on maybe that was it maybe they were like a Nvidia racker and reseller not reseller but I think I think I think that they part of that definitely make server racks similar to like a Dell for example and they're in there in that eco they're in that ecosystem look there I was gonna make the joke that they are what they have been doing is potential SEC fraud but that's that's different well it's interesting because it is I have heard that it's similar to Dell and I was just always curious well you know it or either of these because you know super micro had a crazy stock run did mostly hype I think but but also you know some some revenue growth and I was just curious if that was just the numbers very well but if if Dell experienced any of that too but maybe that's not the spot in the value chain where a lot of red meat is I'm not saying that either it's you know I'm just I only have so much bandwidth and there's gonna be companies that I miss that I should study that I just don't because for one reason or another I gravitate towards the ones that I gravitate towards and I only have so much bandwidth but you know Dell Dell server business is growing I would assume just as fast as super micros it's just that that doesn't show up to the same extent in Dell's overall top line growth because there are a variety of other businesses as well right and so parts of Dell's business are growing every bit as fast as super micros I would assume it's just Dell's overall corporate level growth is not at the same rate because it has some slower growing businesses as well it probably has some legacy shrinking businesses as well but you know I have not looked at Dell that closely either although I do own some Dell personally personally this is this is bringing back all my questions around you know super micro was one of the first I don't know if you'd call them I guess hardware companies that came along that had some meteoric rises you know and not just in the stock price but in revenue you know doubling or something like that it's kind like Nvidia's did a few years ago maybe Nvidia was really the the first one to see the the benefit start to to hit hard when I think about you know we learned on our last conversation with you that you are not allergic to fast growth by any means nor to stock price appreciation it's just what underlies it of course and you know I think about that we talked about a stereo labs last time and that's another one where the growth has been phenomenal in revenue and the stock price has followed with with super micro you saw at some point and again I don't expect you to have an encyclopedic knowledge what happened to them but in my understanding the revenue basically doubled you know in a couple of quarters and then leveled off and I always worry about not every company is going to have the continued growth that sort of tracks Nvidia's right so not every company's going to keep growing like that but they will occasionally get priced like that I think that's what happened to super micro before they their stock crashed back down and a stereo labs has become a large position for me since we last spoke partially because I have been tracking several quarters now and the numbers have not started to default or like super micros that I know you do things a little bit differently with your tracking and probably understand them a lot better than I do but I remember you saying you're interested in them I don't know exactly where I'm going with this other than I know we wanted to talk about them again at some point I just am curious because again like super micro and even more so they were starting from this small small base of revenue that's now growing even faster than Nvidia at this point but it's still small and I'm just curious maybe the question is what would you see them becoming or they do they have a shot to become a huge company or they can they do they have is what's their what's their market I guess what what what types of things are they capable of I think the reason that I don't own a stereo labs yet even though the stock has fallen out of bed significantly recently is because I don't I don't have a I don't have conviction in an answer to that bear you know I do think that the retimer product that they make to make sure that that data does not get corrupted as it travels at very high speeds and very long speeds inside these data centers is an important product and I do think that they are getting more intertwined into the hyperscalar ecosystem I do think that which could indicate to me that they're going to be much bigger company in the future just I don't know enough the retimer market is so new and I haven't studied it enough that I just don't know how how replaceable it is if at all I just don't know if there are any substitute products out there and so until I get a better idea of that I don't think I'm gonna you know buy it but that could change if I if I if I get conviction around that I think I think an important so as there is like this newer company in a lot of ways and retimers are sort of this like newer product category in a lot of ways I have gone about building my portfolio a little differently and so you know a question I get pretty often from products I'm sorry from prospects is talk to us about how you manage your AI portfolio I don't think about it that way like I don't think I'm managing an AI portfolio you know I would say that I'm a generalist I'm definitely not an AI specialist and I and my aim is to invest in the highest quality best managed companies that I can find in the industrial technology and infrastructure space because that's my mandate right I manage an industrial infrastructure strategy now I do spend a lot of time studying AI because it's generational in nature and possibly the most important technological invention of our lifetimes that has the potential to change things in a fundamental and tectonic way and also you know I just I just I just I just do think that you know AI is a is an important part of this larger infrastructure buildout and so it's another reason that I have spent a lot of time studying AI but it's not a name it's it's not an AI portfolio however I do have a lot of AI exposure I don't know I don't know if a lot is the word but I do think I have ample AI exposure but I do it by trying to identify what I think are the best companies in the world and by best I mean companies that find ways to remain relevant and that have a long history of adapting to technological changes in a way that drives new profit cycle growth and value for owners over a long period of time so but so my point is I'm not just trying to identify what companies do I think have the best AI stocks or what AI stocks do I think will perform best over the next year or two or three rather I'm just trying to find the best infrastructure and industrial companies that I can find and by best once again these are companies that are adaptable and find ways to remain relevant no matter what the technological wave is right and so right now the technological wave happens to be AI and so for example let me give you some examples so four of the companies that I that I think I have most AI exposure to are Amphenol which we own Eaton we own TE connectivity and train technologies okay what's interesting about these four companies that are all getting very good AI demand and very good growth from AI is that train was founded the train brand was founded in 1885 so it's 140 years old Eaton was founded in 1911 it's 114 years old Amphenol founded in 1932 it's 93 years old and TE connectivity was founded in 1941 so it's 84 years old these four companies are an average of 108 years old so can we call 100 year old businesses AI companies I mean I don't think so but I'll let the market the market thinks so because my portfolio is suffering right now that's for sure but like I don't I don't think of it that way these are 100 year old companies these companies were around before the word AI had ever been uttered you know by decades so they're 100 years old so you know while these stocks definitely get lumped into the AI trade and and and you know their AI data center businesses are surely growing like gangbust gang gangbusters I don't consider them AI I mean they sell these are these are multi-industrials they sell into you know anywhere from like six to ten to 12 different verticals one of those verticals at each of these companies happens to be AI data centers right one of these verticals if that if that vertical went away bear and drowsy these companies would not go away I'm not not yet sure that's the case for a stair labs right like I don't know the answer to that if AI snap of a finger and AI disappears I'm not sure a stair labs remains relevant maybe it does I'm just not sure but if AI goes away train an amphenol and eaten and te connectivity there's no question in my mind they remain relevant right and so I think that's where I'm coming from I don't I don't think of them as a as AI plays I don't think of them as AI companies you know I just think that they remain relevant their products remain in high demand they're benefiting from much broader trends like electronification digitization decarbonization grid build out electrification and so that's you know that's what that's what I'm trying to look for I'm trying to look for companies that have proven their ability to adapt to find new S curves find new growth profitable growth cycles and I you know I think that's what drives high terminal values and longevity over time and I love I mean this is one of the reasons we love talking to you John is just these names are names that the investors that we mostly swim with high growth hyper growth whatever growth boys whatever you want to call us are not talking about they're not growing fast enough or maybe it's just one vertical of theirs is but the other verticals are not and it kind of drags down that overall number so it's fascinating to finally be able to look up Eden as you're talking about and realize it's a hundred and thirty five billion dollar company like this is not a small company and as you mentioned I've been around for a hundred years I I mean this is just convinced me again that I need to go back and kind of widen my aperture a bit and kind of incorporate some other of these companies that we hear about in that token way that I think a lot of investors approach it but also to realize that one vertical exploding is good durability and adaptability over a lifetime is what you're going for it's what I'm going I mean it's what I'm going for but it's not it's not necessarily the right way right and it's it's it's almost surely not the way that's going to maximize returns in the short term you know but it's just the way that I've chosen yeah yeah and I guess I'm I guess I mean bear bear to some extent but I think I'm more similar as an investor to you just in the I'm looking for making as few moves as possible over as long of a period of not making mistakes as possible and just letting good companies continue to be good so like when I hear about these you know verticals exploding that's great I'm more intrigued that they've had a hundred years of success doing things that had nothing to do with open AI or circular financing until the last few years and these companies are built for it and have CEOs that realize this long before and have kind of aggregated so I'm really interested in that and sure it may not be a hundred percent grower every year I have a few of those in my portfolio I have a few less now in the last couple weeks than I did but I'm really kind of looking for that set in forget it so I'm pretty intrigued by that and not not to throw a bear under the bus or any investor under the bus but others are looking for that kind of quick win of numbers and then to sort of either get out or look for that next one and that's never been my game just because I'm not any good I guess I'm curious you know I have several thoughts like one of them is if these these companies just I'll pick one verdict since we said it before if they're if they have benefited you know I don't know what how much the stock prices up year-to-date but if they've benefited significantly from AI demand and like you said AI demand somehow magically went away and the company was was fine I still think they would take a huge hit probably to revenues and profits but certainly to their stock and so I guess I'm curious why that's a you know like obviously it's better to go down 50% than to go to zero but so is there any other reasoning behind your so the four verdict was not in those four companies I mentioned right so okay well but you pick one but verdicts of no what they're different inverters important to talk about right so I it in a portfolio of 25 stocks I think I have what I would call three AI companies and I would put inverted as one of those so in video is one of those alphabets one of those and verdict is one of those I think that each of those companies are you know nuts to bolts AI companies fully stacked AI companies why verdict because 80% of their revenues come from AI data centers and because demand for their liquid cooling products at least is currently very closely tied to NVIDIA GPU demand and so I do consider a verdict to be an AI company one of three in the portfolio I called out the other four amphenol T connectivity train and eaten so trains you know trains it has a residential business and has a commercial business it's commercial bit just its commercial business sells into 14 different verticals 14 different verticals right and you know eaten sells into I think it's six or eight different verticals same with amphenol same with T connectivity and and you know at these companies I would say on average these four you know AI is 20% of their business maybe 30% of their business somewhere in there not 80% like verdict and not close to a hundred percent like NVIDIA and so I just think that I don't know what the stocks will do bear it in a sell-off and I I don't care and I actually hope that they fall a lot so that I can build up the positions in them my I don't think they will fall as much as an NVIDIA or averted if all of a sudden the AI bubble pops I don't expect they will fall as much because they do so many other things other than just AI I could be wrong about that even though their business fundamentals surely tell me that they do so many other things other than AI they do get caught up in the AI trade and so it's very possible that if the AI bubble pops they will get crushed as well I don't know what will happen but I don't expect they will fall as much as what I'm calling a pure play AI company which is verdict so with our so we have a bit of time left so I wanted to hit on a completely out of left field question so we've talked a bunch about AI portfolio although not really an AI portfolio as you perceive it lots of new companies for Bear and I to dig into and hope our listeners have a lot of joy kind of going through these companies to understand them we'll have additional discussions on them I'm sure Bear and I will dig in on them as well John you go to different colleges and brief and present on stock investing and teach I mean you've always been a teacher whether at Motley Fool or after of investing but I'm curious if you've seen a change in young investors over the last few years or what what are you hearing from that next generation I think what we see is a mixture of people understanding long-term investing and the benefits mixed with time to go on Kalshi or time to bet on whatever Malaysian doubles bad men next week and I wanted to hear like what are you hearing from the students and what change what change can you detect even through 2020 to 2022 bus to 2025 AI hype what are you hearing from the youths yeah great question I spoke to Fordham University and Tulane University both in the last month Fordham was actually in person to in New York so for the first time I'm getting a lot of questions around crypto and a lot of questions around whether AI could disrupt the investing industry could disrupt me out of a job or or disrupt analysts out of a job because you know a lot of these students are in investing programs at the undergrad level and they're trying to get analyst positions and so they are they there's some anxiety around that for sure and the so they actually asked me questions about whether I invest in crypto how I would incorporate crypto into a portfolio they asked me questions about how I use AI in my day-to-day job and whether I think it's a long-term threat to the analyst and portfolio manager role and the last thing I'll say is they are using AI absolutely in their research when they you know do write-ups for class study a business and do a write-up and maybe even evaluation on the stock and on the equity they are they are incorporating AI yeah and last year I spoke to the same schools in 2024 I didn't get those questions and so I mean so the AI anxiety is out there for that sort of entry-level white collar job seeker and certainly you know an analyst kind of in the financial industry is definitely fighting a bit against you know deep research by Google or some aggregate of AI LLMs coming in with some amount of surface-level research that can kind of provide an overview on something so I totally get that I'm curious your answer on the crypto front and I know crypto maybe isn't your forte it doesn't sound like it's close to your portfolio in many realms but I think increasingly crypto is a mainstream form of investing is being taught and certainly folks that I talked to in the industry are putting it as a 5% part of your portfolio you need some exposure because all these other companies are doing it and we're kind of me right now we're living through crypto apocalypse as it falls 30% and people freak out but I'm curious your thoughts and how do you talk about that to kind of the next generation of investors who are you know doing the modeling doing the deep understanding of a company but then they're also seeing their friends or seeing other people in the industry making ungodly amounts of money on things that don't have underlying value I didn't get I didn't ask them specifically whether they whether these students are investing in crypto but I get the sense that they are even if it's just small amounts I get the sense that they are because you know they were focused on those questions and it was definitely top of mind for them you know you asked me about me personally there's 18,000 crypto coins out there I own some Bitcoin personally I could not you know there's 18,000 coins out there I could name another five and that's it and one of those would be Fart and one of those would be Malangia I was gonna say bears long fart coin he's got 20% position portfolio update bears all crypto yeah like I literally could not name five coins to you but I own I own some Bitcoin and I have for not long enough you know I was very late to the story but for maybe two or three years two and a half small whale if you will I'm definitely a small whale for sure you know like just back to like the portfolio and so there's another company Quanta which is an electrical grid solutions provider we actually got a lot of questions on Quanta from our previous conversation to which Bear and I knew nothing so this is perfect so they they do everything from like literally designing if a hyperscaler wants to build a data center and there's currently no grid power to that data center they're at the table before the data center even breaks brown breaks ground out how they can help get grid power to that data center they're at the table okay and Quanta has built over 50% of all the long-distance transmission lines in the US 15% of all combined transmission and distribution so T&D but they're they're special to historically is building these long-distance high voltage transmission lines over 50% market share if I told you they were an AI stock two years ago you know you think I was nuts this is an electrical grid engineering and construction firm but how it's historically been thought of but on deep-seek Monday January 27th a data that I remember quite well because the portfolio got got punched you know it was down more than 20% that one point that day I think it finished the day down 20% but at one point it was down I think 25% but no no one you know would think of it it's only recently been thought of as an AI company now it's revenues currently less than 10% of revenues come from AI less than 10% of revenues so is it an AI company I don't know but that's not how I think about it I think it's got good exposure to AI I think AI will provide a good long-term tailwind to the business right but this is an electrical solutions provider and you know the US utilities themselves and so it's just like I think it's a great business without AI AI is like sort of that gravy on the top type of thing mm-hmm and then just like two observations about AI that maybe I'll close with tech disruption usually means historically has meant upstarts with a disruptive technology take market share from incumbents that's not happening to a large extent this time at least not yet rather it's the incumbent mega cap tech leaders that are the AI leaders so far and so the big business winners so far are the biggest companies in the world I mean you know them NVIDIA alphabet Microsoft Taiwan semiconductor broadcom right like these are the leaders in AI these are the company setting the pace these are the companies building the ecosystem what companies that I just name yeah I own all those companies by the way I need to say that and so that's one observation another observation would be this AI technological revolution is directly and closely tied to and completely dependent on physical infrastructure so I feel very lucky and blessed to be an infrastructure investor at this time I did write about this in my third quarter letter but tech companies these days are just trying to get their hands on GPUs or other ASICs or accelerators gigawatts of power and then all the electrical power and mechanical components that go into and outfitting a data center there was this guy on X dinosaur is his name at AI yoyo asteroid at AI yoyo asteroid it's actually a really thoughtful comment he said I never thought combined gas turbines will one day be analyzed by a semiconductor analyst so he was responding to something that Dylan Patel from semi-analyst put out about how combined cycle gas turbines see CGT's are the equivalent of like 10 jet engines and Dylan Patel put that out you know leading semiconductor analyst in the world a leading semiconductor analyst in the world and so dinosaur responded I never thought large combined cycle gas turbines would be analyzed by a leading semiconductor analyst but that's the world that we live in today very thoughtful comment about just how intimately tied technology companies which used to be thought of as capital light at least the hyperscalers used to be thought of that way or no longer and this is just it's I think I think it's a very good time to be an industrial technology and infrastructure investor at least I hope so I think I think that is the perfect way to wrap up episode 9 today's chat with J. Row J. Row I bear and I just always learn from you we think the world of you we want to just thank you from the bottom of our hearts for coming on we'll have you on again soon I know our listeners have completely kind of inundated how do you like that word bear us with a request to have you back on so we will but J. Row thank you so much for coming on and we look forward to talking to you again soon thank you for having me I learned honestly so much from y'all till you brought in my horizons bear is gonna have me looking at a stare again definitely gonna be top of mind while I go do some some research on it again so thank y'all both very much thank you so much John