AI Transformation: Generative AI is restructuring “WINS” industries, with leaders building hybrid human-digital organizations and laggards seeing valuation discounts.
IT Consulting: Gartner (IT) faces structural headwinds due to misaligned incentives and slow asset building, while Accenture (ACN) is better positioned with asset-focused ROA discipline.
Hyperscalers: Bullish on software-driven productivity at Microsoft (MSFT) and Alphabet (GOOGL), with Google’s vertical integration (TPUs) a strategic edge and MSFT’s internal cost savings a major value driver.
Semiconductors: Cautious on NVIDIA (NVDA) as algorithmic advances (e.g., more efficient models) could reduce primary compute demand and challenge hardware-driven growth.
AI Infrastructure: Massive data center buildout faces power and turbine bottlenecks, though efficiency gains could temper long-term compute needs.
Open Source AI: European OEMs adopting Chinese open models raises strategic and security risks for the U.S., potentially shifting AI standards and influence.
Autonomous & Robotics: Waymo showcases AV maturity while broad robotics adoption should surge; near-term service and maintenance demand supports the robotics ecosystem.
AI in Science: Expect an acceleration in discovery (e.g., repurposing compounds like lidocaine) as models search vast design spaces, benefiting healthcare and materials.
Transcript
Hello and welcome to the Stansbury Investor Hour. I'm Dan Ferris. I'm the editor of Extreme Value and the Ferris Report, both published by Stanberry Research. And I'm Cory McGlaclin, editor of the Stanberry Daily Digest. Today we talk with Dr. John's Viola of G AI Insights. John is an AI guru and get a get a pen and a piece of paper because you will want to take notes. He's just that kind of a guy. He's a great talker. We're going to throw a few questions at him. Start him up and he's got a lot to say. So get ready. We're going to talk with John. Let's do that. Let's do it right now. John, welcome back to the show. It's good to see you again. >> Dan, great to be here. >> So, Corey and I are going to pepper you with lots of good questions. Um, but I wanted to start by um telling our listener, if you could just tell our listeners sort of what you do because most of our guests, they are not like you. you know, most of our guests are are, you know, hedge fund managers and traders and analysts and, you know, equity analysts, all those people. Um, you're a little different. What do you do, John? >> Yeah. Well, um, really kind of a lifelong passion of mine is to try to understand uh what how do machines think, how do people think, and what does that mean for business, society, and individual productivity? Um, and so uh I started that as a professor at Harvard Business School many years ago. And uh what I do right now is run a company called GI Insights. And we believe that every firm is going to have to become a hybrid firm, a hybridization of human workers and digital workers. And our purpose in life is to make people aware of that and to to manage that transition in a productive way. So we do research, training, strategy, and community to help people accelerate that. The training sounds fascinating to me. Maybe we'll get into that. But um Sure. >> I wanted to revisit something that we discussed last time. And you you had told us that folks who work with words, images, numbers, and sounds, the winds wins winds framework >> would be most impacted by AI. >> Yes. >> And then you you kind of put some numbers on it for us as investors. You said we did some research with Valen's research. We looked at which companies are wins intensive. Lots of words and images, numbers and sounds. >> Yes. >> And you said we think 50% of the market cap and 50% of the profit of the entire publicly traded market is up for grabs with generative AI and AI. >> Yes. >> You think you think we're still at 50 or think we're more? >> Uh maybe more. Um because I think there's a knock-on effect of uh lots of market cap getting created to build infrastructure to support AI. So it's a secondary effect. Uh the primary effect is clear. I mean you look at what happened to Cheg's stock price. You look at what happened to Gartner's stock price. You look at what happened at Accenture stock price. Gartner's down 50%. Accentur's down 35%. Because the market is not yet convinced that they know how to be a hybrid organization. >> How do do you think they can do you think you know Accenture we mentioned last time or whichever one you want to take for an example or three or whatever. Do do you think they can? Do you think they're on their way? >> I think Accenture can. Uh Gartner, I'm not so sure. And I had to be a little careful because we compete with Gartner in some areas. Um so I'm not bashing a competitor here, but um the the reason is if you look at if you dig into the DEF 14A and you understand the comp for the board and the senior executives, >> um and I'm I'm on a public board. I've been on a number of public boards. So I and I've been on comp committees and so forth. their um their short-term and long-term compensation is based on two numbers. EBITDA growth >> and uh the growth of long-term contracts. Okay? And the problem with that is that if you're a Gartner, you're going to have to go from 24,000 people probably down to 5,000 people and you're going to have to build a bunch of assets that uh unless you grow phenomenally, right? Okay. So, whatever. I'm talking about the same revenue, >> right? You're going to have yeah you're going to have to build assets that are going to the in the early going could have a return on investment within a given year but it's going to take two three four years for these assets data assets process assets AI assets to pay off. So if you're incenting people on eB that penalizes that investment in those assets and so you're going to be disincented to actually build the assets that are going to be made make you more productive. So I'm unless I see a change in the senior management compensation at a Gartner, I don't think they're going to get there fast enough. Accenture I think understands it because Accenture has an asset measure in the very top of the house in terms of return on assets. Um, if I were running Gartner, I would I would uh give it IBDAR and the the the ad back in on Ebad would be uh research and development and the creation of software and data assets and I would advertise it over a 4-year period. Uh because that that's the operational reality of building um building cognitive assets. >> I see. poor incentives matter in in short >> especially at the top of the house right I mean we're talking you know for for your viewers you know the dev 14a lays out um my comp I'm on the board of infuse systems it lays out my comp everybody else's comp when I was a named officer at diamond same thing even though I was inside the firm so uh yeah incentives drive behavior funny how that works >> yeah executive I mean executive incentives aside which is really important a lot of people overlook in in the investing world. Our the last time we had you on it was November 2024. You said at the time kind of like CEOs, executives didn't really have any hands-on experience with AI tools for the most part. >> Yes. >> Yes. >> Has that changed at all or is that still the case? >> Yeah, it has improved. It's changed. But we're seeing u is a bifurcated market. So there's leaders and there's lagards and the leaders are moving ahead faster. So anytime you see an average number um like uh I MIT had this study which if you look at it was the the methodology is really terrible uh but let's just you know take their thing 95% of generative AI stuff doesn't yield value by the way they're an outlier nobody else is reporting that Wharton's reporting like you know over 70% and open AI anyway >> but let's just say that for an example the problem is that you don't want to look at the average because the le there's a learning effect here and one of the big differences people talk about uh AI and generative AI being a foundational technology um you know general purpose technology I think that's true but it's really not a technology it's a capability and the distinction I make there is a technology can be bought implemented and you get the yield so if I buy a faster welding machine with robots you know I can do the ROI I can get this many more you know autoframes throughout the whole routine when I'm talking about AI or generative AI I'm talking about how people think, how the machine thinks, and I'm evolving that over time. It's much more organic. So, I'm growing a capability and I'm buying a technology. And we still see the minority of companies buying a growing the proper capability. The problem with this, Corey, is that the market is going to mark your com if you're public or if you're a private transaction, you have public comparables. The market is going to discount your company. And we we're seeing this. It's going to discount your company if they're convinced that you have the old model, not the hybrid model. That's what's happening to Gartner. That's what's happening to CHEG. And so Gartner is still a very healthy company in terms of revenues coming in and profit going to the bottom line. But his valuation has been hugely discounted because the market believes they don't have the right model anymore. >> Yeah, that is the uh the stock market looking ahead effect that you've heard so much tell about all my life anyway. Uh and we'll I guess it remains to be seen. You know, >> the other thing, Dan, we're seeing it in private equity because private equity, uh, we're serving a number of private equity firms, and private equity really looks at generative AI and AI, I think, as uh, both something to help with growth, but also to radically decrease labor. >> And, you know, when you create a when you create an innovation, there's only three places the the the value can go. It can go to the customer in a lower price, like with Craigslist, right? Most of the stuff's free. uh it can go to the labor and uh it can go to the investor in a higher return or or it can go to the the labor in either training or higher comp. Uh private equity of course gives it almost all to the investor and the customer and very little to the labor um except at the top of the house, right? Traditionally and u so what we're seeing is that there's a decoupling between hiring labor and and revenue growth. And so that's the premium. So you look at somebody like Curser um you know the new AI startup focused on um uh software creation. They are at $5 million per employee in terms of revenue. >> You know OpenAI is over three million. So we're going to see massive revenue growth separate from labor growth. >> Yeah. I'm glad you brought up labor because that of course is an enormous part of the narrative from all kinds of people. >> Yes. >> From from the you know like the uh what was his name? Hinton the godfather of AI that from from everybody they're saying you know it's going to be apocalyptic. It'll put millions of people out of work. Um historically of course that has been predicted again and again and generally transformative new technologies have created whole new industries multiple new industries sometimes. Um what where are where are you on this? >> Well uh first of all we have to recognize that um I think through two things um u the lack of uh enforcement of antitrust which I think has been a bad thing for competition. I'm a real capitalist. I think you need to enforce antitrust >> and the like the Google thing, you know, that they let him keep the browser I think is nuts. Okay. Uh uh and so u I think there's that. The second thing is if you look at the percentage of uh profit that's going to labor, it's decreased by about 50% over the past 20 years. >> So labor is remployed, but they're getting less of the value. So I think it's really important when people say, "Hey, there's new jobs." jobs. Yeah, but those new jobs don't pay as well, right? And that's absolutely true. And the middle class has been going, you know, u has been going down in terms of standard of living and wealth for the past 40 years. So I think that's an important undercurrent to to remind folks. >> The the second thing is that we've I think we've had two great uh labor transformations in this country. One was the industrial revolution and the second was the computer revolution. Okay, in the industrial revolution, we don't teach this anymore, but it was incredibly violent, right? I live here in Chicago and riots down here, you know, the, you know, 12 year olds working in factories, you know, no benefits, nothing. Uh, 60-hour days, uh, you know, right before there was that and that was bloody people killing each other, the president sending the troops, I mean, the whole routine, right? >> Okay. So, we forget about that and we don't teach it anymore. The second one was the computer revolution, right? Because you think about like insurance companies, for example, literally employed thousands of accountants and and and bookkeepers and all that other stuff, all gone. >> Mhm. >> But at the in in the World War II period, not only were we spending 12% of our uh federal budget on new innovations like rockets and the internet and stuff like that. So you had all that investment which is now down to below 6%. So less than half, right, in terms of forward-looking stuff. The other thing is we have massive investments in labor liquidity. What happened was they were worried as to uh over 10 million guys were coming back from Europe. They were afraid of political instability in the United States because they'd seen all this stuff. They've seen communists and all routine, right? >> Mhm. >> Okay. So we had the GI bill, right, which made for education. It also made for housing. uh the relative cost of upskilling yourself was trivial compared to today, right? You could do it for less than one year savings from a job. Now you you basically have to save for 20 years to be able to afford a college education. >> Uh you didn't you didn't have the massive health care risk. You didn't have both spouses working. So labor liquidity was very high. Hey, I want to go move out to Arizona to get a job. I used to live in the Bronx. That's a lot harder now because I have to get two if I'm going to have healthcare, I have to have two people working like the whole routine. So labor liquidity is way it's more like labor liquidity was in the industrial revolution, not now. There's this great quote from Levit, the guy who did Levittown, the the cheap houses outside New York. Oh yeah. >> He said near about Yeah. They were worried about you know this this issue of communism political unrest. And Levit's quote went something like this isn't exactly right. They said, "Hey, look. If a guy has a house, has a mortgage payment, a car payment, three kids, a dog, and a wife, he's going to be too goddamn busy to be a communist." Okay. [laughter] You know, there's a lot of truth in that. You know, there is. >> So, my worry is that we don't have those same. If anything, we're disinvesting in all those labor liquidity things because I again, I'm a capitalist. I believe I want labor to be liquid, right? I don't want to be able to move around. And u all the variables are uh are much much worse now. They're much more like the first revolution, not the second one. >> And we have a a a openly socialist mayor of the biggest city in the country. >> Yeah. I mean, if you if you if we really talk about socialism, he's not a socialist. Okay. He's not going to try to grab hold of the we have a tremendous amount of socialism, right? Like we have uh the military-industrial complex is largely socialized. It's a cost plus business, very little bidding, you know, we have that, right? We have uh the government buys most of the healthcare, right? And so we already have uh uh we 12 billion is going to the the soybean farmers. I mean uh you know farming is socialist in this country, water is socialist in this country. um you know uh roads are socialists in this country. Those are the real socialist things. He's not going to touch any of that stuff, you know, and he's not gonna he's not also not going to have stateowned uh means of production. So I don't know why he calls himself a socialist. If you know anything about socialism, he's not one. >> Okay. So to him, socialism looks like um you know more free stuff is you know he wants well stateowned grocery store or city own city >> own grocery store. >> You think that's going to make a dent? Come on. Stateowned grocery stores compared to Target, Walmart and Kroger, forget it. >> Yeah. >> You know, I just >> Oh, I don't think it's going to succeed. I'm just >> No, no. But I'm but he's not saying a real socialist would say, "Hey, look, we need to take over Google, right? Because Google is absolutely an essential facility. It's got no governance except I mean that's what a socialist would do. >> I mean what did the socialist do in um in England? They took over the uh uh the coal mining, right? I mean this people they threw on socialism and communism. They don't even know what the heck they're talking about. >> Well, it's a Yeah, I'm aware that it's a sales pitch, but so is conservativism and progressivism and the others. They're all a sales pitch to gain control and get power and stuff. So to me they're all >> yeah I think going back to like the u why why the messaging is is appealing though you know say to New Yorkers that's the point kind of what you're saying John right about like >> the just the situation of people with labor and all people trying to keep up and and afford things afford to live basically >> there's no question that the majority of labor is getting less of the economic value out of our society forget the wealth effect that to the other thing I think people totally misread one of the core things which is that it's not wealth differential it's risk differential so the average worker today has longevity risk they don't have enough money they have healthc care risk which was never look uh one of my kids has MS and the cost of his MS medicine is 230,000 bucks a a year okay So yeah, this personalized medicine stuff and everything, it's expensive. And so there's that. There's uh education risk. I mean, kids, the fact that kids that they have that I I I I really believe in bankruptcy because bankruptcy is a critical part of capitalism. The fact that student loans survive bankruptcy, I think, forget it. I don't Look, if you're a bank and you give some 20-year-old a quarter million bucks, it's your freaking problem. It's not my problem and it's not his problem. That's ridiculous. So I I mean we have car I think of that as socialism. I mean you know what the heck? Why are we supporting >> Huh? >> Yeah they're they they don't have the bank doesn't have any risk. The loans guaranteed. >> It's ridiculous. I mean that let's go back to capitalism. You loan somebody something you go at risk. That's how you make the money >> right? Anyway, so u you know, so those risks, longevity risk, health risk, >> u education risk, those are the things that people wake up at night about, >> right? I know I do and I and I and I've got a great situation. >> Absolutely. >> Yeah. >> Yeah. I mean, I did a age calculator, you know, I'm going to be live to 90 according to the age calculator. Okay. And I probably got like a one in three chance of having a, you know, uh mental disease, right? to have some kind of dementia or some like that. Like, it's expensive. You know, in the old days, I'd be smoking pots, I'd be drinking booze, and I'd die at a reasonable 75. You know, [snorts] >> I I'll I'll send you a case of cigarettes, John. >> Well, actually, what the >> know when the dementia kicks in that we laugh, but the the French as they put in their anti-smoking campaigns, >> the smoking consumption went like this. Body mass index went like this. I talked to some folks at a a large industrial company I don't want to disclose. What a lot of people don't know is smoking sessation actually increased their health care cost because people instead of dying of lung cancer fast they die of diabetes slow. Okay. So smoking sessation is actually bad for health care costs because people still put stuff in their mouth but the m the thing they put in their mouth is food. I've never heard this, but it makes perfect sense, doesn't it? >> It's crazy. >> And and I have to tell you, my [clears throat] my wife decided just to stop drinking. She got some values on a liver test a couple years ago, and she said, "I don't like that." And she just stopped on a dime because it wasn't like the most important thing in the world to her. And her son did the same thing a year ago, and they both report exactly what you're saying. They're they're both like all practically candyholics at this point. I mean, they look great, actually. They're they're both in better shape than ever. >> Sure. >> But but they've they've definitely substituted the candy, you know, and I could see in their case, they're cool, but I could see it getting way out of hand with a lot of people. >> So, it makes sense. >> No, it's it's not anyway. So, I think those the risk differential is at least as important to me as the wealth differential. >> Right. So, we we start we got into this by talking about labor. >> Yeah. and and you know like like Musk is Elon Musk you know his famous quote is like in however many years nobody will have a job and you know everybody will get everything they need or something like that >> but but you know that's that's a very that's that's a huge risk right to to say the end of scarcity or you know extinction I don't know if you saw that chart in the Financial Times actually the Dallas Fed put out the chart and and it was like extinction on one risk and and then you the end of scarcity meaning we're all just fine on the other end I mean this uh >> a wide range is a lot it means high risk right yeah if you treasury bills have a small range of outcomes right >> mining stocks have a big range of outcomes they're much riskier less you get it >> and and and AI from that perspective looks extremely risky to me >> and labor is at the that's in the crosshair anyway >> yeah I think look I'm I'm a real believer in absolute wealth, not relative wealth. Okay? >> And so, uh, I think a lot of wealthy people would rather have relative wealth than absolute wealth. And what I mean by that is that if you take a small amount of the profits that are now going to capital, and you reinvest those in forward-looking stuff like we used to, like the like pretty much every major invention that we that made great companies were started by the government or by academic institutions. So the internet, GPS, all the drugs we use, modern farming, like all of it, right? Began with research that was done by the government or by academic institutions or what I call gift culture institutions. Like all of it, all this stuff. Look, Elon Musk, you and I have invested in Elon Musk at least twice, probably three times. Uh first we invested in all the technology he used to build his cars, right? of the the battery technology and all that stuff. Second, we gave him a direct loan when he was about to go bankrupt. Third, the NASA contracts kept Tesla and SpaceX afloat. So that's our tax dollars. >> So he's a socialist leader, right? That happens to have I mean socialism has saved him, right? Um and so the the the fundamental science is uh almost universally um built by the government. Okay, we're disinvesting on those and on a and and the reason Okay, so that's one thing. The second thing is take a little bit and reinvest in the labor the way we did with the GI bill. That makes my capital more productive because those people are building stuff that's productive and we're winning on a global basis. So they're getting economic rents from the rest of the world, right? And then I have places to put my capital because those people are now consumers, right? This current approach is weakening both of those, which means that my capital in the future will be chasing returns more and more, right? The question is how do I chase those returns? Well, if I don't have growing productivity, the way I chase returns is to seek scarcity. So I get hold of water or I get hold of oil or stuff like that, right? If productivity is not happening. And so I want the world to be like Henry Ford. Henry Ford doubled the living wage of his people from two and a half bucks to five bucks for two reasons. He wanted to he wanted to skim the market on better talent and he wanted to have a consumer there. So my capital is more productive if I have a healthy middle class that's consuming and being more productive. So I'm going to make more. That's why absolute wealth. Now, if I want relative wealth, I can go back to 1066 and sit in the Tower of London freezing my butt off in a dead animal, but I have I have, you know, three dead deer in the closet that nobody else has. So, I'm I'm the wealthiest guy there, right? [laughter] >> I like the absolute thing better. >> I like the absolute thing better, too. >> That would be better. The deer the deer outside aren't aren't very happy with that comment by the way that I got out here. >> Hey, by the way, you know what? What do you think the uh the the uh uh main course was at the first Thanksgiving? >> And it's not turkey. >> Oh, not turkey. Oh, what was it? I don't know. >> Was deer venison? >> It was deer. Okay. >> Yeah, >> makes sense. >> Yeah. But I think that So, back to your question. Yes. If if we do not invest in um new science um what Ben Franklin did with the um the public library there is no public library for AI right now. you need one, right? And more uh funding of fundamental research that open sources, data, weights, models, availability, there's some, there's not enough in my opinion. The Chinese are actually doing much, much more of it. And this is really scary on a global basis because the Europeans now with our new aggressive antagonistic stance with our allies, our traditional allies, >> what's happening is they are now building Chinese AI into their product. So dollar Benz is using the Quen models right and this is I think a big mistake because for the past 60 80 years we've exercised a tremendous amount of security and economic value by uh having our partners like build in our telecommunications I don't know if you remember they pushed back against the uh Huawei switches for example right when the Chinese were trying to penetrate the European we put pressure Don't buy Huawei, buy American stuff. Right? We've done it with our with our software. We've done it with our computer chips. Right? What we're doing now is we're saying, "Okay, that next generation, go ahead and build the Chinese models and we don't care." I think that's a strategic issue both from a productivity standpoint and from a national security standpoint. And we are actually making it super easy for the Chinese to infiltrate our traditional allies. >> Okay. I need you to flesh this out for me a little better. I >> sure say Sam Dmer Benz. Okay. >> Right. >> If you look at now in in AI, there's there's two big uh uh u you know, leaderboards. One is the proprietary model. You have to pay for it. Open AAI, that stuff, right? >> And then there's the open model, you know, Deep Seek, Quen, Z, so forth. >> Okay. If you look at the leaderboards, you take the top 10, there's only one Chinese company that's in the top 10 of the paid models. There's only one American company in the top 10 of the open models. >> Okay. So there that now there open's a little tricky. There's three parts to open. There's data, there's models, and there's the the software. There's data, there's weights, and there's models. Okay. The Chinese are not opening source open sourcing the data but they are doing the weights in the models. So I can use if I'm dollar benz and I want to I want some intelligence in my dashboard and I want to run a 70 billion parameter model I can use the Chinese models dropped right in there I don't have to pay IP back to the Chinese. Okay. If I do that with all but a couple of the American models I do have to pay IP back to the Americans. In addition, when the president is saying things like NATO's gone and all that other stuff, of course, I'm going to go to the Chinese. Who else am I going to go to? I mean, I've got Mistl and I've got, you know, a deaf alpha out of Germany. I've only got two models and I got Falcon out of the UAE. I've only got two or three options to do anything like a good model. I got the Chinese who are going open and the Americans who are going closed. >> I don't think that's great strategically for us. It's ironic too to hear that, right? >> Yeah. US closed, China open or and >> well except the China they they have they the idea of intellectual property in China is just you know >> right >> it's non-existent basically >> you know so it makes all the sense in the world >> it's the same thing the Americans did like Moody Street and Waltham mass where the American industrial revolution started is named after it's the main street in Waltham Massachusetts where the Charles River drops three feet so that's where they first put it and then they went over to Marramac because it dropped 13 ft anyway for more power. Moody's claim to fame. He was an engineer, American engineer. And when the shuttlecock loom uh was making uh British um you know um British uh weaving way more productive. Uh if you left England with blueprints of the shuttlecock loom, you would be put to death. Okay. >> Whoa. >> If you're caught. Legend has it that Moody memorized it and came back to the United States and put it in there not paying any royalties back to the Brits. So, you know, and the same with the books, you know, like Charles Dickens and all that other stuff, we stole all their stuff and just published it locally, right? So anyway, but the we our problem is that when we let the Chinese in, we let them for market access and I talked to a bunch of folks in in the intelligence agencies. They would go to the big companies and you can imagine who they are, aerospace, industrials and stuff like that. They say, "Hey, the Chinese are stealing all your stuff." And the the the companies wouldn't raise a stink because they still wanted access to Chinese markets. Okay? And I think that was a tremend I think that's where folks like the the the government should have stepped in and said, "No, no, no, no. You can't because guess what? American consumers and American taxpayers built that IP. You can't simply give those decades of intellectual property to them for your incremental market access, >> right?" Which is what they did. So, let me give a specific example. I forget the name of there's a Chinese drone. Looks just like the General Dynamics Predator and at the weapons uh fairs they're selling it for onetenth the cost. Sure looks a lot like it. Looks a lot like it. >> Yeah. Yeah. There's a lot of >> I saw that a lot of other stuff in the west. >> Yeah. I saw that recently with what the when the humanoid robots were having their moment a couple months ago and there was a Chinese one posted online through like a Walmart affiliate site for like >> Sure. >> three days before it got pulled down for >> Yeah. >> No public. >> We have to Oh, sorry, Cory. >> No, go ahead. Yeah. >> Yeah. I I I I think that's true. I think that I think that we have to update the narrative, though. They're not just copying. I mean, you look at some of the stuff they did with the Deep Seek model, for example. They're advancing algorithmic innovation and that's really important because the reason that these that uh chat GPT and Gemini and so forth consume so much power and compute is because the transformer algorithm which is underneath most of these things and and gave us this unbelievable I mean I never anticipated the quality of the interaction you have with these things even back in November 2022 when chat GPG35 hit the scene right >> is that they are computationally inefficient Well, if somebody comes up with a really efficient algorithm because the algorithm, the transform algorithm brought us this step change in in in uh in performance, right? >> Yeah. >> Okay. If somebody comes up with a step change in performance on the compute side and there are some some folks who are using other kinds of models, well, guess what? Then primary demand for all that compute goes down. That's why uh personally I don't I don't uh invest in the hardware stocks because I think algorithmic innovation can radically change um the um the demand for hardware >> and it sort of needs to go down if you do all you know if you do the arithmetic on the power and what's needed to be built and it's sort of we it it kind of can't be done very quickly and and you know if you look at for example you know they're the the turbine makers just you know natural gas turbine makers >> right >> it's all that's the that's the uh the bottleneck right so >> I've heard I've heard it's about a fiveyear weight right now We're we're working with one of the folks who are building was building the largest data center shell in the world and they they have eight >> they have uh five gawatts of turbines lined up and that's a huge part of their market valuation is the fact that they have the contracts and they have the permits. >> Yeah. They just need the machines. They need the turbines. So, so in other words, I'm just saying that it it it makes sense to me what you said that you will you will need you will need that greater efficiency. In other words, like when when deep it makes it made all the sense in the world for Nvidia's market cap to get hit, you know, the day Deep See came out with its big announcement. And there ought to there well ought to be more such announcements in the near future, shouldn't there? I mean this this is you know it's it seems an extremely important part of the equation for someone not to be saying hm you know maybe if we maybe we wouldn't have to build all these power plants if we just had you know better software basically better algorithms. >> Sure. Yes. And there's certainly a lot of people working on that problem. Uh and the Chinese are working on it. I I actually think it's good news that we're starting to sell the Nvidia chips because that'll keep them fat and happy and they won't be so they won't be so desperate to improve the algorithms which I think is a good thing that net >> um for us u I mean I should I should share um two two u points of view I have first of all I I do believe even with all our faults that uh I do believe in America as it in its value systems and its capabilities and and its role in the world and I'm kind of sad we're pulling back as we are. Uh and the second thing is I'm a rail capitalist so I I like competition. Um I'm I'm sad that um you know we've allowed these massive concentrations in so many industries. If you go back to the Sherman Antitrust Act, you know, uh, John Sherman, who was Tecumpsa Sherman's brother, the guy from the, you know, the north who did the whole Sherman's march in Georgia and all that other stuff in in some of the words about that, there were two parts to the antitrust thing. One was consumer harm and the other was companies having power as powerful as governments, right? And Sherman had this great quote. He said, "Why? We just fought a war to not serve a king. Why would we allow a company to control an essential facility where we have to bow down to them just like a king? Okay. Now, the the interpretation of that act has gotten rid of the market power piece and only kept consumer harm largely due to the guys down the street here at University of Chicago. >> Uh but, you know, I think these I think these mega scalers have way too much uh political power. Mhm. >> Um, and I think and I don't think we should have a company owning most of the satellites up there, which we have right now, and not being underneath military control, you know, again, I I believe in America and I believe in competition, so I don't think we should have people with monopolies on essential facilities. >> Yeah. Um, let's talk about um [clears throat] your last last time you were here, you introduced the four levels of generative AI adoption. And I'm and I'm wondering how much that has changed, if at all, if you still use that framework, because I know that since last time we talked, I went from what you would call a toe dipper, if that even to um someone who is trying very very hard to be an intelligence leverager, and to load as much of the best data I can into the best, you know, um system that I can get, you know, that a guy like me can get on his lap. top. Okay, just put it that way. Um to to make the to do the best research I can on on all the public companies that we write about. Um so, you know, I really I I just can sense the power of it. I haven't really, you know, I've learned all this. I've learned how to collect all the data I want and put it all in one place. Now, I just need to learn how to prompt it and analyze it and and get what I want out of it. Um, so I it would seem to me you're probably in the business of teaching people how to be intelligence leveragers uh to to a great degree. Yeah. But I'm wondering something else though, John, just before you tell me talk about intelligence leveraging and what you guys do. Um, what I'm wondering is, um, since you last spoke with us a little over a year ago, um, have you heard the same experience that you just heard from me from lots of people? >> Yeah, absolutely. and and the a key differentiator is how for the firms because think of it again as these two populations and and we're seeing significant return on assets differentials in the intelligence leveragger kind of folks >> like you know ROAs of you know 21 22 23% growth much faster things like that um the the capability level is the key indicator of if you are going to get value from AI and >> capability level >> capability level and so to your question yes we absolutely are not only using that framework we updated we have a little bit slightly different language to make it easier to remember we call it the rise framework so there's research and experiment and education >> then there's islands of innovation >> then there's scaling and synchronization >> and then there's uh uh emergent intelligence that's the intelligence leveragers okay It's a rise and we have we have yet to find a company that can jump straight to level three to go system level stuff. we see more what happened with you Dan which is you experimented you probably did some little innovations right now you're trying to think okay how can I really reinvent the way I do my work I don't want to put words [clears throat] >> but >> every day that's all I'm concerned with now is that I'm I'm re I'm refiguring out how to do my job all over again >> right but if you had I would say and feel free to disagree if you had started that at the very beginning before you had those other experiences you wouldn't have known what you were talking Yeah, that's been my >> Sure. Yeah. >> Yeah, that's been my experience, too. You know, I I originally, you know, I was playing around with all these tools a year or two, you know, a year and a half ago, and there's run run into a certain level of friction, right? And and now, you know, you you learn the pro, you learn what this model is, you learn the prompting, the importance of the prompting. You know, I'm talking about for writing and editing purposes and research. And then now it's to me it's a matter of work somehow working it into like my daily workflow is seems to be the the biggest challenge >> like what's worth spending time on versus what's not. >> Um that's like the biggest thing for me and I als you know I also you know struggle to make decisions sometimes. So that doesn't help either, you know. >> Um so that's I I don't know that is that that's a is that a common uh thing that you're trying to address as well? >> Yes, absolutely. and and the the kind of dirty little secret, you know, of of of executive education and, you know, I'm I'm a fellow at uh Harvard Business School and I taught there for over a dozen years and so forth, is that if you go to an executive education class at Harvard Business School or any any good uh you know, academic institution, a core skill when you're teaching executives is to teach them things that they should have known anyway, but they can't admit they don't know in an unemb embarrassing way. Okay. [laughter] >> All right. >> And so that's where AI is in most companies. People like it's the because the windows closed to say AI, what the heck is that? Could you like slow this down and tell me what you're talking about? >> Right. >> And that's how most executives really think, but they can't say it because it's not it's not socially acceptable. AI, what's that? >> I mean, if you go back to the basics, we can't really define artificial and we really can't define intelligence. you put those things together, you know. Um, so that's why that's why this model is so important. You have to have hands-on experience. There's a great uh there's a great quote from Frank Zappa when he was somebody's asking about a a music critic and he said, "Writing about music is like dancing about architecture, right?" >> Oh, dancing about architecture. Yeah. >> Yeah. And so if you don't experience what it's like to talk to a silicon intelligence who can literally talk to you in any language at any level of specificity and now with some of the new models, especially Gemini 3, you can say, "Give me an illustration. Give me a cartoon. Give me a what? Give me a movie." Right? If you haven't experienced talking with this silicon hive mind that has read literally everything that's been written, you don't know what that's like, right? And and now we have all these low code no code things. Anthropic just came out with a open no anthropic just came out with an analysis that said inside their company almost a quarter 23% of the low code no code stuff they're doing would never have been done because it was either too low down the priority for it or the person who understood the problem thought it was too much of a hassle to learn the technology. So back to what you're saying Corey people are using this stuff just you say okay should I automate that do I do it enough and there and I think the ripple effect of that productivity at the individual level is going to be fantastic but then it's like special forces like we used to compete with regular army and special forces we crossrain we upskill we use we take traditional technology we modify it to make it better and just like with special forces you know regular army might have taken a 100 people to to do something special forces might be able to do it with six, you know, and that's what's happening in organizations and and to the organizations that are upskilling. So, back to your question, Dan is absolutely central. Senior executives resonate with it. The other thing is it helps it helps sort what the vendors are trying to sell you. So, there'll be one of the big four went to a and I don't want to use names, right? But one of the big four went to a a consumer brand that we all know, right? And they came in and said, 'Look, give us 5 million bucks and we'll go find 25 million bucks at of system level improvement. Complete and utter failure. Why? They tried to jump straight to level three. They tried to jump straight to system level change. So when they're talking to people, they don't I mean, we're having this AI kabuki where we're saying stuff, but we really don't understand each other. And then they have no hands-on experience. They haven't done islands of innovation. So they don't know what it's good for and what it's bad for with any even if they're not building it. You know, you have to have you have to you have to drive a car to understand what's valuable in a car even if you don't know how to build the car, right? And so yeah, that and so they jump straight to system level stuff, burn through the five million bucks, total abysmal failure. And we see that happen again and again. >> John, I wonder um I I am not a uh company executive, so I'm not afraid to say I don't know. [laughter] So I don't know what low code no code means. >> Okay, that that's basically u think about programming in English. Hey, I would like to you know I would like you to go and look at this uh set of spreadsheets and then look at this set of objectives and then generate a six-page PowerPoint thing every Tuesday at 10 a.m. in the morning and to say it like that not to have to go into the language of all the different software that would generate that. Ah, >> so low low code no code sounds like people know who know how to prompt AI. Well, >> well, yes, but often prompt. Yes, absolutely. In the with a capital P prompt. So, if I'm prompting to build an artifact which then will do stuff for me as opposed to prompting in a dialogue to get an answer. >> Ah, okay. I see. So, um could I see. So not even so the PowerPoint is a good example, but it sounds like um low code, no code creation of of uh you know modules of code that then do things too. >> Exactly. >> Okay, I see what we're talking about. >> Look, we've got >> that's that's that's really cool. And that's why I guess Zuckerberg took out all his mid-level managers, right? This isn't what we're talking about. Loco, no. Okay. >> Exactly. And so 10 million people speak Java, 15 million people speak Python, everybody speaks their own language. >> Yeah. >> So that's the thing that we've just taken, you know, everybody can be a programmer, >> right? >> Yeah. So the winds crowd um may may have a bit of an advantage here if they're really good with words and numbers. >> Absolutely. The >> It's not hopeless. >> No, no, no. But back to employment. We've already seen a tail off of employment of new kids >> coming in. We see a tail off of software programmers who have less than three years experience. >> People ask me what their kids should do. I say two things. Well, sorry, three things. No, sorry, four things. U the the I'll stop at four. No, the first of all, if you want to learn a trade, that's a good thing because we're going to be short about a million folks at least of plumbers, carpenters, all that other stuff. As well, if wealth differential continues, you look at the consumption of the top, you know, 1 to 10%. We got a lot of Maria Antonet going on, right? People building houses they don't need, you know, and they're super specific. They're going to absorb all kinds of trades, right? So, you know, that electrician that, you know, used to be building, you know, houses for the middle class and making whatever 20 grand a house is now building, you know, some 6,000 foot mansion in Tucson that they get used twice a year, but the electrical bill is 100 grand, right? >> And so, and we're seeing that happen. So, the whole Marine Antinet effect, right, that that's going to create primary demand. There's also just a raw need. So, you want to go in the trades, go for it. And and by the way, the trades, this thing is going to help the trades phenomenally. 26 is going to be the year of of uh mobile intelligence. And I don't know if you've done this, but just take this thing and if you have any question, you know, you're uh like I was looking at my water heater the other night and I was trying to figure out if I had a low water shut off. I just hold this freaking thing up and I take a picture of it and it'll tell me if it's got a low water shut off. How do I repair it? Where can I get somebody? The whole routine. Okay, so just imagine what that does for the tray is what's this part? How do I do this? How do I do a hip roof? Right? >> This will tell you today retail with no right, which is unbelievable. >> So you have that. So trades are going to go like crazy. Second thing, mathematics, especially matrix math. Math is the lingua frana is the universal language under all of this. So if you understand mathematics, fantastic. Third thing, sell. Anybody can sell stuff. as the as as as as we go from army to special teams and you get smaller and smaller groups, the premium and the value of people who can do demand generation is going to go up, not down, >> right? Because you still have to sell stuff to people to get them to, you know, commit. And and the the last thing and perhaps most importantly is do what you're doing, Dan, and what you're doing, Corey, which is show me your posy, right? So, I don't think any company today should hire somebody who hasn't built a GPT, a gem, low code, no code. Show me how you're using AI to automate your life. And I don't care if that's in work, your passion, family stuff. I don't care what it is. Show me. And I only want to hire those people who are doing that because individuals are not how if I've got no experience doing software coding and I want to get to be and I want to compete with the three-year folks which is what the numbers show according to signal fire venture capital company you know the massive fall off on if you have less than three years experience how can I simulate three years of experience I can come in with my posi oh let me show you my code review thing let me show you my documentation robot right so you're hi people should be hiring posies of a human and a bunch of robots, not just humans. >> I'm glad that you broached the topic of robots like because when I think of robots, I think of the physical ones in factories. >> Yes. >> And and um that that has some impact in the trades, doesn't it? >> Uh yeah, it's going to increase primary demand for the trades because all those robots are going to need to be serviced. >> Okay. Of course. >> Right. and and I we'll have robots that'll service robots, but there'll be a I it's just like when um the uh I was at a a conference, it's kind of funny. I was at a conference in San Francisco when you know the self-driving thing just started taking off and and people were saying, "Oh, what are we going to do with these three million truck drivers?" And I felt like saying, "What are we going to do about the 15 people I walked past on the street on the way here?" But anyway, the uh [laughter] >> let's solve that problem. Anyway, it turns out actually if you remember what happened when when between when self-driving started which is brought to you by the government defensive research projects agency did the Mojave Desert Challenge right and only after the two number of teams CMU and a couple others successfully did that under government funding that's when the the only after they've done that translational research that's when Bin called a guy named Chris Mson from CMU and started the Google car project which was the first among the self-driving car projects. Right? So all that Whimo and all that other stuff again started like the internet started with government investment. Um the the the what happened instead of 3 million truck drivers being out of work is with the microtargeting of Uber and micro logistics of Uber and my supermarket and everybody else driver demand actually went up in the near term. Now, in the long term, as Whimo comes in and the little Whimos, right, little non-car size stuff, will driver demand go down? Sure. But it's going to take a while. Same thing's going to happen with with robotics. Going to have massive adoption of robots. Huge increase in the need for mechanical engineering and repair over time. Will the robots build the robots? Sure. But it's going to take 20 years, just like it's taking 20 years for the driving thing to happen, right? Mhm. >> So I think we have it's just like the paperless office. I I have a friend who made a boatload of money when Xerox announced a paperless office. He invested in paper companies because as soon as they had as soon as they [laughter] had the the laser printer, the first salvo of the paperless office was an explosion in paper consumption, >> right? >> You know, same thing happened with driving. Same thing's going to happen with robots. >> I see. Do you use Whimo? Do you Whimo in Chicago, John? >> Uh, the answer to the first question is yes. The answer to the question, the second question is no. I wish we did. I love Whimo. I mean, I was so impressed. I mean, when I use Whimo, it's just it's a totally different experience. And >> I've heard that universally from everyone. >> Heard that from every single person. Yeah. >> I was staying in a club on the top of Knobill and so we got in the thing and it went up over Knobill. It was an illegally parked 18-wheeler at the top right in San Francisco. Then there were some people who were jaywalking and then there was a car that was going around the illegally parked thing. It managed all that seamlessly. I was like, "Wow, that was just like [laughter] I I just happened to go that way. My first Whimo ride." It's like I am impressed. >> Wow. Yeah. >> Yeah. >> Yeah. How long before Go ahead, Cy. >> Yeah. How long before Whimos are They're coming to coming to Baltimore, uh Philly and St. Louis now. uh Pittsburgh too I saw. Uh so we'll see which which brings me to like one of the things I wanted to ask you well you already answer I mean you said so many fascinating things already has been great and one of the things I wanted to ask you was like you know what what what for you know what skills for uh somebody trying to to learn all this and you already said that so um I'm also our crowd is also interested in like attractive investments or sectors um as this whole you know as the whole story plays out you already mentioned you weren't really not into the hardware companies. Um are there any areas like that you're uh particularly optimistic about or sectors or or that sort of thing? Um >> sure. >> Yeah. >> Yeah. >> Yeah. Absolutely. Well, first I think um in the in the um AI world, the hyperscalers, the hyperscalers I like are the ones that have massive u productivity enhancement through software creation that improves an already great model. So who is that? That's Amazon, Google, U Meta and um and uh uh uh Microsoft, right? Uh Apple should but we've seen no evidence of their ability to do it. maybe they can pull it off with a you know their their version of the AI iPhone or whatever it's going to be right but um so and then of those you have to remember Google is the only one that's vertically integrated in the whole AI stack right from the customer all the way down to the silicon right so they have their TPUs and I think that's going to be a huge advantage because the the the performance characteristics of delivering instantaneous intelligence across any language in any location is is a massive engineering challenge right just the the latency see the response time for for you not to perceive. For example, 11 Labs has got this new thing out where it can real time listen to the three of three of us real time uh instantaneously transcribe and then rebroadcast in 20 languages with 150 millisecond delay. >> The reason 150 milliseconds is important is because that's the threshold for perceiving a delay in conversation. >> Wow. >> So you have to engineer the living daylights out of that, right? And Google is amazing at engineering. Um, something like Nvidia, I'm not saying they're not worth it. It's just so many things have to go right. And I look at the hyperscalers, you know, these idiots who say stuff like, "Oh, [snorts] it drives me nuts about the the AI bubble." Okay, first of all, let's separate two things. Is there overinvestment overpricing of certain things? Of course, it's a market, you know. On the other side, is there an AI bubble on adoption? Absolutely not. Adoption is accelerating. Okay. So, pricing is different than adoption. >> Sure. Sure. >> Then the second thing is will everybody make money? Of course not. That's why it's called investing. We sit here, you and you and me and you, Coran, you you know, the three of us make three different bets going forward. The market can't support all three at the level we're hoping it'll get to. So, you guys pick the right ones. I pick the wrong one. Of course, I'm going to lose. That's the nature of investing before the fact, right? >> Yeah. And so that's nutty. Um, and last but not least, so, so of course I can sit like Norio Rabini and say, "Oh yeah, stuff's going to go down in the future." Yeah, no joke. You know, the sun's going to come up, too. I mean, give me a break. >> The um, >> yeah, >> the but the most important thing is if you want the clearest use case for productivity with AI, bar none and adoption, >> software software creation. >> 40% of Microsoft's cost base is software. >> Okay, >> if I can take that number down by 12%, 30% of 40%. >> Right? 30% of that is a massive increase in in Microsoft's market cap. They can spend billions. They generate just under $2 billion of EB a week. They can spend billions of dollars on this. And if they never sell any co-pilot to anybody, it's going to be worth it to them. So, you know, people are now open AAI, I I personally wouldn't have the courage to invest in Open AI at the valuations they're talking about because they don't have $2 billion of IBIDA a week coming in and they don't have a a 30% 40% cost base that they can shrink. So, I would say that. So, that's that. The second thing is I would look carefully at these [clears throat] new companies that are um that are doing uh vertical uh rappers and implementation. So you take a company like Harvey in the law. Okay. So Harvey is now adopted by a number of law firms. You know, I think >> uh my oldest boy is an attorney. He's a prosecuting Chicago uh attorney in Chicago. The good news is he's going to still be employed because Chicago's not going to get any less crime written and and so he's he's going to have a job as long as the state doesn't go bankrupt, right? Which it could also do anyway. But but you know, if you're working for some corporate law firm working on deals and stuff like that, they're going to have half, a third, a quarter of the people, right? And certainly if you're working in a because you know a lot of the I mean think about it, the law is essentially a badly designed large language model, right? um kind of you know um and so the you look at somebody like Harvey they not they've not only succeeded in getting good market share but now they're learning how to really tune those models and they're having a whole longitudinal thing of data you know just like um you know Facebook has my whole longitudinal thing of my social network they're doing that in different vertical domains I think those are areas that are very promising and so there's that uh the second thing if you have attackers who are coming in underneath and we like to think of ourselves as one in the research area that if you if we can prove that we can scale you know I think folks like us coming in with the with the Amazon model when they're competing against Kohl's and Macy's right >> in any given industry and those are going to start to appear they usually appear about five to seven years in for the retail customer if you're not in venture So you know for example the the financial disruption usually takes somewhere between four and seven years. So if you look at for example the the the price of New York taxi medallions 5 years after Uber entered the after Uber already entered not only being company but had entered the New York market the price of their taxi medallions was between 1.2 and 1.3 million bucks per medallion. The sixth year it was 35,000 bucks. >> Now they've recovered to about 200 to 300,000. >> They're never going back up to a million two. >> No. >> So financial disruption, you know, my my ex-colague and God rest his soul, you know, Clay Christensen talked about disruption, but he missed financial disruption. He got operational disruption right. Like you're going to buy from a mini mill, not from an integrated steel mill, right? >> That thing. But the financial disruption, that's what's happening to Gartner right now. They're getting financial disruption. Unless they can convince the market that they can get that they can do the new model, they're never going to recover. And so if you invest in those companies that are coming up underneath, the Ubers of the world that have the new model in professional services, in science and discovery, what a lot of people don't understand is we're going to have an explosion of science like we've never had before. Right? the ability of these models. If you think about science, okay, think think about how you improve stuff and I call this stuff practice. Um, practical knowledge. >> So, let's let's let's separate knowledge and I know I've read enough philosophy to know that this is grotesqually oversimplified. So, [snorts] anybody actually knows what they're talking about, I apologize in advance, but let's keep it for for us. >> Two hunks of knowledge. There's what I call practis, which is a provable thing that regardless of what your belief system is, works. So Mike Johnson, you know, is a flat earth guy, right? >> And I mean, he's a young earth guy. He thinks the earth was started 6,000 years ago and stuff like that. The speaker of the house, >> he uses technology that his belief system would never create. Okay? Like he like his belief system would never get you a cell phone, would never get you an antibiotic, right? The whole routine because he doesn't believe in science. So he never he I don't think he should be able to use all that other stuff. But that's another thing. The if you know I think if your belief system would never discover the things you use, you shouldn't be able to use them. But anyway, um so if he wants to go be Amish or something like that, that's okay. Go for it. But um the so what's hap so prais is provable like we can't argue about the fact that you know if you have this bacterium and I put this penicellin on it it's going to be dead at this time we can't argue about the fact if I design a nuclear weapon like this and I do this you know kind of explosion it's going to create this kind of bomb right >> that's a very specific kind of knowledge doesn't mean doesn't matter what you believe it's true >> and we can repeatedly do it >> right >> right Okay, that and then there's all the other stuff like political beliefs and religious beliefs, those are all socially constructed and navigated, right? >> Right. >> You know, and back to Mike Johnson, his phone works even though his belief system doesn't believe it should be there, >> right? >> Because nothing about anyway, you know what I mean? So, so what's what we're gonna what's beautiful about AI and the math underneath AI, the matrix algebra, is that it is incredibly good at creating practice, right? Because I can say here's what you need to look for as an outcome to prove that it's right. Okay, so what's an example of this? something like will this molecule um for example lidocaine lidocaine completely commoditized molecule. It turns out at Google they were looking at um they were creating um a a system to help uh think about new molecules that might be relevant to different disease states >> and they look at them functionally and they look at the trials and they look at all the data they can get to say oh okay here's something we haven't thought of might be useful. Lidocaine turns out is very good for certain kinds of breast cancer. Okay. Now, nobody even thought to look at lidocaine. Okay. And it doesn't reverse it, but it does stop it. Okay. That's only possible because the AI allowed for a robust description and an unbelievable search space where it could go look for stuff that no one no scientist had ever thought of, right? It's just too big a space. So, that's a that's a perfect example of the creation of practice. >> Okay? Because now I've got rational reasons. I'm looking. I've got an objective function. I can say does it help stop cancer and does and then what's the mechanism of action and then like that. Okay. It turns out for example in drugs we believe that we've maybe maybe explored 1% of the useful compounds to make humans more healthy. Maybe 1%. In material science it's less than 1%. >> Okay. I see what you're saying. This is this is um Thomas you know, structure of scientific revolutions, not on steroids, but you know, the Ferrari, the the you know, the the rocket ship, >> acid, iaska, whatever, you know, it's like >> Yeah. >> Yeah. And and and the thing is that all the models speak math. >> Yeah. >> So they all can talk to each other. So you'll find all these random things that you that that would take a long time that nobody would ever think of. Some outlier student that nobody listens to might come up with it. Now you just describe a broad outcome and allow the the the model to search with all of the data, you know, that you could ever think and then it puts those things together that would have >> taken sounds like a treasure map to uh Yeah. So yeah, therefore scientific revolution inbound. That's right. >> Yeah. And and and and you're also articulating like um Deis Hasselbliss, the guy won a Nobel Prize for AlphaFold. >> I mean his passion in life is he wants to create a silicon version of the cell, right? A completely simulation of the cell, right? And that's alpha folds on the way there. So that's the kind of stuff science is going to go wild, right? >> All right. Um I worry that at the same time we're going to have a fragmentation of consensus reality like what >> you know 40% of Republicans believe in QAnon and that you know there's whatever I mean there's some crazy out there that people believe >> and you know I think that's increasing and so we're I think we're fragmenting consensus reality but we're increasing scientific knowledge. That's going to be a weird tension, right? >> Yeah. You're going to have more people who believe in stuff that would never have the power, would never have the science that they have in their hands, like their cell phone, but those same people are going to have more influence. So, it's going to be weird. >> That is weird. This is a perfect time to ask our final question. >> This has been great. By the way, I could do another three hours of this. This is you're I I said in the beginning, I said John's a great talker. He's a great guest and and the proof is now uh recorded for an hour. So the the question is the same for every guest no matter what the topic even if it's a non-financial topic identical question. >> Sure. >> If you've already said the answer feel free to repeat it. The question is simply for our listener you know to can you provide him with a single takeaway a single thought today. What would that thought be if you could do that? >> Absolutely. Ask the robot. >> Okay. >> Ask the robot. All right. No matter what you're doing, you're planning a meal, you're going shopping, you want to advise your kid, you feel sad, you want to study nuclear physics, ask the robot. >> All right. Uh, that is certainly one of the most concise answers to that question we've ever gotten. And we do it all the time at I we're constantly, you know, having conversations with our phone and my wife is taking pictures and saying, "How do I fix this?" and all kinds of stuff. So, it's it's we're there. We're there. >> Dan, can I say one last thing? >> If we get the if we get the investment in science back to where it used to be, >> right? And Doge did tremendous damage. Uh why he did that stuff, I don't know. But he he cut off he he cut stuff that was similar to the stuff that he built his business on. I just don't get that. Right. Anyway, so if we can get back to the the investment in science, if we can get back to the brain drain, which I think is critical to the United States success, you look at, you know, Musk, Theel, um, Sachs, uh, U Sergey Bren, um, the guy who started Selectron, these are all immigrants who came here to the United States and got capital. If we keep that going, right? I want to keep importing geniuses from the world and giving them capital and letting them create jobs, right? >> Got to get that right. Got to get science, right? If we do that and then if we think about what's the public library, what's the public education for AI, okay, that will and the Chinese, by the way, are doing this intensively. They're teaching their whole population how to use AI, which is really scary for us, I think, because there's no question that, and I can show you good data that shows an individual with AI can outperform a team. A team with AI outperforms everybody. Okay? >> It's a big study at Proctor and Gamble showed this definitively. Anyway, there's that. If we get that right, we are going to have uh a level of innovation, entrepreneurship we haven't seen since at least the industrial revolution. Why is that the case? You have all this expertise. Capital's more efficient. Uh you know you you can an individual can do the work of many and this a team can do the work of hundreds or thousands. I mean and there's so many unmet needs in the world. I we are just going to see an explosion of entrepreneurship. But we have to have enough people who understand it so we don't end up with people wanting wealth transfer instead of wealth creation. Right? But we have to the the people with means and the people in government need to understand that this is this is a GI Bill moment, right? We have to think about how we insill the population, how we continue to suck the brains out of the entire world and have them come here and do great stuff and and keep capital formation going, you know, and enforce any trust, right? Have some competition. Uh those to me, we we will it is an American century if if we do those things. If we push out, if we screw up bankruptcy with things like student loans, if we monopol, if we let a few monopolists control the core of AI, if we if we don't let the geniuses into this country, if you know, and we don't promote uh enough knowledge of AI so that people don't fight it, but they use it, then we're going to have another whole problem. So, I I think we're at a I think we're at a a fork in the road where it could be greatness or it could be ugliness. >> All right. Um I'm glad we I'm glad we got got an extra thought out of you. So, thanks for that and thanks for being here, John. It was great to hear from you again. >> It's great to be with you, Dan. Corey, take care, man. And just let me know if you need anything else. I love working with you guys. >> You bet. You'll be hearing from us. >> I told you he was a good talker, didn't I? Right. I [laughter] mean, um, sure is. >> John's a great guest. You can just >> Awesome. >> Yeah, we start him up and let him go and he and he tells us everything we'd ever want to know and and then some. It's great. >> I I've taken I took so many notes during that. I'm going to have to download maybe I I might have to listen to this myself again when it comes out just to uh get up to speed on everything that he said. I mean it's just it's it's all fascinating and he but he brings some practical um takeaways from it. I I loved his like the four things for for you know what what would he tells you know parents when they say what should my kid do? >> Um you know the trades, math, >> marketing, selling essentially and then just kind of like ingenuity you know >> experimenting with these things. Um but so much there uh that was awesome. >> It was it was and we will definitely have him back again at some point. And I just want everyone to know, you know, I'm not going to debate people on things like antitrust and government involvement in everything. People know where I stand on this, but it's really not a part of the show. So, anybody who knows me and didn't hear me push back on some of these things, that's just not a part of this show. Um, but hearing from people like John who know a lot about what people are doing with like the most transformational technology yet, um, is really, really important to us. And that was great. It was, it's another great interview and another episode of the Stanberry Investor Hour. I hope you enjoyed it every bit as much as we really truly did. Opinions expressed on this program [music] are solely those of the contributor and do not necessarily reflect the opinions of Stanbury Research, its parent company or affiliates.
AI Is the 'Special Forces' of Investing
Summary
Transcript
Hello and welcome to the Stansbury Investor Hour. I'm Dan Ferris. I'm the editor of Extreme Value and the Ferris Report, both published by Stanberry Research. And I'm Cory McGlaclin, editor of the Stanberry Daily Digest. Today we talk with Dr. John's Viola of G AI Insights. John is an AI guru and get a get a pen and a piece of paper because you will want to take notes. He's just that kind of a guy. He's a great talker. We're going to throw a few questions at him. Start him up and he's got a lot to say. So get ready. We're going to talk with John. Let's do that. Let's do it right now. John, welcome back to the show. It's good to see you again. >> Dan, great to be here. >> So, Corey and I are going to pepper you with lots of good questions. Um, but I wanted to start by um telling our listener, if you could just tell our listeners sort of what you do because most of our guests, they are not like you. you know, most of our guests are are, you know, hedge fund managers and traders and analysts and, you know, equity analysts, all those people. Um, you're a little different. What do you do, John? >> Yeah. Well, um, really kind of a lifelong passion of mine is to try to understand uh what how do machines think, how do people think, and what does that mean for business, society, and individual productivity? Um, and so uh I started that as a professor at Harvard Business School many years ago. And uh what I do right now is run a company called GI Insights. And we believe that every firm is going to have to become a hybrid firm, a hybridization of human workers and digital workers. And our purpose in life is to make people aware of that and to to manage that transition in a productive way. So we do research, training, strategy, and community to help people accelerate that. The training sounds fascinating to me. Maybe we'll get into that. But um Sure. >> I wanted to revisit something that we discussed last time. And you you had told us that folks who work with words, images, numbers, and sounds, the winds wins winds framework >> would be most impacted by AI. >> Yes. >> And then you you kind of put some numbers on it for us as investors. You said we did some research with Valen's research. We looked at which companies are wins intensive. Lots of words and images, numbers and sounds. >> Yes. >> And you said we think 50% of the market cap and 50% of the profit of the entire publicly traded market is up for grabs with generative AI and AI. >> Yes. >> You think you think we're still at 50 or think we're more? >> Uh maybe more. Um because I think there's a knock-on effect of uh lots of market cap getting created to build infrastructure to support AI. So it's a secondary effect. Uh the primary effect is clear. I mean you look at what happened to Cheg's stock price. You look at what happened to Gartner's stock price. You look at what happened at Accenture stock price. Gartner's down 50%. Accentur's down 35%. Because the market is not yet convinced that they know how to be a hybrid organization. >> How do do you think they can do you think you know Accenture we mentioned last time or whichever one you want to take for an example or three or whatever. Do do you think they can? Do you think they're on their way? >> I think Accenture can. Uh Gartner, I'm not so sure. And I had to be a little careful because we compete with Gartner in some areas. Um so I'm not bashing a competitor here, but um the the reason is if you look at if you dig into the DEF 14A and you understand the comp for the board and the senior executives, >> um and I'm I'm on a public board. I've been on a number of public boards. So I and I've been on comp committees and so forth. their um their short-term and long-term compensation is based on two numbers. EBITDA growth >> and uh the growth of long-term contracts. Okay? And the problem with that is that if you're a Gartner, you're going to have to go from 24,000 people probably down to 5,000 people and you're going to have to build a bunch of assets that uh unless you grow phenomenally, right? Okay. So, whatever. I'm talking about the same revenue, >> right? You're going to have yeah you're going to have to build assets that are going to the in the early going could have a return on investment within a given year but it's going to take two three four years for these assets data assets process assets AI assets to pay off. So if you're incenting people on eB that penalizes that investment in those assets and so you're going to be disincented to actually build the assets that are going to be made make you more productive. So I'm unless I see a change in the senior management compensation at a Gartner, I don't think they're going to get there fast enough. Accenture I think understands it because Accenture has an asset measure in the very top of the house in terms of return on assets. Um, if I were running Gartner, I would I would uh give it IBDAR and the the the ad back in on Ebad would be uh research and development and the creation of software and data assets and I would advertise it over a 4-year period. Uh because that that's the operational reality of building um building cognitive assets. >> I see. poor incentives matter in in short >> especially at the top of the house right I mean we're talking you know for for your viewers you know the dev 14a lays out um my comp I'm on the board of infuse systems it lays out my comp everybody else's comp when I was a named officer at diamond same thing even though I was inside the firm so uh yeah incentives drive behavior funny how that works >> yeah executive I mean executive incentives aside which is really important a lot of people overlook in in the investing world. Our the last time we had you on it was November 2024. You said at the time kind of like CEOs, executives didn't really have any hands-on experience with AI tools for the most part. >> Yes. >> Yes. >> Has that changed at all or is that still the case? >> Yeah, it has improved. It's changed. But we're seeing u is a bifurcated market. So there's leaders and there's lagards and the leaders are moving ahead faster. So anytime you see an average number um like uh I MIT had this study which if you look at it was the the methodology is really terrible uh but let's just you know take their thing 95% of generative AI stuff doesn't yield value by the way they're an outlier nobody else is reporting that Wharton's reporting like you know over 70% and open AI anyway >> but let's just say that for an example the problem is that you don't want to look at the average because the le there's a learning effect here and one of the big differences people talk about uh AI and generative AI being a foundational technology um you know general purpose technology I think that's true but it's really not a technology it's a capability and the distinction I make there is a technology can be bought implemented and you get the yield so if I buy a faster welding machine with robots you know I can do the ROI I can get this many more you know autoframes throughout the whole routine when I'm talking about AI or generative AI I'm talking about how people think, how the machine thinks, and I'm evolving that over time. It's much more organic. So, I'm growing a capability and I'm buying a technology. And we still see the minority of companies buying a growing the proper capability. The problem with this, Corey, is that the market is going to mark your com if you're public or if you're a private transaction, you have public comparables. The market is going to discount your company. And we we're seeing this. It's going to discount your company if they're convinced that you have the old model, not the hybrid model. That's what's happening to Gartner. That's what's happening to CHEG. And so Gartner is still a very healthy company in terms of revenues coming in and profit going to the bottom line. But his valuation has been hugely discounted because the market believes they don't have the right model anymore. >> Yeah, that is the uh the stock market looking ahead effect that you've heard so much tell about all my life anyway. Uh and we'll I guess it remains to be seen. You know, >> the other thing, Dan, we're seeing it in private equity because private equity, uh, we're serving a number of private equity firms, and private equity really looks at generative AI and AI, I think, as uh, both something to help with growth, but also to radically decrease labor. >> And, you know, when you create a when you create an innovation, there's only three places the the the value can go. It can go to the customer in a lower price, like with Craigslist, right? Most of the stuff's free. uh it can go to the labor and uh it can go to the investor in a higher return or or it can go to the the labor in either training or higher comp. Uh private equity of course gives it almost all to the investor and the customer and very little to the labor um except at the top of the house, right? Traditionally and u so what we're seeing is that there's a decoupling between hiring labor and and revenue growth. And so that's the premium. So you look at somebody like Curser um you know the new AI startup focused on um uh software creation. They are at $5 million per employee in terms of revenue. >> You know OpenAI is over three million. So we're going to see massive revenue growth separate from labor growth. >> Yeah. I'm glad you brought up labor because that of course is an enormous part of the narrative from all kinds of people. >> Yes. >> From from the you know like the uh what was his name? Hinton the godfather of AI that from from everybody they're saying you know it's going to be apocalyptic. It'll put millions of people out of work. Um historically of course that has been predicted again and again and generally transformative new technologies have created whole new industries multiple new industries sometimes. Um what where are where are you on this? >> Well uh first of all we have to recognize that um I think through two things um u the lack of uh enforcement of antitrust which I think has been a bad thing for competition. I'm a real capitalist. I think you need to enforce antitrust >> and the like the Google thing, you know, that they let him keep the browser I think is nuts. Okay. Uh uh and so u I think there's that. The second thing is if you look at the percentage of uh profit that's going to labor, it's decreased by about 50% over the past 20 years. >> So labor is remployed, but they're getting less of the value. So I think it's really important when people say, "Hey, there's new jobs." jobs. Yeah, but those new jobs don't pay as well, right? And that's absolutely true. And the middle class has been going, you know, u has been going down in terms of standard of living and wealth for the past 40 years. So I think that's an important undercurrent to to remind folks. >> The the second thing is that we've I think we've had two great uh labor transformations in this country. One was the industrial revolution and the second was the computer revolution. Okay, in the industrial revolution, we don't teach this anymore, but it was incredibly violent, right? I live here in Chicago and riots down here, you know, the, you know, 12 year olds working in factories, you know, no benefits, nothing. Uh, 60-hour days, uh, you know, right before there was that and that was bloody people killing each other, the president sending the troops, I mean, the whole routine, right? >> Okay. So, we forget about that and we don't teach it anymore. The second one was the computer revolution, right? Because you think about like insurance companies, for example, literally employed thousands of accountants and and and bookkeepers and all that other stuff, all gone. >> Mhm. >> But at the in in the World War II period, not only were we spending 12% of our uh federal budget on new innovations like rockets and the internet and stuff like that. So you had all that investment which is now down to below 6%. So less than half, right, in terms of forward-looking stuff. The other thing is we have massive investments in labor liquidity. What happened was they were worried as to uh over 10 million guys were coming back from Europe. They were afraid of political instability in the United States because they'd seen all this stuff. They've seen communists and all routine, right? >> Mhm. >> Okay. So we had the GI bill, right, which made for education. It also made for housing. uh the relative cost of upskilling yourself was trivial compared to today, right? You could do it for less than one year savings from a job. Now you you basically have to save for 20 years to be able to afford a college education. >> Uh you didn't you didn't have the massive health care risk. You didn't have both spouses working. So labor liquidity was very high. Hey, I want to go move out to Arizona to get a job. I used to live in the Bronx. That's a lot harder now because I have to get two if I'm going to have healthcare, I have to have two people working like the whole routine. So labor liquidity is way it's more like labor liquidity was in the industrial revolution, not now. There's this great quote from Levit, the guy who did Levittown, the the cheap houses outside New York. Oh yeah. >> He said near about Yeah. They were worried about you know this this issue of communism political unrest. And Levit's quote went something like this isn't exactly right. They said, "Hey, look. If a guy has a house, has a mortgage payment, a car payment, three kids, a dog, and a wife, he's going to be too goddamn busy to be a communist." Okay. [laughter] You know, there's a lot of truth in that. You know, there is. >> So, my worry is that we don't have those same. If anything, we're disinvesting in all those labor liquidity things because I again, I'm a capitalist. I believe I want labor to be liquid, right? I don't want to be able to move around. And u all the variables are uh are much much worse now. They're much more like the first revolution, not the second one. >> And we have a a a openly socialist mayor of the biggest city in the country. >> Yeah. I mean, if you if you if we really talk about socialism, he's not a socialist. Okay. He's not going to try to grab hold of the we have a tremendous amount of socialism, right? Like we have uh the military-industrial complex is largely socialized. It's a cost plus business, very little bidding, you know, we have that, right? We have uh the government buys most of the healthcare, right? And so we already have uh uh we 12 billion is going to the the soybean farmers. I mean uh you know farming is socialist in this country, water is socialist in this country. um you know uh roads are socialists in this country. Those are the real socialist things. He's not going to touch any of that stuff, you know, and he's not gonna he's not also not going to have stateowned uh means of production. So I don't know why he calls himself a socialist. If you know anything about socialism, he's not one. >> Okay. So to him, socialism looks like um you know more free stuff is you know he wants well stateowned grocery store or city own city >> own grocery store. >> You think that's going to make a dent? Come on. Stateowned grocery stores compared to Target, Walmart and Kroger, forget it. >> Yeah. >> You know, I just >> Oh, I don't think it's going to succeed. I'm just >> No, no. But I'm but he's not saying a real socialist would say, "Hey, look, we need to take over Google, right? Because Google is absolutely an essential facility. It's got no governance except I mean that's what a socialist would do. >> I mean what did the socialist do in um in England? They took over the uh uh the coal mining, right? I mean this people they threw on socialism and communism. They don't even know what the heck they're talking about. >> Well, it's a Yeah, I'm aware that it's a sales pitch, but so is conservativism and progressivism and the others. They're all a sales pitch to gain control and get power and stuff. So to me they're all >> yeah I think going back to like the u why why the messaging is is appealing though you know say to New Yorkers that's the point kind of what you're saying John right about like >> the just the situation of people with labor and all people trying to keep up and and afford things afford to live basically >> there's no question that the majority of labor is getting less of the economic value out of our society forget the wealth effect that to the other thing I think people totally misread one of the core things which is that it's not wealth differential it's risk differential so the average worker today has longevity risk they don't have enough money they have healthc care risk which was never look uh one of my kids has MS and the cost of his MS medicine is 230,000 bucks a a year okay So yeah, this personalized medicine stuff and everything, it's expensive. And so there's that. There's uh education risk. I mean, kids, the fact that kids that they have that I I I I really believe in bankruptcy because bankruptcy is a critical part of capitalism. The fact that student loans survive bankruptcy, I think, forget it. I don't Look, if you're a bank and you give some 20-year-old a quarter million bucks, it's your freaking problem. It's not my problem and it's not his problem. That's ridiculous. So I I mean we have car I think of that as socialism. I mean you know what the heck? Why are we supporting >> Huh? >> Yeah they're they they don't have the bank doesn't have any risk. The loans guaranteed. >> It's ridiculous. I mean that let's go back to capitalism. You loan somebody something you go at risk. That's how you make the money >> right? Anyway, so u you know, so those risks, longevity risk, health risk, >> u education risk, those are the things that people wake up at night about, >> right? I know I do and I and I and I've got a great situation. >> Absolutely. >> Yeah. >> Yeah. I mean, I did a age calculator, you know, I'm going to be live to 90 according to the age calculator. Okay. And I probably got like a one in three chance of having a, you know, uh mental disease, right? to have some kind of dementia or some like that. Like, it's expensive. You know, in the old days, I'd be smoking pots, I'd be drinking booze, and I'd die at a reasonable 75. You know, [snorts] >> I I'll I'll send you a case of cigarettes, John. >> Well, actually, what the >> know when the dementia kicks in that we laugh, but the the French as they put in their anti-smoking campaigns, >> the smoking consumption went like this. Body mass index went like this. I talked to some folks at a a large industrial company I don't want to disclose. What a lot of people don't know is smoking sessation actually increased their health care cost because people instead of dying of lung cancer fast they die of diabetes slow. Okay. So smoking sessation is actually bad for health care costs because people still put stuff in their mouth but the m the thing they put in their mouth is food. I've never heard this, but it makes perfect sense, doesn't it? >> It's crazy. >> And and I have to tell you, my [clears throat] my wife decided just to stop drinking. She got some values on a liver test a couple years ago, and she said, "I don't like that." And she just stopped on a dime because it wasn't like the most important thing in the world to her. And her son did the same thing a year ago, and they both report exactly what you're saying. They're they're both like all practically candyholics at this point. I mean, they look great, actually. They're they're both in better shape than ever. >> Sure. >> But but they've they've definitely substituted the candy, you know, and I could see in their case, they're cool, but I could see it getting way out of hand with a lot of people. >> So, it makes sense. >> No, it's it's not anyway. So, I think those the risk differential is at least as important to me as the wealth differential. >> Right. So, we we start we got into this by talking about labor. >> Yeah. and and you know like like Musk is Elon Musk you know his famous quote is like in however many years nobody will have a job and you know everybody will get everything they need or something like that >> but but you know that's that's a very that's that's a huge risk right to to say the end of scarcity or you know extinction I don't know if you saw that chart in the Financial Times actually the Dallas Fed put out the chart and and it was like extinction on one risk and and then you the end of scarcity meaning we're all just fine on the other end I mean this uh >> a wide range is a lot it means high risk right yeah if you treasury bills have a small range of outcomes right >> mining stocks have a big range of outcomes they're much riskier less you get it >> and and and AI from that perspective looks extremely risky to me >> and labor is at the that's in the crosshair anyway >> yeah I think look I'm I'm a real believer in absolute wealth, not relative wealth. Okay? >> And so, uh, I think a lot of wealthy people would rather have relative wealth than absolute wealth. And what I mean by that is that if you take a small amount of the profits that are now going to capital, and you reinvest those in forward-looking stuff like we used to, like the like pretty much every major invention that we that made great companies were started by the government or by academic institutions. So the internet, GPS, all the drugs we use, modern farming, like all of it, right? Began with research that was done by the government or by academic institutions or what I call gift culture institutions. Like all of it, all this stuff. Look, Elon Musk, you and I have invested in Elon Musk at least twice, probably three times. Uh first we invested in all the technology he used to build his cars, right? of the the battery technology and all that stuff. Second, we gave him a direct loan when he was about to go bankrupt. Third, the NASA contracts kept Tesla and SpaceX afloat. So that's our tax dollars. >> So he's a socialist leader, right? That happens to have I mean socialism has saved him, right? Um and so the the the fundamental science is uh almost universally um built by the government. Okay, we're disinvesting on those and on a and and the reason Okay, so that's one thing. The second thing is take a little bit and reinvest in the labor the way we did with the GI bill. That makes my capital more productive because those people are building stuff that's productive and we're winning on a global basis. So they're getting economic rents from the rest of the world, right? And then I have places to put my capital because those people are now consumers, right? This current approach is weakening both of those, which means that my capital in the future will be chasing returns more and more, right? The question is how do I chase those returns? Well, if I don't have growing productivity, the way I chase returns is to seek scarcity. So I get hold of water or I get hold of oil or stuff like that, right? If productivity is not happening. And so I want the world to be like Henry Ford. Henry Ford doubled the living wage of his people from two and a half bucks to five bucks for two reasons. He wanted to he wanted to skim the market on better talent and he wanted to have a consumer there. So my capital is more productive if I have a healthy middle class that's consuming and being more productive. So I'm going to make more. That's why absolute wealth. Now, if I want relative wealth, I can go back to 1066 and sit in the Tower of London freezing my butt off in a dead animal, but I have I have, you know, three dead deer in the closet that nobody else has. So, I'm I'm the wealthiest guy there, right? [laughter] >> I like the absolute thing better. >> I like the absolute thing better, too. >> That would be better. The deer the deer outside aren't aren't very happy with that comment by the way that I got out here. >> Hey, by the way, you know what? What do you think the uh the the uh uh main course was at the first Thanksgiving? >> And it's not turkey. >> Oh, not turkey. Oh, what was it? I don't know. >> Was deer venison? >> It was deer. Okay. >> Yeah, >> makes sense. >> Yeah. But I think that So, back to your question. Yes. If if we do not invest in um new science um what Ben Franklin did with the um the public library there is no public library for AI right now. you need one, right? And more uh funding of fundamental research that open sources, data, weights, models, availability, there's some, there's not enough in my opinion. The Chinese are actually doing much, much more of it. And this is really scary on a global basis because the Europeans now with our new aggressive antagonistic stance with our allies, our traditional allies, >> what's happening is they are now building Chinese AI into their product. So dollar Benz is using the Quen models right and this is I think a big mistake because for the past 60 80 years we've exercised a tremendous amount of security and economic value by uh having our partners like build in our telecommunications I don't know if you remember they pushed back against the uh Huawei switches for example right when the Chinese were trying to penetrate the European we put pressure Don't buy Huawei, buy American stuff. Right? We've done it with our with our software. We've done it with our computer chips. Right? What we're doing now is we're saying, "Okay, that next generation, go ahead and build the Chinese models and we don't care." I think that's a strategic issue both from a productivity standpoint and from a national security standpoint. And we are actually making it super easy for the Chinese to infiltrate our traditional allies. >> Okay. I need you to flesh this out for me a little better. I >> sure say Sam Dmer Benz. Okay. >> Right. >> If you look at now in in AI, there's there's two big uh uh u you know, leaderboards. One is the proprietary model. You have to pay for it. Open AAI, that stuff, right? >> And then there's the open model, you know, Deep Seek, Quen, Z, so forth. >> Okay. If you look at the leaderboards, you take the top 10, there's only one Chinese company that's in the top 10 of the paid models. There's only one American company in the top 10 of the open models. >> Okay. So there that now there open's a little tricky. There's three parts to open. There's data, there's models, and there's the the software. There's data, there's weights, and there's models. Okay. The Chinese are not opening source open sourcing the data but they are doing the weights in the models. So I can use if I'm dollar benz and I want to I want some intelligence in my dashboard and I want to run a 70 billion parameter model I can use the Chinese models dropped right in there I don't have to pay IP back to the Chinese. Okay. If I do that with all but a couple of the American models I do have to pay IP back to the Americans. In addition, when the president is saying things like NATO's gone and all that other stuff, of course, I'm going to go to the Chinese. Who else am I going to go to? I mean, I've got Mistl and I've got, you know, a deaf alpha out of Germany. I've only got two models and I got Falcon out of the UAE. I've only got two or three options to do anything like a good model. I got the Chinese who are going open and the Americans who are going closed. >> I don't think that's great strategically for us. It's ironic too to hear that, right? >> Yeah. US closed, China open or and >> well except the China they they have they the idea of intellectual property in China is just you know >> right >> it's non-existent basically >> you know so it makes all the sense in the world >> it's the same thing the Americans did like Moody Street and Waltham mass where the American industrial revolution started is named after it's the main street in Waltham Massachusetts where the Charles River drops three feet so that's where they first put it and then they went over to Marramac because it dropped 13 ft anyway for more power. Moody's claim to fame. He was an engineer, American engineer. And when the shuttlecock loom uh was making uh British um you know um British uh weaving way more productive. Uh if you left England with blueprints of the shuttlecock loom, you would be put to death. Okay. >> Whoa. >> If you're caught. Legend has it that Moody memorized it and came back to the United States and put it in there not paying any royalties back to the Brits. So, you know, and the same with the books, you know, like Charles Dickens and all that other stuff, we stole all their stuff and just published it locally, right? So anyway, but the we our problem is that when we let the Chinese in, we let them for market access and I talked to a bunch of folks in in the intelligence agencies. They would go to the big companies and you can imagine who they are, aerospace, industrials and stuff like that. They say, "Hey, the Chinese are stealing all your stuff." And the the the companies wouldn't raise a stink because they still wanted access to Chinese markets. Okay? And I think that was a tremend I think that's where folks like the the the government should have stepped in and said, "No, no, no, no. You can't because guess what? American consumers and American taxpayers built that IP. You can't simply give those decades of intellectual property to them for your incremental market access, >> right?" Which is what they did. So, let me give a specific example. I forget the name of there's a Chinese drone. Looks just like the General Dynamics Predator and at the weapons uh fairs they're selling it for onetenth the cost. Sure looks a lot like it. Looks a lot like it. >> Yeah. Yeah. There's a lot of >> I saw that a lot of other stuff in the west. >> Yeah. I saw that recently with what the when the humanoid robots were having their moment a couple months ago and there was a Chinese one posted online through like a Walmart affiliate site for like >> Sure. >> three days before it got pulled down for >> Yeah. >> No public. >> We have to Oh, sorry, Cory. >> No, go ahead. Yeah. >> Yeah. I I I I think that's true. I think that I think that we have to update the narrative, though. They're not just copying. I mean, you look at some of the stuff they did with the Deep Seek model, for example. They're advancing algorithmic innovation and that's really important because the reason that these that uh chat GPT and Gemini and so forth consume so much power and compute is because the transformer algorithm which is underneath most of these things and and gave us this unbelievable I mean I never anticipated the quality of the interaction you have with these things even back in November 2022 when chat GPG35 hit the scene right >> is that they are computationally inefficient Well, if somebody comes up with a really efficient algorithm because the algorithm, the transform algorithm brought us this step change in in in uh in performance, right? >> Yeah. >> Okay. If somebody comes up with a step change in performance on the compute side and there are some some folks who are using other kinds of models, well, guess what? Then primary demand for all that compute goes down. That's why uh personally I don't I don't uh invest in the hardware stocks because I think algorithmic innovation can radically change um the um the demand for hardware >> and it sort of needs to go down if you do all you know if you do the arithmetic on the power and what's needed to be built and it's sort of we it it kind of can't be done very quickly and and you know if you look at for example you know they're the the turbine makers just you know natural gas turbine makers >> right >> it's all that's the that's the uh the bottleneck right so >> I've heard I've heard it's about a fiveyear weight right now We're we're working with one of the folks who are building was building the largest data center shell in the world and they they have eight >> they have uh five gawatts of turbines lined up and that's a huge part of their market valuation is the fact that they have the contracts and they have the permits. >> Yeah. They just need the machines. They need the turbines. So, so in other words, I'm just saying that it it it makes sense to me what you said that you will you will need you will need that greater efficiency. In other words, like when when deep it makes it made all the sense in the world for Nvidia's market cap to get hit, you know, the day Deep See came out with its big announcement. And there ought to there well ought to be more such announcements in the near future, shouldn't there? I mean this this is you know it's it seems an extremely important part of the equation for someone not to be saying hm you know maybe if we maybe we wouldn't have to build all these power plants if we just had you know better software basically better algorithms. >> Sure. Yes. And there's certainly a lot of people working on that problem. Uh and the Chinese are working on it. I I actually think it's good news that we're starting to sell the Nvidia chips because that'll keep them fat and happy and they won't be so they won't be so desperate to improve the algorithms which I think is a good thing that net >> um for us u I mean I should I should share um two two u points of view I have first of all I I do believe even with all our faults that uh I do believe in America as it in its value systems and its capabilities and and its role in the world and I'm kind of sad we're pulling back as we are. Uh and the second thing is I'm a rail capitalist so I I like competition. Um I'm I'm sad that um you know we've allowed these massive concentrations in so many industries. If you go back to the Sherman Antitrust Act, you know, uh, John Sherman, who was Tecumpsa Sherman's brother, the guy from the, you know, the north who did the whole Sherman's march in Georgia and all that other stuff in in some of the words about that, there were two parts to the antitrust thing. One was consumer harm and the other was companies having power as powerful as governments, right? And Sherman had this great quote. He said, "Why? We just fought a war to not serve a king. Why would we allow a company to control an essential facility where we have to bow down to them just like a king? Okay. Now, the the interpretation of that act has gotten rid of the market power piece and only kept consumer harm largely due to the guys down the street here at University of Chicago. >> Uh but, you know, I think these I think these mega scalers have way too much uh political power. Mhm. >> Um, and I think and I don't think we should have a company owning most of the satellites up there, which we have right now, and not being underneath military control, you know, again, I I believe in America and I believe in competition, so I don't think we should have people with monopolies on essential facilities. >> Yeah. Um, let's talk about um [clears throat] your last last time you were here, you introduced the four levels of generative AI adoption. And I'm and I'm wondering how much that has changed, if at all, if you still use that framework, because I know that since last time we talked, I went from what you would call a toe dipper, if that even to um someone who is trying very very hard to be an intelligence leverager, and to load as much of the best data I can into the best, you know, um system that I can get, you know, that a guy like me can get on his lap. top. Okay, just put it that way. Um to to make the to do the best research I can on on all the public companies that we write about. Um so, you know, I really I I just can sense the power of it. I haven't really, you know, I've learned all this. I've learned how to collect all the data I want and put it all in one place. Now, I just need to learn how to prompt it and analyze it and and get what I want out of it. Um, so I it would seem to me you're probably in the business of teaching people how to be intelligence leveragers uh to to a great degree. Yeah. But I'm wondering something else though, John, just before you tell me talk about intelligence leveraging and what you guys do. Um, what I'm wondering is, um, since you last spoke with us a little over a year ago, um, have you heard the same experience that you just heard from me from lots of people? >> Yeah, absolutely. and and the a key differentiator is how for the firms because think of it again as these two populations and and we're seeing significant return on assets differentials in the intelligence leveragger kind of folks >> like you know ROAs of you know 21 22 23% growth much faster things like that um the the capability level is the key indicator of if you are going to get value from AI and >> capability level >> capability level and so to your question yes we absolutely are not only using that framework we updated we have a little bit slightly different language to make it easier to remember we call it the rise framework so there's research and experiment and education >> then there's islands of innovation >> then there's scaling and synchronization >> and then there's uh uh emergent intelligence that's the intelligence leveragers okay It's a rise and we have we have yet to find a company that can jump straight to level three to go system level stuff. we see more what happened with you Dan which is you experimented you probably did some little innovations right now you're trying to think okay how can I really reinvent the way I do my work I don't want to put words [clears throat] >> but >> every day that's all I'm concerned with now is that I'm I'm re I'm refiguring out how to do my job all over again >> right but if you had I would say and feel free to disagree if you had started that at the very beginning before you had those other experiences you wouldn't have known what you were talking Yeah, that's been my >> Sure. Yeah. >> Yeah, that's been my experience, too. You know, I I originally, you know, I was playing around with all these tools a year or two, you know, a year and a half ago, and there's run run into a certain level of friction, right? And and now, you know, you you learn the pro, you learn what this model is, you learn the prompting, the importance of the prompting. You know, I'm talking about for writing and editing purposes and research. And then now it's to me it's a matter of work somehow working it into like my daily workflow is seems to be the the biggest challenge >> like what's worth spending time on versus what's not. >> Um that's like the biggest thing for me and I als you know I also you know struggle to make decisions sometimes. So that doesn't help either, you know. >> Um so that's I I don't know that is that that's a is that a common uh thing that you're trying to address as well? >> Yes, absolutely. and and the the kind of dirty little secret, you know, of of of executive education and, you know, I'm I'm a fellow at uh Harvard Business School and I taught there for over a dozen years and so forth, is that if you go to an executive education class at Harvard Business School or any any good uh you know, academic institution, a core skill when you're teaching executives is to teach them things that they should have known anyway, but they can't admit they don't know in an unemb embarrassing way. Okay. [laughter] >> All right. >> And so that's where AI is in most companies. People like it's the because the windows closed to say AI, what the heck is that? Could you like slow this down and tell me what you're talking about? >> Right. >> And that's how most executives really think, but they can't say it because it's not it's not socially acceptable. AI, what's that? >> I mean, if you go back to the basics, we can't really define artificial and we really can't define intelligence. you put those things together, you know. Um, so that's why that's why this model is so important. You have to have hands-on experience. There's a great uh there's a great quote from Frank Zappa when he was somebody's asking about a a music critic and he said, "Writing about music is like dancing about architecture, right?" >> Oh, dancing about architecture. Yeah. >> Yeah. And so if you don't experience what it's like to talk to a silicon intelligence who can literally talk to you in any language at any level of specificity and now with some of the new models, especially Gemini 3, you can say, "Give me an illustration. Give me a cartoon. Give me a what? Give me a movie." Right? If you haven't experienced talking with this silicon hive mind that has read literally everything that's been written, you don't know what that's like, right? And and now we have all these low code no code things. Anthropic just came out with a open no anthropic just came out with an analysis that said inside their company almost a quarter 23% of the low code no code stuff they're doing would never have been done because it was either too low down the priority for it or the person who understood the problem thought it was too much of a hassle to learn the technology. So back to what you're saying Corey people are using this stuff just you say okay should I automate that do I do it enough and there and I think the ripple effect of that productivity at the individual level is going to be fantastic but then it's like special forces like we used to compete with regular army and special forces we crossrain we upskill we use we take traditional technology we modify it to make it better and just like with special forces you know regular army might have taken a 100 people to to do something special forces might be able to do it with six, you know, and that's what's happening in organizations and and to the organizations that are upskilling. So, back to your question, Dan is absolutely central. Senior executives resonate with it. The other thing is it helps it helps sort what the vendors are trying to sell you. So, there'll be one of the big four went to a and I don't want to use names, right? But one of the big four went to a a consumer brand that we all know, right? And they came in and said, 'Look, give us 5 million bucks and we'll go find 25 million bucks at of system level improvement. Complete and utter failure. Why? They tried to jump straight to level three. They tried to jump straight to system level change. So when they're talking to people, they don't I mean, we're having this AI kabuki where we're saying stuff, but we really don't understand each other. And then they have no hands-on experience. They haven't done islands of innovation. So they don't know what it's good for and what it's bad for with any even if they're not building it. You know, you have to have you have to you have to drive a car to understand what's valuable in a car even if you don't know how to build the car, right? And so yeah, that and so they jump straight to system level stuff, burn through the five million bucks, total abysmal failure. And we see that happen again and again. >> John, I wonder um I I am not a uh company executive, so I'm not afraid to say I don't know. [laughter] So I don't know what low code no code means. >> Okay, that that's basically u think about programming in English. Hey, I would like to you know I would like you to go and look at this uh set of spreadsheets and then look at this set of objectives and then generate a six-page PowerPoint thing every Tuesday at 10 a.m. in the morning and to say it like that not to have to go into the language of all the different software that would generate that. Ah, >> so low low code no code sounds like people know who know how to prompt AI. Well, >> well, yes, but often prompt. Yes, absolutely. In the with a capital P prompt. So, if I'm prompting to build an artifact which then will do stuff for me as opposed to prompting in a dialogue to get an answer. >> Ah, okay. I see. So, um could I see. So not even so the PowerPoint is a good example, but it sounds like um low code, no code creation of of uh you know modules of code that then do things too. >> Exactly. >> Okay, I see what we're talking about. >> Look, we've got >> that's that's that's really cool. And that's why I guess Zuckerberg took out all his mid-level managers, right? This isn't what we're talking about. Loco, no. Okay. >> Exactly. And so 10 million people speak Java, 15 million people speak Python, everybody speaks their own language. >> Yeah. >> So that's the thing that we've just taken, you know, everybody can be a programmer, >> right? >> Yeah. So the winds crowd um may may have a bit of an advantage here if they're really good with words and numbers. >> Absolutely. The >> It's not hopeless. >> No, no, no. But back to employment. We've already seen a tail off of employment of new kids >> coming in. We see a tail off of software programmers who have less than three years experience. >> People ask me what their kids should do. I say two things. Well, sorry, three things. No, sorry, four things. U the the I'll stop at four. No, the first of all, if you want to learn a trade, that's a good thing because we're going to be short about a million folks at least of plumbers, carpenters, all that other stuff. As well, if wealth differential continues, you look at the consumption of the top, you know, 1 to 10%. We got a lot of Maria Antonet going on, right? People building houses they don't need, you know, and they're super specific. They're going to absorb all kinds of trades, right? So, you know, that electrician that, you know, used to be building, you know, houses for the middle class and making whatever 20 grand a house is now building, you know, some 6,000 foot mansion in Tucson that they get used twice a year, but the electrical bill is 100 grand, right? >> And so, and we're seeing that happen. So, the whole Marine Antinet effect, right, that that's going to create primary demand. There's also just a raw need. So, you want to go in the trades, go for it. And and by the way, the trades, this thing is going to help the trades phenomenally. 26 is going to be the year of of uh mobile intelligence. And I don't know if you've done this, but just take this thing and if you have any question, you know, you're uh like I was looking at my water heater the other night and I was trying to figure out if I had a low water shut off. I just hold this freaking thing up and I take a picture of it and it'll tell me if it's got a low water shut off. How do I repair it? Where can I get somebody? The whole routine. Okay, so just imagine what that does for the tray is what's this part? How do I do this? How do I do a hip roof? Right? >> This will tell you today retail with no right, which is unbelievable. >> So you have that. So trades are going to go like crazy. Second thing, mathematics, especially matrix math. Math is the lingua frana is the universal language under all of this. So if you understand mathematics, fantastic. Third thing, sell. Anybody can sell stuff. as the as as as as we go from army to special teams and you get smaller and smaller groups, the premium and the value of people who can do demand generation is going to go up, not down, >> right? Because you still have to sell stuff to people to get them to, you know, commit. And and the the last thing and perhaps most importantly is do what you're doing, Dan, and what you're doing, Corey, which is show me your posy, right? So, I don't think any company today should hire somebody who hasn't built a GPT, a gem, low code, no code. Show me how you're using AI to automate your life. And I don't care if that's in work, your passion, family stuff. I don't care what it is. Show me. And I only want to hire those people who are doing that because individuals are not how if I've got no experience doing software coding and I want to get to be and I want to compete with the three-year folks which is what the numbers show according to signal fire venture capital company you know the massive fall off on if you have less than three years experience how can I simulate three years of experience I can come in with my posi oh let me show you my code review thing let me show you my documentation robot right so you're hi people should be hiring posies of a human and a bunch of robots, not just humans. >> I'm glad that you broached the topic of robots like because when I think of robots, I think of the physical ones in factories. >> Yes. >> And and um that that has some impact in the trades, doesn't it? >> Uh yeah, it's going to increase primary demand for the trades because all those robots are going to need to be serviced. >> Okay. Of course. >> Right. and and I we'll have robots that'll service robots, but there'll be a I it's just like when um the uh I was at a a conference, it's kind of funny. I was at a conference in San Francisco when you know the self-driving thing just started taking off and and people were saying, "Oh, what are we going to do with these three million truck drivers?" And I felt like saying, "What are we going to do about the 15 people I walked past on the street on the way here?" But anyway, the uh [laughter] >> let's solve that problem. Anyway, it turns out actually if you remember what happened when when between when self-driving started which is brought to you by the government defensive research projects agency did the Mojave Desert Challenge right and only after the two number of teams CMU and a couple others successfully did that under government funding that's when the the only after they've done that translational research that's when Bin called a guy named Chris Mson from CMU and started the Google car project which was the first among the self-driving car projects. Right? So all that Whimo and all that other stuff again started like the internet started with government investment. Um the the the what happened instead of 3 million truck drivers being out of work is with the microtargeting of Uber and micro logistics of Uber and my supermarket and everybody else driver demand actually went up in the near term. Now, in the long term, as Whimo comes in and the little Whimos, right, little non-car size stuff, will driver demand go down? Sure. But it's going to take a while. Same thing's going to happen with with robotics. Going to have massive adoption of robots. Huge increase in the need for mechanical engineering and repair over time. Will the robots build the robots? Sure. But it's going to take 20 years, just like it's taking 20 years for the driving thing to happen, right? Mhm. >> So I think we have it's just like the paperless office. I I have a friend who made a boatload of money when Xerox announced a paperless office. He invested in paper companies because as soon as they had as soon as they [laughter] had the the laser printer, the first salvo of the paperless office was an explosion in paper consumption, >> right? >> You know, same thing happened with driving. Same thing's going to happen with robots. >> I see. Do you use Whimo? Do you Whimo in Chicago, John? >> Uh, the answer to the first question is yes. The answer to the question, the second question is no. I wish we did. I love Whimo. I mean, I was so impressed. I mean, when I use Whimo, it's just it's a totally different experience. And >> I've heard that universally from everyone. >> Heard that from every single person. Yeah. >> I was staying in a club on the top of Knobill and so we got in the thing and it went up over Knobill. It was an illegally parked 18-wheeler at the top right in San Francisco. Then there were some people who were jaywalking and then there was a car that was going around the illegally parked thing. It managed all that seamlessly. I was like, "Wow, that was just like [laughter] I I just happened to go that way. My first Whimo ride." It's like I am impressed. >> Wow. Yeah. >> Yeah. >> Yeah. How long before Go ahead, Cy. >> Yeah. How long before Whimos are They're coming to coming to Baltimore, uh Philly and St. Louis now. uh Pittsburgh too I saw. Uh so we'll see which which brings me to like one of the things I wanted to ask you well you already answer I mean you said so many fascinating things already has been great and one of the things I wanted to ask you was like you know what what what for you know what skills for uh somebody trying to to learn all this and you already said that so um I'm also our crowd is also interested in like attractive investments or sectors um as this whole you know as the whole story plays out you already mentioned you weren't really not into the hardware companies. Um are there any areas like that you're uh particularly optimistic about or sectors or or that sort of thing? Um >> sure. >> Yeah. >> Yeah. >> Yeah. Absolutely. Well, first I think um in the in the um AI world, the hyperscalers, the hyperscalers I like are the ones that have massive u productivity enhancement through software creation that improves an already great model. So who is that? That's Amazon, Google, U Meta and um and uh uh uh Microsoft, right? Uh Apple should but we've seen no evidence of their ability to do it. maybe they can pull it off with a you know their their version of the AI iPhone or whatever it's going to be right but um so and then of those you have to remember Google is the only one that's vertically integrated in the whole AI stack right from the customer all the way down to the silicon right so they have their TPUs and I think that's going to be a huge advantage because the the the performance characteristics of delivering instantaneous intelligence across any language in any location is is a massive engineering challenge right just the the latency see the response time for for you not to perceive. For example, 11 Labs has got this new thing out where it can real time listen to the three of three of us real time uh instantaneously transcribe and then rebroadcast in 20 languages with 150 millisecond delay. >> The reason 150 milliseconds is important is because that's the threshold for perceiving a delay in conversation. >> Wow. >> So you have to engineer the living daylights out of that, right? And Google is amazing at engineering. Um, something like Nvidia, I'm not saying they're not worth it. It's just so many things have to go right. And I look at the hyperscalers, you know, these idiots who say stuff like, "Oh, [snorts] it drives me nuts about the the AI bubble." Okay, first of all, let's separate two things. Is there overinvestment overpricing of certain things? Of course, it's a market, you know. On the other side, is there an AI bubble on adoption? Absolutely not. Adoption is accelerating. Okay. So, pricing is different than adoption. >> Sure. Sure. >> Then the second thing is will everybody make money? Of course not. That's why it's called investing. We sit here, you and you and me and you, Coran, you you know, the three of us make three different bets going forward. The market can't support all three at the level we're hoping it'll get to. So, you guys pick the right ones. I pick the wrong one. Of course, I'm going to lose. That's the nature of investing before the fact, right? >> Yeah. And so that's nutty. Um, and last but not least, so, so of course I can sit like Norio Rabini and say, "Oh yeah, stuff's going to go down in the future." Yeah, no joke. You know, the sun's going to come up, too. I mean, give me a break. >> The um, >> yeah, >> the but the most important thing is if you want the clearest use case for productivity with AI, bar none and adoption, >> software software creation. >> 40% of Microsoft's cost base is software. >> Okay, >> if I can take that number down by 12%, 30% of 40%. >> Right? 30% of that is a massive increase in in Microsoft's market cap. They can spend billions. They generate just under $2 billion of EB a week. They can spend billions of dollars on this. And if they never sell any co-pilot to anybody, it's going to be worth it to them. So, you know, people are now open AAI, I I personally wouldn't have the courage to invest in Open AI at the valuations they're talking about because they don't have $2 billion of IBIDA a week coming in and they don't have a a 30% 40% cost base that they can shrink. So, I would say that. So, that's that. The second thing is I would look carefully at these [clears throat] new companies that are um that are doing uh vertical uh rappers and implementation. So you take a company like Harvey in the law. Okay. So Harvey is now adopted by a number of law firms. You know, I think >> uh my oldest boy is an attorney. He's a prosecuting Chicago uh attorney in Chicago. The good news is he's going to still be employed because Chicago's not going to get any less crime written and and so he's he's going to have a job as long as the state doesn't go bankrupt, right? Which it could also do anyway. But but you know, if you're working for some corporate law firm working on deals and stuff like that, they're going to have half, a third, a quarter of the people, right? And certainly if you're working in a because you know a lot of the I mean think about it, the law is essentially a badly designed large language model, right? um kind of you know um and so the you look at somebody like Harvey they not they've not only succeeded in getting good market share but now they're learning how to really tune those models and they're having a whole longitudinal thing of data you know just like um you know Facebook has my whole longitudinal thing of my social network they're doing that in different vertical domains I think those are areas that are very promising and so there's that uh the second thing if you have attackers who are coming in underneath and we like to think of ourselves as one in the research area that if you if we can prove that we can scale you know I think folks like us coming in with the with the Amazon model when they're competing against Kohl's and Macy's right >> in any given industry and those are going to start to appear they usually appear about five to seven years in for the retail customer if you're not in venture So you know for example the the financial disruption usually takes somewhere between four and seven years. So if you look at for example the the the price of New York taxi medallions 5 years after Uber entered the after Uber already entered not only being company but had entered the New York market the price of their taxi medallions was between 1.2 and 1.3 million bucks per medallion. The sixth year it was 35,000 bucks. >> Now they've recovered to about 200 to 300,000. >> They're never going back up to a million two. >> No. >> So financial disruption, you know, my my ex-colague and God rest his soul, you know, Clay Christensen talked about disruption, but he missed financial disruption. He got operational disruption right. Like you're going to buy from a mini mill, not from an integrated steel mill, right? >> That thing. But the financial disruption, that's what's happening to Gartner right now. They're getting financial disruption. Unless they can convince the market that they can get that they can do the new model, they're never going to recover. And so if you invest in those companies that are coming up underneath, the Ubers of the world that have the new model in professional services, in science and discovery, what a lot of people don't understand is we're going to have an explosion of science like we've never had before. Right? the ability of these models. If you think about science, okay, think think about how you improve stuff and I call this stuff practice. Um, practical knowledge. >> So, let's let's let's separate knowledge and I know I've read enough philosophy to know that this is grotesqually oversimplified. So, [snorts] anybody actually knows what they're talking about, I apologize in advance, but let's keep it for for us. >> Two hunks of knowledge. There's what I call practis, which is a provable thing that regardless of what your belief system is, works. So Mike Johnson, you know, is a flat earth guy, right? >> And I mean, he's a young earth guy. He thinks the earth was started 6,000 years ago and stuff like that. The speaker of the house, >> he uses technology that his belief system would never create. Okay? Like he like his belief system would never get you a cell phone, would never get you an antibiotic, right? The whole routine because he doesn't believe in science. So he never he I don't think he should be able to use all that other stuff. But that's another thing. The if you know I think if your belief system would never discover the things you use, you shouldn't be able to use them. But anyway, um so if he wants to go be Amish or something like that, that's okay. Go for it. But um the so what's hap so prais is provable like we can't argue about the fact that you know if you have this bacterium and I put this penicellin on it it's going to be dead at this time we can't argue about the fact if I design a nuclear weapon like this and I do this you know kind of explosion it's going to create this kind of bomb right >> that's a very specific kind of knowledge doesn't mean doesn't matter what you believe it's true >> and we can repeatedly do it >> right >> right Okay, that and then there's all the other stuff like political beliefs and religious beliefs, those are all socially constructed and navigated, right? >> Right. >> You know, and back to Mike Johnson, his phone works even though his belief system doesn't believe it should be there, >> right? >> Because nothing about anyway, you know what I mean? So, so what's what we're gonna what's beautiful about AI and the math underneath AI, the matrix algebra, is that it is incredibly good at creating practice, right? Because I can say here's what you need to look for as an outcome to prove that it's right. Okay, so what's an example of this? something like will this molecule um for example lidocaine lidocaine completely commoditized molecule. It turns out at Google they were looking at um they were creating um a a system to help uh think about new molecules that might be relevant to different disease states >> and they look at them functionally and they look at the trials and they look at all the data they can get to say oh okay here's something we haven't thought of might be useful. Lidocaine turns out is very good for certain kinds of breast cancer. Okay. Now, nobody even thought to look at lidocaine. Okay. And it doesn't reverse it, but it does stop it. Okay. That's only possible because the AI allowed for a robust description and an unbelievable search space where it could go look for stuff that no one no scientist had ever thought of, right? It's just too big a space. So, that's a that's a perfect example of the creation of practice. >> Okay? Because now I've got rational reasons. I'm looking. I've got an objective function. I can say does it help stop cancer and does and then what's the mechanism of action and then like that. Okay. It turns out for example in drugs we believe that we've maybe maybe explored 1% of the useful compounds to make humans more healthy. Maybe 1%. In material science it's less than 1%. >> Okay. I see what you're saying. This is this is um Thomas you know, structure of scientific revolutions, not on steroids, but you know, the Ferrari, the the you know, the the rocket ship, >> acid, iaska, whatever, you know, it's like >> Yeah. >> Yeah. And and and the thing is that all the models speak math. >> Yeah. >> So they all can talk to each other. So you'll find all these random things that you that that would take a long time that nobody would ever think of. Some outlier student that nobody listens to might come up with it. Now you just describe a broad outcome and allow the the the model to search with all of the data, you know, that you could ever think and then it puts those things together that would have >> taken sounds like a treasure map to uh Yeah. So yeah, therefore scientific revolution inbound. That's right. >> Yeah. And and and and you're also articulating like um Deis Hasselbliss, the guy won a Nobel Prize for AlphaFold. >> I mean his passion in life is he wants to create a silicon version of the cell, right? A completely simulation of the cell, right? And that's alpha folds on the way there. So that's the kind of stuff science is going to go wild, right? >> All right. Um I worry that at the same time we're going to have a fragmentation of consensus reality like what >> you know 40% of Republicans believe in QAnon and that you know there's whatever I mean there's some crazy out there that people believe >> and you know I think that's increasing and so we're I think we're fragmenting consensus reality but we're increasing scientific knowledge. That's going to be a weird tension, right? >> Yeah. You're going to have more people who believe in stuff that would never have the power, would never have the science that they have in their hands, like their cell phone, but those same people are going to have more influence. So, it's going to be weird. >> That is weird. This is a perfect time to ask our final question. >> This has been great. By the way, I could do another three hours of this. This is you're I I said in the beginning, I said John's a great talker. He's a great guest and and the proof is now uh recorded for an hour. So the the question is the same for every guest no matter what the topic even if it's a non-financial topic identical question. >> Sure. >> If you've already said the answer feel free to repeat it. The question is simply for our listener you know to can you provide him with a single takeaway a single thought today. What would that thought be if you could do that? >> Absolutely. Ask the robot. >> Okay. >> Ask the robot. All right. No matter what you're doing, you're planning a meal, you're going shopping, you want to advise your kid, you feel sad, you want to study nuclear physics, ask the robot. >> All right. Uh, that is certainly one of the most concise answers to that question we've ever gotten. And we do it all the time at I we're constantly, you know, having conversations with our phone and my wife is taking pictures and saying, "How do I fix this?" and all kinds of stuff. So, it's it's we're there. We're there. >> Dan, can I say one last thing? >> If we get the if we get the investment in science back to where it used to be, >> right? And Doge did tremendous damage. Uh why he did that stuff, I don't know. But he he cut off he he cut stuff that was similar to the stuff that he built his business on. I just don't get that. Right. Anyway, so if we can get back to the the investment in science, if we can get back to the brain drain, which I think is critical to the United States success, you look at, you know, Musk, Theel, um, Sachs, uh, U Sergey Bren, um, the guy who started Selectron, these are all immigrants who came here to the United States and got capital. If we keep that going, right? I want to keep importing geniuses from the world and giving them capital and letting them create jobs, right? >> Got to get that right. Got to get science, right? If we do that and then if we think about what's the public library, what's the public education for AI, okay, that will and the Chinese, by the way, are doing this intensively. They're teaching their whole population how to use AI, which is really scary for us, I think, because there's no question that, and I can show you good data that shows an individual with AI can outperform a team. A team with AI outperforms everybody. Okay? >> It's a big study at Proctor and Gamble showed this definitively. Anyway, there's that. If we get that right, we are going to have uh a level of innovation, entrepreneurship we haven't seen since at least the industrial revolution. Why is that the case? You have all this expertise. Capital's more efficient. Uh you know you you can an individual can do the work of many and this a team can do the work of hundreds or thousands. I mean and there's so many unmet needs in the world. I we are just going to see an explosion of entrepreneurship. But we have to have enough people who understand it so we don't end up with people wanting wealth transfer instead of wealth creation. Right? But we have to the the people with means and the people in government need to understand that this is this is a GI Bill moment, right? We have to think about how we insill the population, how we continue to suck the brains out of the entire world and have them come here and do great stuff and and keep capital formation going, you know, and enforce any trust, right? Have some competition. Uh those to me, we we will it is an American century if if we do those things. If we push out, if we screw up bankruptcy with things like student loans, if we monopol, if we let a few monopolists control the core of AI, if we if we don't let the geniuses into this country, if you know, and we don't promote uh enough knowledge of AI so that people don't fight it, but they use it, then we're going to have another whole problem. So, I I think we're at a I think we're at a a fork in the road where it could be greatness or it could be ugliness. >> All right. Um I'm glad we I'm glad we got got an extra thought out of you. So, thanks for that and thanks for being here, John. It was great to hear from you again. >> It's great to be with you, Dan. Corey, take care, man. And just let me know if you need anything else. I love working with you guys. >> You bet. You'll be hearing from us. >> I told you he was a good talker, didn't I? Right. I [laughter] mean, um, sure is. >> John's a great guest. You can just >> Awesome. >> Yeah, we start him up and let him go and he and he tells us everything we'd ever want to know and and then some. It's great. >> I I've taken I took so many notes during that. I'm going to have to download maybe I I might have to listen to this myself again when it comes out just to uh get up to speed on everything that he said. I mean it's just it's it's all fascinating and he but he brings some practical um takeaways from it. I I loved his like the four things for for you know what what would he tells you know parents when they say what should my kid do? >> Um you know the trades, math, >> marketing, selling essentially and then just kind of like ingenuity you know >> experimenting with these things. Um but so much there uh that was awesome. >> It was it was and we will definitely have him back again at some point. And I just want everyone to know, you know, I'm not going to debate people on things like antitrust and government involvement in everything. People know where I stand on this, but it's really not a part of the show. So, anybody who knows me and didn't hear me push back on some of these things, that's just not a part of this show. Um, but hearing from people like John who know a lot about what people are doing with like the most transformational technology yet, um, is really, really important to us. And that was great. It was, it's another great interview and another episode of the Stanberry Investor Hour. I hope you enjoyed it every bit as much as we really truly did. Opinions expressed on this program [music] are solely those of the contributor and do not necessarily reflect the opinions of Stanbury Research, its parent company or affiliates.