Mark Mills - The Cloud Revolution

27:11You can trace this across every class of technology, including chat GPT right now. I mean, it doesn't matter what the technology is. In the early stages, we have very low efficacy. Whatever the efficacy measure is, whether it's making drugs, an artificial intelligence engine, a radio, even cars, if you just think about airplanes. and all of them is the engineering follows the science and chases the curve down, not just on cost, but in energy efficacy. What happens is people get seduced by that magical trend and think it can happen forever and it's infinite. The energy one, like everything in nature, back to laws of nature, everything has an asymptote or everything reaches limits.

27:57So as you approach the limits, It doesn't mean you can't find new innovation. It means you have to use the proverbial workaround. You have to change their engines. You have to change the modalities. You have to have new inventions. This reminds us of the Google Data Center and Gilder's book. Cole's first cousin put the smooth floor in the Amazon Data Center near Prineville, close to the Dallas Dam. Google was the Dallas Dam, but the other one was the dam up the river from there. It needs energy like the aluminum smelters did. The smelters are gone now. Doesn't this explain that even the data center will find their own death at some point?

28:33So George and I have a side debate on this. I think the answer is no for the reasons of the nature of the architecture of systems. You know, the peach analogy is what is used many times about systems and networks. The networks are where all the action is on the edge because the surface expands faster as you make the balloon bigger or the peach bigger. The core is still needed, and the core has a different – But what happens is the function of the core changes. And this is what's going on already. So edge networks are part of the cloud. So rather than having, and you're absolutely right, there is a natural limit to the scales of all things.

29:07And engineers usually figure it out by going too far. So we did this with airplanes in a sense. The A380 is not a successful airplane, not because it's not a great aircraft. It's an incredible aircraft. It's because it's limited utility. As you make them too big, it takes away the rest of the utility function. Or the Spruce Goose, right? Howard, he's Bruce Goose. Exactly. We did it with nuclear power plants. The French built the Super Phoenix Breeder Reactor. I never finished it. It was huge. I actually visited it back when I was under construction in France. The nature of engineers is not to try to figure out by a paper or PowerPoint where the limit of scales matter.

29:44They just keep building it. It's not the physical limit of scale. You could build big airplanes. It's the system limit of the scale, which actually has high utility value. And of course, that varies by definition depending on the product. There always is a limit. But that doesn't mean that it's the end of data centers. There's a natural limit to hyperscale data centers in the sense of how many data centers do you need that have more square footage than the Burj Covey Tower and that use the power of a city. I mean, a data center that gobbles 100 megawatts, which is what the big ones do now, it has a location challenge.

30:18Once you put a few of them in a place, you can't actually power it. I have a theory how they're going to power them. I mean, we've already seen hints of what will happen next, which is perhaps obviously small nukes. Those are the people who should invest in small nukes. You just made me think of the Washington public power supply system. Yeah. So on this, I want to pivot to – because you make a comment about 5G. You wrote, as 5G becomes ubiquitous, each mobile user will have access to 100-fold increase in bandwidth capacity. and each network cell will handle 100 times the number of connected devices and 1 ,000 times more data traffic per square mile, end quote.

30:56While we don't disagree with your view and statement of 5G, we also think about the rate of change because to your point, there's maxims and there's limits. So 4G was a big pickup in terms of speed and what we do. 5G is a big pickup in terms of what can be done, but relative to 4G, it's, to quote Mike Freeze, is a former holding of ours, our investors, he said, the CEO of Liberty Global, he said that 5G will be evolutionary, not revolutionary. Okay. And so I want to ask, I want to throw that out to you and ask you like 4G was revolutionary. Just look at me watching Hulu on my phone. It proves that 4G is revolutionary.

31:36Do you look at that, that same way? Do you agree with Frieze's view of that? Well, not entirely. I mean, this is, these are the nuances of that matter. I agree that going to 4G was a big step function and therefore revolutionary in terms of the bandwidth that handled. And I agree that the incremental change from 4G to 5G is less of a smaller leap. However, the question that's been raised about 5G correctly is the use case. Because to your point, I don't need 5G to stream live video at a quality that people find. I know where you're going with that, Mark, so I'm going to ask the second question because I know exactly where you're going.

32:15So would you agree with Freeze from a consumer perspective versus what you're about to say is how do you look at this from a commercial perspective? Well, again, yes and no. So the consumer use case is fully met with 4G, with the products the consumers are now using. Correct. And the commercial use cases are egregiously underserved with things we already know we'd like to use in the commercial environment. The problem is the commercial use cases in terms of volume for the providers is too small to justify the capital. It's missing right now, in effect. So let's use two examples that are obvious.

32:56You know these examples. And these are classic examples of getting timing right and looking at the nature of the underlying cycles of the technologies. For the consumer side, the use case for needing more bandwidth than 4G can deliver is obviously going to be things like VR and AR. Rendering voxels instead of pixels going to real three-dimensional, not gaming, but anything that involves that class of visual imaging, 4G will have a problem. you're going to need 5G. But we also know that the products that are being offered, and we'll wait, let's see what Apple's going to unveil soon, but the products that are being offered are not much better than the embarrassing introduction of Google Glass, right?

33:40I mean, it was called the glass hole when these little scenes came out. So there's a challenge on the use case on the consumer side. So you'd have to ask yourself, if you look at the use case stuff on the consumer side, Do we really believe that we're never going to deliver, by never I mean within the usefully foreseeable future, that we can deliver products that are fully immersive, that have voxels and not pixels? I think the data and evidence on that are overwhelming in the case. That's what's coming. But we haven't seen a product equivalent to the iPhone, the smartphone, be introduced yet that will stimulate that demand.

34:18Commercial side, the use cases are, I'll tell you the one that's most likely to come sooner but the commercial side it's the edge requirements for true automation in the majority of industries in transportation the combination of bandwidth and latency required for the scale of information to control systems like cars and dead serpent cores or like robots in industrial environments these are extremely bandwidth intensive very very data rich machines that You can't put enough processing horsepower in the machines to do the function you need. It's going to have to be resonant nearby the edge data centers.

34:59That use case is, in fact, you could argue, almost here now, if not here now in many cases. But if you're the industrial customer that wants to have this kind of control system, you're buying private networks. In fact, the company that I like, you guys may not have heard of it, it's a private company called Regent. they've made uh a what you could call a 5g level wireless mesh network for mining and industrial environments and military that is it's just astonishingly robust and fast because their first customer was the military which is trying to handle high data requirements of autonomy in very difficult environments you know there's no networks so they have to bring their own network in a sense.

35:42So the technology is clearly there because it exists. You can buy it now, but it's not there at scale. So I, long way of saying, I agree with them that the revolution is not, is not that going from 4G to 5G is itself revolutionary, but in combination with the other technologies, back to my sort of rule threes, the combination of all these products are in fact revolutionary. Yeah. So I want to hit on the data centers because I heard an interesting podcast recently where Jim Chanos was talking about how he was shorting the data centers. Okay. And as you talked about, you know, I mean, the growth of the data center business has been remarkable the last 20 years in a way that, you know, most people would have never predicted that, you know, data centers, you know, in the early 2000s would get to, you know, where they are today.

36:29But he also contrasts what he calls the super scalers, Microsoft, Apple, Google, and Amazon versus the kind of the legacy data centers. And he looks at the legacy data centers, which are obviously REITs. And he looks at it as kind of like it's the biggest misallocation of capital he's seen in a long time. So how do you look at the big players, those big tech companies going into hosting versus kind of the legacy data center model? It's interesting. So this is where we get into the intersection of the sort of three domains. What does the technology permit issue? Where does it want to go, I guess, is the way of putting it.

37:06Then what are the markets doing behaviorally? Because the use cases for data centers are now incredibly varied beyond obviously. I mean, banking and hospital use cases are very different from streaming movie use cases. And everybody has a use case now because everything is getting digitalized. and then the the third domain that always intersects the real world which i wrote very little about in my book for deliberate reasons which is the political public policy world i would say i think there's three things going on and again set aside the short-term cycles of whether it is over or under build because these industries are always cyclically build over building and under building it's in a sure by definition you can't know the future well enough not over or under built.

37:57So I don't care if it's roads or data centers or pharmaceuticals. So set that aside, we know a few things, I would say, in principle. We know that the world is under-digitalized, that the number of things for which we'd like to collect information and do analytics on it that are meaningful is underserved and under-digitalized profoundly. And we know that because if you talk to anybody in any business about what their use experience with most of the new things you're doing, you don't get a good report card, right? It's pretty hard to use except for what we'll call the 20 % of the economy that's heavily digitalized, which is news, entertainment, and so on.

38:40So we have an under-digitalized environment with an infinite demand for data. I mean, this is the only place in which there's infinite demand. Everything else has saturation. There's essentially no limit to the granularity of quantity of data I might want to collect. about some commercial activity product, whatever. Because granularity is everybody, everybody wants more granularity and they want higher velocities. And they want to know more about what they've collected, which is, again, processing power. The combination of processing power and the data would suggest that it's going to have to happen somewhere, and it's going to happen at data centers.

39:15And as the leading edge pulls the users along, which is where you're going to find Exascale, exa-flop computers in the Amazon and Microsoft data centers. But the trailing edge, it doesn't need exa-flop. It needs peta-flop, but you don't put peta-flop in your basement. So the trailing edge is filled up with another class of users. I think that scale of demand for data centers, again, we'll still go a decade up, fastly exceeds the supply that exists today. Question is, where will they physically be located? And which businesses need which class of data centers? The The idea that all the data centers of the world and the functionality of data centers will be owned by, pick a number, three to six players, it seems to me a very, very hard case to make because, again, it's quite unlike the electric utility model, which I use a little bit and many others have used to analogize the cloud.

40:12The problem with the electric utility model is, it's a utility function in democratizing stuff. It's a good model. But from a physics perspective, it's a really bad model because there are only so many lumens anybody wants in any building. Well, also from a security perspective, too, you tie everyone together. I mean, to your point, if you want to ruin the world, you kill three to six companies and you can ruin the world. Yeah. You know, and this is true for electric grids for most countries, not America. You know, we have we have over half a dozen grids, arguably 10. Yeah. There's no way to kill the grid because there is no the grid.

40:47and it's true for every financial institution of any consequence is not in any shared cloud amazon's or anybody's they have their own data centers that's not going to change totally and that's going to be increasingly true for uh we'll call high value assets and then you add the political feature to this which is of course the eu rule which affects the rest of the world is that the primary data has to be resident in the country in which the primary activities occurred. Well, okay. By definition, you're going to overbuild data centers because the economic efficiency would suggest you don't do that.

41:21Well, we overbuilt car manufacturing for the same reason. We actually over... Well, and you haven't mentioned John Sherman yet either. Well, so I would real quick on the data center. So you mentioned the one in Reno, which is, you know, one of the largest in the world. I love Reno. Here's my question though. And you point out there's higher rent per square foot there than there is for the Burj Khalifa in Dubai. Okay. Now, let me go one step further. We ask and want to kind of think about this question because we've thought a lot about internally here about this tug of war that suddenly showed up between labor and capital.

41:55So that place has, to your point, that is a capital game. That data center is all about capital. But the Burj Khalifa is actually about labor. And so how do you think about that? It's like a paradox in a way about that tug of war. The tug of war is inherent in every business decision. on whether you deploy capital for growth or you hire people. So CapEx versus OpEx, right? And it's always a tug of war. And the tug of war is always dynamic because of the two things that are hard to predict. But you sort of have to watch how they evolve. You're trying to guess when the CapEx, the equipment, the hardware, is a better bet than more people.

42:34And the people question, it really relates to not just what the people cost, but it also has to do with the dynamism of the business that you're in if things change and you make a capex decision typically the equipment is locked into the present it doesn't change in the future to say the obvious whereas the person can be retrained on the fly the humans are really really adaptable they're really good at that so that that that tug of war is is wired into things and well again this is why i think what's incredible about the time you live in so that That tug of war has had the sort of a fixed battle lines because the humans are adaptable.

43:13You can pay them more or less, retrain them. The machines are not, but they're more efficient. So for high volume stuff, you can't beat the machine. For the lower volume or more dynamic intuitive stuff, you can't beat the human. And then so never the twain shall meet until now. This is incredible. You talk about a commercial data tsunami coming in the 2020s from all these connected advices. think of our conversation, what we just talked about, like 5G and manufacturing and things like that as an example, is there a risk? And this is kind of a framework question. Is there a risk that we have to let some data go as we won't have enough capacity to retain?

43:50And I asked this because in the brokerage firms, the 1960s, there were a lot of failures, not because the brokerage business wasn't doing well, they were booming. Stock trading was off the charts. It was because the paper trail caused them to fail because they couldn't settle trades. Therefore they failed on their equity. Is there a possibility where we run into this situation where great, we got all this data and oh, by the way, we got to get rid of it because the energy or the capacity don't meet that data? No, I don't think so. Okay. But I don't think so from a capacity or energy perspective.

44:22I do think so from the viewpoint of the utility function. So we're already learning this. So But the deal with the energy and resource side, if what you're storing data, storing information is accelerating, the efficacy of storing information is accelerating at about double the Moore's Law rate. So the Moore's Law rate in CPUs and GPUs continues. But if you look at the energy costs of storage and draw the same curve, it's an exponential that is far, far faster than Moore's Law. And it hasn't ended. This is back to Feynman's rule, you know, playroom at the bottom. And you can store information passively.

45:00Once you've stored it, if you don't need quick access, I mean, you probably know Facebook stores a lot of their legacy data. Literally, in, you know, CD-ROM, you know, Blu-ray, discs that are giant robot-operated cassette players. Yeah. You could go find your disc because you don't need the cat video in a millisecond. You can wait one second for it to find the cat video. Totally. So the energetics are the problem. The problem is data overload in this sense. If you collect more data than you need, it could be for a regulatory sense, but let's just say more data than you need to do research or analysis, it doesn't cost you much of the story.

45:38It costs you to process it. It takes time, even with ExaFlock computers. So a lot of what is happening now in one of the sort of innovations in software, of course, is what you might call data curation or analytics on the fly. So this began in the astronomy community. I mean, I wrote about this in my book, but not as deeply as I could have. But in order to collect the data to make the first image of a black hole, that project was a multi-exaflop project in terms of data that was coming in and collected on a virtual telescope the size of the Earth. It's astonishing, astonishing engineering feat.

46:18But the quantity of data... Oh, by the way, they called their information sharing for that network the sneaker net because the data quantities are too great to send either by satellite or by fiber. You take the hard drives and physically carry them on a plane to the central place to upload them around the world. So the sneaker net was cheaper, faster. But what they did is they developed a software system, which is starting to show up in commerce and research, that on a use case specific basis determines what data is useful and should be stored. So in effect, what we do with compression, when we want to send a picture, we take white space out and reassemble the picture, that compression.

47:05But now you're doing it at the front end in the data acquisition because the quantities of data are too great to be functionally useful. Now, there's risk in that, right? I mean, obviously, it's kind of like, you know, the obviously, am I going to scrub out data that would, in hindsight, I should have had because it would have been useful. But the reality is you have no choice because the data quantities are so great. Yeah, so back to space, because we're there. NASA has cataloged many products, and many of these products have become indirect spinoffs. And you talk about like the X-ray technology, for example, that NASA had in space.

47:46This brought us to a really interesting conversation that we just had with Margaret O'Mara in her book, The Code, where, I mean, we run into these technology executives who act like they're these egalitarian, libertarian, capitalistic, when in reality, the government spending at the beginning was the largesse that made the technology, okay? So now you think of like Palantir, who's the first customer? Government, that was the only customer. And by the way, they funded the business in effect as well. So, you know, how do you look at the government's role in, I'll call it the initial explosion of capital in these new technologies?

48:27Well, this is a really important question in terms of the intersection of the domains that my book centers on. What leads to innovation? And then what leads to innovation to be useful, commercialized? What's the role of government? And right now we have a full-fledged assault on innovation in the sense that governments everywhere are convinced they know who the innovators are and what needs to be innovated. Just innovation, Mark? We're going to stick to the subject of the book as opposed to how you choose to identify all the rest of the things that are going on. Look, I am not only a believer in the role of government, but I think it's clear from history, to your point, there's many examples of an extremely important catalytic role of government in early stages of many technologies.

49:27An extremely important and underappreciated and now egregiously underfunded role of government supporting basic, unpredictable research. And I use the word unpredictable for obvious reasons. Governments tend to think, and policymakers today especially, including the potentates of technology and the technorati, all think that innovation is like a cafeteria. And you go in and you can pick and choose. It's not how foundational innovation happens. It happens fundamentally by serendipity that's directed, but directed only in the sense that the general area of inquiry might be chemistry or physics or mechanics or flying or computing.

50:11But beyond that, we have no idea where the breakthrough is going to come from. And that's never funded. Not never. In our times, it's not funded by private industry. And it's underfunded by government. So what we have, however, are the people who correctly observe that the Don of LSI, courtesy of the government, or the Palantir owes a lot to the government's contracts and orders. The aircraft itself owes a lot to the government role early on. The computer, obviously, famously, with ENIAC and predating that Colossus. But the first computer predated the wars. It was, you know, in Iowa in 1936. Mark, you're jumping the gun on the next question, so I'm going to jump in.

50:55Yeah, you're spot on. You're spot on. You talk a lot about serendipity and technological innovation. You discuss Jan, and I have a hard time pronouncing his name, Sarkovsky, accidentally found pure crystalline silicon. When I read that story, it's like God revealed to him what was already true. Nothing is created. It's found. Aren't these just revelations? Well, yeah, isn't there a chapter in the Bible called Trevely? Anyway, yes. So the Tchaikovsky process for making crystalline silicon, named after Jan Tchaikovsky, is a great example of serendipity. And it's exactly like the discovery of penicillin.

51:40So two things happened. Serendipity happens, but it happens to somebody who recognizes that it is important. So you can't... They understand what they found, in other words. Exactly. You can't fund the backhoe driver to get you the serendipity. Nothing wrong with the backhoe drivers. I used to write a amount of backhoes and work on them as a young man. But that's a skill that doesn't have the skill to recognize the value of penicillin or pure... utterly magical crystal and silicon. To your point, it is a the revelation of something that's possible whether you imagined it in your head because you're doing research on something collateral or you saw it happen.

52:26You know a trivial example of course was Teflon which NASA accidentally discovered and the but the chemist who discovered it was a chemist and he understood why Teflon was important once it accidentally appeared in his formulation. that is one of the most bizarre features of discovery and innovation and most magical philosophically and theologically fascinating. We don't have, we invented none of the atoms that exist in the universe but we don't have, the suite of atoms is fixed we aren't discovering any new atoms on the periodic table and the forces that exist that enable different combinations of those atoms are very limited in number, we know what they are We know a lot about them.

53:10And yet we keep discovering, unveiling new combinations of those forces and atoms that do things that are utterly magical. You know, the old Arthur C. Clarke line about magic and technology. So yes, but how do you get more of that? Well, you let smart people who are curious pursue their passions. All of history has shown that that has led to all the kinds of magic that we really like and products that are profound and products that are trivial. It's like throwing spaghetti up on the wall and seeing what sticks, but it's also a needle. It's a needle in a haystack, really, isn't it? Like I see all these funds formed to invest in innovation, and it just looks to me like a lot of searching for needles in a haystack.

53:56Well, yeah, but the question is, is it going to create the new technology or is it going to heavily use the technology way better than the prior companies that did? I mean, that's kind of the two sides of it. Well, I think that the, let's use, as an investor, this is the challenge I think a lot of investment funds have, looking at a, trying to invest in a profound innovation, a radical change, something that comes from the serendipity. The only place that comes from fundamentally is that you give money with no strings attached to a university to give endowments and high salaries to really bright people to pursue their passions.

54:33But that's not investable in the sense of getting a better product. And pretending that you can make that happen, back to my analogy that it's like a cafeteria, you just go in, pick the domain, it's solar photovoltaics, or it's a battery, I'm going to come up with a magic battery. The next generation of electricity storage is not going to come from a venture fund. I just don't believe it. The next generation of electricity storage will be revolutionary, and a bigger step than going from lead acid to lithium, will probably come from an unexpected researcher somewhere sometime figuring out how to make a stable and manufacturable room temperature superconductor.

55:12We have no idea where that will come from. You point out the progress being made between wealthy and poor nations. Wealthy nations have 800 cars per 1 ,000 people. Poorer nations have 100 people per car. Is the Western world just being smugged to the realities of economic growth and progress by trying to treat every country like it's a USA or Amsterdam? Well, yeah, that's the question that answers itself, as they say. Of course. I mean, the arrogance to think that we've reached the apotheosis of technology development and human flourishing because we're so well off, we can tolerate some deprivation in our daily lives, which is sort of the goals of governments to reach other political and social objectives.

55:54I mean, it really is profoundly arrogant. And worse than that, it's profoundly amoral, immoral. Or also inhuman too. It's a hate of humanity versus to your point, if this is not the climax of society, why sacrifice? Let's continue to build what we have. Therefore, you don't sacrifice because you're not there yet. Well, and I think that the goals that we're being told that we're pursuing, broadly speaking, are always about utilization of the planet's resources. I don't care whether it's the land or the food or fuel or materials. It's always about that. And we know that the solution to that is a combination of wealth and technology.

56:41It's kind of funny that Elon Musk's latest plan he announced about energy for the future. The one thing he and I, I certainly agree with their plan, was how we began. He began by taking it from a perspective of optimism that the world could sustainably support farmer human beings that exist now. And I think he might actually have said, you know, hundreds of billions or more. And I agree. In fact, I think that's unequivocally the case that we know that the inherent capacities exist. We just haven't discovered. We haven't had the revelations yet of what those technologies are. Agreed. And that doesn't mean they won't happen, but we do know how they will not happen, which is governments spending more money on yesterday's technologies as opposed to unleashing sort of the imaginations of engineers and scientists and innovators to pursue passions and dreams.

57:41Well, great, because they assume that there is a lot of finiteness to this earth and this world. And as we know, there's far more infinite than there is finite time and time again. I want to jump to kind of a particular case because I always love this and I always heighten this example. And you had this in your prior book. You explain that LEDs have reduced coal use because of its efficiency. But LEDs didn't cause a decline of energy use, right? Part of the lie of today is like, oh, we're going to be so much more efficient, so we're going to use less energy compared to the luddites of yesterday, and therefore we're all better off.

58:19That's a zeitgeist. Are we just using more LEDs in our house? I mean, like your house is so much better lit. Why? Because the LED is more efficient, but you use 20 times the lights that you used to. Isn't it the wealth effect that you commented on just a second ago really what policymakers miss? because that enhanced wealth causes you to use more. Well, certainly, absolutely. So this is the old misnomer, the Jevons paradox, the rebound effect that efficiency doesn't cut demand, it actually fuels demand until, again, there's always saturation. There's only so many lumens you want in a room. So you get to a point where you have enough lumens even with LEDs.

59:00Lumens, of course, are the illumination itself, not the power to create the illumination. So if I make the power to create the illumination sensory-free, then the market will saturate to the level of lumens people really want. And that's essentially what's happening. And until everybody in the world is lit up, and we know this from the famous pictures from space, that about 80 % of the world is not lit up. Correct. So we got sort of a 10x increase in unmet demand for lumens, let's just say. And thank goodness for the invention of the semiconductor light bulb, the LED. That'll make it achievable.

59:36But that will saturate too eventually. But if you think about it, what's happened to your point is that it didn't mean electric demand went away. It is true that the absolute demand for electricity for lumens dropped faster than the absolute quantity of lumen consumption, if you like, in the Western world. So the net quantity of electricity directed to illumination still slid down instead of up even as the quantity of the product lumens went up. but electric demand roughly stayed flat so the question i pose to people who look at innovation especially as it tends to efficiency is that we know that air conditioners are today are roughly 40 to 50 percent more efficient than they were 20 years ago refrigerators are almost twice as efficient light bulbs are 10 times more efficient motors have improved in their efficiency so how come electric demand didn't go down because the population only went up you know 10 20 percent time period.

1:00:31I mean with this massive increases in efficiency the reason absolute demand to go down is because we have thought other uses not just more lumens but much more importantly to use the obvious example data centers and computing and telecommunications filled the gap if you like that we ended up with a net new source of electric demand and so if you look at the future from the viewpoint of what are the unmet needs of humans what else would we like never mind people who are poor, who don't have anything what we have, but what are the unmet demands that most people still have? Well, what are they? Well, there was no demand for cars before the invention of the car, by definition.

1:01:11And to pretend that there's nothing being invented that isn't as consequential in material energy and human use terms as the car, means that you believe that we've invented everything that could ever be invented, which of course is, you know, pretty fast-y silly. So let's go one step further on cars, because, I mean, the other part of this technology side, I got thinking and such a great theme in your book, Mark, was the idea that the new pulls the old forward, okay? The new pulls the old forward. So you get into your section on 3D printing and immediately it hits me. Just so in full disclosure, Mark, and I think our listeners already know this, but my weekend driver is a 2005 Ford Excursion, the king of the suvs okay and it has as a turbo diesel v8 in it right so i get to see what diesel's work was up front um the catch is that it's a 2005 and so i had a situation where you know i had to uh try and get a piece for an ac line that had gone about a metal line and so it had to be custom made by the human when in reality in the world we're going to ford could send the specs to that to a 3D printer that does a metal fabrication, and I would pick it up later that week or have it delivered to my house, and it would cost far less than having the person bend it for me.

1:02:34Okay? Well, this has been, of course, back to your original point, probably the government's been involved in this particular benefit of keeping old equipment around without having to have expensive spares on hand for everything that's a piece of equipment that's maybe 50 years old, not 15 years old, and which is the military and of course they have long sought not only keeping spares in digital form but if you don't have the digital version of the spare you can scan it with high resolution imaging and x-rays to look at what the part is and create it i had a personal experience with this 3d printing on a car part myself i have a 89 560 sl which is a classic last year that mercedes made that soft top car and the plastic pivot that holds the sun visor aged out in the sun because I leave the top off all the time broke.

1:03:25Sure. And you can't get the part. First world problem, by the way. Well, that's a classic first world problem. Thing is flopping around and the shop that does it got, you know, a 3D printed part. There's a whole orbit, a whole ecosystem of manufacturers with 3D printers and plastics and metals designed for that kind of part production, either because they've got the original part specs or they, you know, take one and scan it and they make it. But that feature alone, both for, it's not just for prototyping, but for repair and extension of life of an old product, really big deal. Creation of new class of products that weren't possible to make before, a really big deal.

1:04:08I mean, again, we're back into not just is the new pull the old along is pulling old along not because it's not the equivalent of the car pulling a the buggy that the horse used to pull correct it's enabling the extension of the value of something that was already inherently valuable so correct it's the additive feature and this is where a lot of pundits and forecasters make a mistake they they think just because it's old that's not useful. And I usually, you know, point out, well, okay, if the old isn't useful, tell me why we're still using stone to build so many buildings from. That's a good segue into the next question.

1:04:47Is there a risk we spend a bunch of money on lithium as the mode of transport and serendipity or revelation shows us to move forward to hydrogen instead, ruining the government investment in lithium? Well, yes, there's a risk. Anytime the government mandates an allocation of capital, the scales that we're talking about. So we're creating a systemic risk of unprecedented proportions, probably in dollar terms, not quite as big as, but beginning to approach the 2008 financial collapse. That's the scale of money governments are subsidizing and mandating being put to work for lithium-based transportation in electric cars.

1:05:27it's moving to the trillions of dollars of capital being uh dedicated to that effort and it will it will fail it will not work and i pray and hope that sanity will restore before all the capitalists deploy because it'll be very damaging is that an al agori by the way mark to your point i mean the how we've how we've been effectively we're just on the other side of the room and everything else that they don't want to touch, praying that that doesn't hurt us in the process of this foolishness. Yeah. Why do so many people look at technology as the destroyer of human flourishing instead of the tool of human flourishing?

1:06:09We went from 40 % agriculture employment to 2 % between about 1925 and 1970 because of tech. And now we're the breadbasket of the world with only one and a half percent of adult deployment in the United States? Well, you know, as you know, I begin my book with the observation about technology is human. We have this sort of construct in the popular debate that technology is this external thing that's foisted on us, and we have to tame it or control like a beast or a devil. I mean, humans are inventing machines. We've always been inventing. Humans have always We're wired to invent, to build, to make technologies.

1:06:53The very definition of a hominid is a tool maker and wielder. That's what defines us. And the tools get better. We instantiate our mental ideas. The smart tools become dynamic and adaptable. This is what we're about. And it's for human flourishing, to your point. I want to come back to hydrogen. We don't have to beat this to death. but I will tell you that the affection with hydrogen is profoundly misplaced. The fact that we've got a lot of environmental groups switching to hydrogen as their fuel of choice over electric vehicles and lithium is this indication of perhaps the recognition, implicit if not explicit, that the lithium experiment will fail as a mandate.

1:07:38There's going to be lots of electric cars, but they're not going to replace them all. But hydrogen is far more problematic, far worse, And I'll be happy to go on record and take a bet with that. I will put it this way. The IEA's forecasts, which are the most wildly optimistic, the International Energy Agency's forecasts, for the share of world's energy that will come from hydrogen 20 years from now, they give it five percentage points of the world's energy from hydrogen in 20 years. And I would say they're probably wrong by a factor of five. Well, yeah, because they're pretty terrible forecasters on anything is what we've seen.

1:08:14And so we're running short on time, but I want to get you into one of our favorite subjects. You wrote about the skills gap. You point out that it's not in STEM. It's in non-STEM degrees. It's also in non-college skilled trades. Why do people over-hype college for all and STEM as the goal? Is the overproduction of STEM like having too many people in manufacturing, in the manufacturing industries at the end of the 60s? Yeah, it is. And, of course, STEM has been oversupplied broadly for, you know, as long as we've been keeping data on it, as I point out in my book. So we know. And there's always episodic, narrow shortages in specific engineering or science domains.

1:08:56It's inevitable. And if you get lucky, you get caught in the right side of that shortage with a degree. Look, the first problem we have, broadly speaking, on the labor markets is not a shortage of engineers and scientists, broadly speaking. As you say, it's a shortage of skilled labor. And the good news is that the engineers and scientists are now getting better at upskilling people with lower skills. That is, chat GPT and AI allow upskilling. Robots allow the ability to either amplify or replace use of people in really menial, repetitive manual tasks and allow the person to be upskilled to do the cognitive or more difficult tasks, the more varied tasks.

1:09:34We're at a really fascinating pivot that's exactly the opposite of what the conventional wisdom is. There's no reason in the world why you wouldn't, if you're good at, and I talked to a lot of parents who have children like I do, you know, what should my children study? I said, well, you know, make sure they're well-educated, flexible, you know, moral people. But be a coder? I mean. Well, let's get at that. What if the robots are the cheapest coders? Well, that's what ChatGPT was originally, Semantic Web, as you know, was originally written. But, Mark, we're producing more computer science majors right now than any point in the history of the United States.

1:10:10That have a hard time having a conversation with another human being and can't get a date until they're 30 years old by default. We have no data on that, by the way. That's just our best guess. It's a hypothesis. I lived in Seattle for 40 years. I can tell you about the body politic. I know. Well, you know, Google did a famous study. This was a few years ago now looking at they have enough employees that can do an internal study to figure out which employees, which skill sets led to employees moving up the chain of promotions and management. You know, were they the coders? Anyway, the short story is they found that, to your point, human skills, creativity, and the ability to work with other people was far more valuable and correlated more success than whether they were a mathematician, a coder, or an engineer.

1:11:00And yes, what we're doing is we're oversupplying the market with computing science degrees. And what will happen is that, like every other engineering discipline that's been oversupplied in the past, most of them will not get jobs in those disciplines, which is, again, we have a lot of data on this. We've been overgraduating engineers and scientists for a very, very long time. This is not new. One last favorite question we've been looking forward to asking you is, why do you think that Berkshire Hathaway, J.P. Morgan, and Amazon were unsuccessful in their effort to apply technology to the healthcare system in a way that made significant improvements?

1:11:39Economic sense for them. Economic sense and significant improvements in life. well they did the equivalent of try to they did it too early so it would be no different than trying to form a company like FedEx in 1937 or the capabilities to bring a combination of AI which is important here that's actually effective and demonstrably so with superior personal diagnostics which are emerging in devices and all manner of things Those things we know are going to bring productivity revolutions to health care. But they're all pre-commercial or just any commercial viability. And they took a step out on the limb before the technologies were ready, frankly.

1:12:28And it was a noble idea. It's too early. Probably too early by, when did they do that? Six years ago, I think, five years ago. So they're too early by a decade. So a few years from now, I think you'll begin to see the tipping over in the productivity function of health care. You know, productivity in health care is like everything else. You want better outcomes for fewer inputs. And the inputs here are, you know, labor hours and dollars. And the outputs are about better diagnostic results because it's all about diagnosis. It's a data problem. It's an information problem. We have to amplify the doctors.

1:13:04And that's a classic science problem. Well, you know, as a former practitioner of real work, I mean, I used to look at all the medical diagnostic tools that any way any physicist would. They're all the same class of tools. My best friend in college went into medical research. He was a terrific physicist working in ultrasound. The tools are starting to get astonishingly better. And as we make the tools astonishingly better, the ability to collect information on the edges with us personally. And by the way, back to use case, a use case for 5G. If I can collect the quantity of information that I would really want to collect about my own biochemical physiological conditions and create a virtual digital twin just for me as I think about how I want to manage my personal health, the amount of data bandwidth required for that are off the charts.

1:14:00Huge. And it's not a crazy idea anymore. It's just not implementable quite yet. Sure. So there's quite a few things we didn't get to. But again, like I just love the themes that you pulled out of your book, Mark, where you're looking at old frameworks to understand the new frameworks. This is a must read, folks. The new technologies, you know, moving forward, old things. Like we're in the mall business. Who are our best new tenants? Online only retailers of the past, right? It's like, ha, perfect example. The mall's coming back. I've got a whole thesis. size, you know, retail. I love retail. Retail, you know, I almost wrote that chapter.

1:14:37I was going to steal, you know, Je pense dans un je suis. I was going to title the chapter, Je Chez dans un je suis. I buy, therefore I am, because people are natural buyers of stuff. They want to shop. Exactly. Yeah. So, Mark, is there anything else that we haven't talked about that you think needs to be mentioned? Because again, we're just, you know, huge fans of this book and we think it's just such a wonderful framework of looking at the world today. Well, I really appreciate the endorsement, the love, and the conversation. There's so much more to talk about in terms of the state of the world.

1:15:12If you like a competition with China, as you know, my appendix, my conclusion, I conclude, you know, I love China. I've been to a dozen cities in China. I like the Chinese, not Chinese government. But this is our century, not the Chinese century. We haven't finished. And I think that that's a reason for Americans to be very optimistic about thinking about our place in the world and trying to restore some optimism and sanity to our political process, because that matters, too. Yeah, you mentioned you mentioned the optimism. We're optimistic about the millennial generation. There's 92 million people and that's great demographics.

1:15:46And no one besides us talks about that. Well, that, and I can tell you, I'm much more optimistic than most people about the Gen Zs that follow them. Because I think if you look at the, if you tease out the demographic trends that are now emerging, behavioral trends, in terms of their buying behaviors, what they're looking to do, how they're, I mean, a jury's out, you know, all the naysayers could be right. This could be, you know, what's a lazy, you know, entitled. I don't know that that's the case. I'm not buying it. And I'm optimistic about them as well. Well, you point out that society always emulates those with wealth ultimately.

1:16:23And so watching college educated women have more children than non-college educated women shows you something about what is a normal good, something with higher income, you attain more of it. Right. And I think that's a really interesting dynamic that you insert in your book. I mean, that's a whole discussion in and of itself, Mark. I insert, I was going to do a whole chapter on that. And I I decided it was the distraction from the book. All right, well, when you get that, why don't you just write a book on it? We'll buy your book when you write that one. It's one of our favorite subjects. Oh, it's incredibly important.

1:16:53I mean, I - We agree. There's something really fascinating going on on the leading demographic edge of wealth with more children. In fact, you probably saw the data are out about the, you should have expected a little echo boom bump from the evil lockdowns. And it looks like that's, I mean, I've watched it anecdotally. It looks like that's what's happened, but it's highly wealth correlated. They know that, you know, to be simplistic, the trope is, this is the trope, societies become wealthier, they have fewer children because that's what's been going on for the last. Well, that's a lie that's been peddled.

1:17:26Well, that's what they say. But maybe that's true up to a point. And then as wealth goes beyond some point, it tips over, goes to a different direction. Again, your cycles. They recycle. We don't know that because we've never had a society where that experiment has been done at scale. and we're now doing it. So Mark, for our listeners, I mentioned your podcast, The Last Optimist, that they can find out there. Where else can people follow you? Oh, the magic Dr. Google. There's a lot more about me there than I prefer, but my website, the tech-pundit website has got all the things that I've written.

1:18:04Awesome. And you're on Twitter as well. I do the Twitter, LinkedIn things because they seem to be the best professional environments to operate in. I don't do personal public policy, politics stuff in my public sphere because I just don't. Sure. It's bad for health. That's bad for health. Neither do we. Awesome. Well, we want to thank you so much, Mark. This has been a lot of fun. I want to thank my dad for hosting this. Mark's book, The Cloud Revolution, provides, like I said earlier, an incredible framework for the exponential progress of technology and how existing human spheres will move forward at a more than linear, like our discussion on babies or the old economy, et cetera.

1:18:43Our audience should get this book. You need this book. Tech execs, to Mark's point earlier, need this book. Energy execs need this book. I can't think of an executive that doesn't need this book, frankly. If you enjoyed our discussion with Mark on this, go to Apple, Spotify, wherever you get your podcasts. Give us a review, rating, or recommendation for other podcast listeners. For our listeners, if you have a great book like Mark's that you'd like to recommend, email podcast at smeedcap.com. That's podcast at smeedcap.com. You can also send suggestions to us on our Twitter handle at smeedcap.

1:19:17Thank you for joining us for a Book With Legs podcast. We look forward to the next episode. Thank you for listening to A Book With Legs, a podcast brought to you by Smeed Capital Management. The material provided in this podcast is for informational use only and should not be construed as investment advice. You can learn more about Smeet Capital Management and its products at SmeetCap.com or by calling your financial advisor.