Mayhem: Big Tech Is Over Hyping AI? AI, Robots, Drones & Data Center Boom As Gold Prices Gap Up

Summary

Market Outlook: The podcast discusses the recent surge in gold prices, with gold crossing above $4,200 an ounce, driven by a weak US dollar and global currency debasement.
Big Tech and AI: There is skepticism about the sustainability of Big Tech's massive capital expenditure on data centers and AI, with concerns that the economic impact may not meet expectations.
Economic Pillars: The US economy currently relies on three main pillars: tech spending, government deficit spending, and consumption by the top 10% of earners, with risks if any of these weaken.
Gold Investment: Despite recent gains, gold remains underallocated by institutional investors, but increasing global demand and central bank purchases could drive further price increases.
AI and Productivity: While AI is touted as a transformative technology, its current impact is more about enhancing productivity rather than replacing jobs, with many companies not seeing significant ROI from AI investments.
Energy Demand: The anticipated growth in AI and data centers is expected to increase energy demand significantly, highlighting potential opportunities in natural gas and electricity sectors.
Geopolitical Risks: The podcast touches on geopolitical tensions, such as US-China trade relations and the impact of tariffs, which could influence global economic dynamics and investment strategies.
Investment Opportunities: The discussion suggests potential investment opportunities in sectors like energy, particularly natural gas, due to expected increases in energy consumption driven by technological advancements.

Transcript

Hi everyone, this is Jason Ber of Wall Street from Main Street. Welcome back for another Wall Street from Main Street podcast interview. We're recording this interview on Wednesday, October 15th, 2025. The US dollar gold price has rallied a lot over the last 6 to9 months. It's it just crossed above $4,200 an ounce. The S&P 500 is also over 6,600. So all these asset prices are going bonkers here in the United States, probably on a weak dollar, but also probably all these other currencies that have been weak over time. But the dollar has been weak a lot. so far in 2025. Today's special guest is a returning guest. He runs the newsletters Macrovisor and Trader Aid. Mayhem from Markets and Mayhem. Thank you for joining me again. >> Thanks for having me, Jason. Pleasure to be back with you. >> So, Mayhem, for our listeners out there who are not familiar with your work, you have a tech background. You worked in tech in the past. I want to get your thoughts on some of these uh investment banks comments because they're basically saying if the big tech companies were not so like Microsoft, Amazon, uh Google, Meta, those guys, if they were not spending like insane amounts of capital expenditure on data centers, artificial intelligence that the US would already be in a recession. Do you agree with that? >> I think that it is a question that deserves to be asked. And I would say that the caveat might be also if we didn't have such aggressive government deficit spending. I think if you put the two of those together without a doubt, yes, we'd see a recession in economic activity. We'd see, you know, many subsequent quarters of negative growth backtoback. Just AI spend alone, we may not yet be at that point. I'll have to double check the math. Maybe we are. But I think with the plans that have been made moving forward, certainly looking forward that would be the case. If those plans didn't materialize, given how much of a centerpiece that has been, that would be problematic. And there's sort of three pegs to this economy right now. It's really a lot of the tech spend and the euphoria around AI. It's the government deficit spend or sort of the fiscal aggressive uh agenda that's happened throughout subsequent administrations. It's not a partisan issue. Everyone's doing it. And then it's also the top 10% of consumers who account for over 50% of consumption. So those are the three pegs of the stool that the US economy is standing on right now. And I think if any of those three get wobbly, we start to see more shock waves in the way that data may look, if we're a ever able to get it back to a regular cadence of data releases again, being that we're in a shutdown right now. but also the knock-on impacts to various asset classes and I think that if you have two of the legs of that stool come out at any given time then a recession becomes near certain. >> Yeah. So we are on a government shutdown but Trump um do you think that like a lot of professional managed money so these institutional investors that had basically 0% gold exposure around 40% of all professionally managed money institutional investors pension funds had 0% gold exposure according to Bank of America's research. Do you think that they started to get like weary with the budget deficits when they saw that Trump in order to get the tax cuts passed in order to get some of the spending he had to get Congress negotiate do all trade all these votes and then still the budget deficits are just blowing up. >> I think it creates a lot more concern about the future of the dollar its purchasing power and how you might want to hedge against some of those risks. But I also think that we have a global backdrop of fiscal expansion more often than not, monetary easing or at least policies that are going back into an easing position, which also helps to create more demand for gold. You have central banks around the world that have been increasingly concerned after the US used the dollar as a weapon against Russia after it invaded Ukraine. Say what you will about that policy, but the one after effect we can know is that governments and central banks around the world are more weary of the dollar and some of the dollar denominated assets, particularly debt. And so they're buying more gold, particularly global central banks. And this has been a big driver of gold demand. And like you said, Wall Street and really investment professionals in the West and primarily in the United States have been woefully underallocated to gold. And I think that with what's happened, the momentum can't be ignored, the trend of what's happening, not just to the dollar, but every currency around the world, which is that the more of it that's in circulation, the more new credit that's created, anything that resembles that, that can create demand, that over time debases purchasing power. And so that leaves us to, well, what can we buy to hold on? Well, obviously people are buying stocks. That helps to offset against inflation pretty effectively as long as it doesn't run too hot. And then people are buying gold, they're buying silver, they're buying Bitcoin, they're buying some of these other things that, you know, are considered hedges or offsets. And I think that gold especially here benefits from a backdrop of increasing uncertainty across the board. Gold is a really powerful hedge against uncertainty. And you'll see there'll be days when things happen in the headlines geopolitically or economically or otherwise and everything else is risk off. And gold's got this very steady strong bid behind it. Even things like Bitcoin are de-risking, you know, even you'll see sometimes selling on the long end of the yield curve, which used to be a flight to safety area or even the dollar, but gold will still be bid. This is a paradigm shift that's in its early innings. That's not to say that gold hasn't run really far pretty fast. If you zoom out to a weekly or monthly chart, it starts to look pretty parabolic, but I think as long as we can see some healthy consolidation, gold has a lot more room to run before this bull market in the metal exhausts itself. than one key area that speaks to your point. If you just look at commitment of traders reports in gold futures, the number of large asset managers that are long here with gold hitting historic new all-time highs is still low compared to other runs. >> And if you look at the gold price in other currencies, it was at or near an all-time high at least 12 months, normally like 18 to 24 months before the dollar gold price started to make its move. So I would argue that um the macro conditions were already favorable for gold before this big run over the last uh 10 to 12 months. So because what gold was uh a little bit above 2200 2500 in a rangebound and then we just had a huge gap up from about 2500 up to like 3,000. We blew through 3,000 and now we're here we are we're above 4,000. >> I agree 100% with what you're saying. I think those are all exactly the backdrop. And also again going back to Russia, weaponizing the dollar again because we've done this before as a government, the United States government has gone after countries using the dollar and Swift and other parts of the Western financial system as a weapon. And all that does every time is send ripples through the global financial system to increase the steady pace but hastening pace of dolization. Now, is it going to get to a point anytime in the reasonable future where the dollar is not the global reserve currency? Maybe not. But we are seeing increasing concern about, well, what does the purchasing power of the dollar look like 5 10 20 years out and then all these other currencies as well because we're all sort of in this global race to the bottom printing money. And you could say, well, technically they're not. You're right. Technically, it's a lot of central banks creating credit that governments have to pay back. I don't know that you can argue that's better, but the end effect is aggregate demand expansion and debasement of purchasing power across the board. >> Well, they're subsidizing government debt because you're not seeing a lot of these governments cut back. So really like uh the currency creation that you're talking about. The private sector is getting crowded out. It's not the consumer. It's not the small business owner who's benefiting from all this currency creation. They're getting killed with higher inflation, higher input costs, shrinkflation, the tariff bill. It's like the large corporations. It's the governments because they're not cutting back. The governments are getting artificially cheap interest rates, especially if these central banks are expanding the money supply to cap uh yields on government bonds. >> Absolutely. You have a situation where the lower tiers from the middle class, even the upper middle class in some cases, depending on their composition of debt and asset exposure. A lot of people are falling behind. And a lot of those same people are small business owners and operators. And so it all kind of goes into this mix of this K-shaped distribution of socioeconomic outcomes. We talked a little earlier about the top 10% making up in aggregate about 50% of consumption. That's a really alarmingly high level for such a small segment of the population to account for. >> Is that like the 1920s? Is that like the roaring 20s? The wealth disparity. >> It's very very similar. and and also the wealth disparity that we have now, and I'm glad you brought that up, is worse than back then. So, you have a situation where underneath the surface, you've had these rolling periods of what you might describe as sector or industry or classdriven or business size specific recessions. And I'd say that applies to the working class. It applies to a lot of small businesses. It applies to manufacturing. And so you have a lot of these areas where there's been pain for years, but because there's been this big K-shaped distribution of outcomes where the rich are getting so much richer and companies are growing, like there's so much concentration in these mag seven and even just the top 50 S&P companies in their growth and their hiring trends and otherwise, it's kind of masked a lot of the weakness underneath the surface. So that's one of the things that I think also creates some consternation about what the economic trajectory looks like without one or two pegs of that three pegged stool we talked about which represents our economy the really wealthy consumer spending which a part of that is because of wealth effects. If the S&P wasn't continuing to go up, if housing prices weren't continuing to go up, crypto, etc. you'd see that spending come down and in fact it was threatened a bit after the April May swoon on tariffs. You saw that trend start to come back as wealth effects went negative. If government aggressive deficit spending comes back that and and it comes back meaningfully from where it is right now where it's about 2 trillion that's going to be uh something else that could be delletterious to the expansion of this economy. I know the mantra right now is run it hot. let's grow the economy faster than our debt. That's kind of a fantasy and that's not really something that you can do for any durable period of time with the amount of debt we have with interest rates where they are and with the risk of inflation causing the Fed to have to be a little more careful about dropping rates or you know going back to balance sheet expansion at a time where again this is a backdrop where more of our trade partners, foreign governments and central banks have become more reluctant to buy our debt. So, we have to generate more demand. You know, we don't necessarily have a way to do that internally too meaningfully because if you're a fixed income manager and you're looking at, you know, twos and 30s trading with the spread they are, you're going to say, well, I want a higher premium for the longer end of the curve, even up to 10. So, you know, where's that demand going to come from? And you mentioned earlier, well, a lot of these central banks are subsidizing the government through things like QE and artificially low rates. Well, eventually you have to get there, but you can't really get there if you're running things too hot. And that's kind of the paradox that we're in right now. So, I think that there's going to be over time some degree of pull, whether it's one or two or all three legs of that stool coming out, cuz you know, when these things move, they move pretty fast. Eventually, you're going to have an economy that's going to be much harder to repair because we have so much slack from the most aggressive deficit spending with unemployment this low across two administrations. It's it's fiscally irresponsible, but it also creates pain. Like you said, small businesses, the middle class, the working class, they're not at the money spigot. Those are the financial elite. They're the S&P 500 that are able to both park short-term cash, you know, at or just above Fed funds and then also have refinanced a lot of their debt at those COVID lows or not too much higher. And then you have small businesses that are able to get abysmal savings if they're lucky enough to have cash in their treasury and they have to deal with revolving interest rates that are whatever the bank puts on top of the Fed funds rate. So it's such a dichotomy of outcomes in the businesses in the households and I think that's the bigger concern is when there is an economic fracture and you see the top tier of that K-shaped distribution of economic outcomes for the rich and the successful companies and the amount of government spending and its impact on an economy with so much slack. when those start to dry up, that's going to be a much bigger problem. And I don't exactly have a time frame as to when that can happen, but I think there are signs along the way we could look for. And the one that we're starting to see more warning signs about is that narrative around AI driving growth in the economy now and in the years ahead. >> Well, we also have hints from Jerome Powell, Eric Trump, uh Scott Patent, President Trump that basically more QE is coming. Uh drone pal basically said that the labor market is a lot lot weaker than they thought. The Bureau of Labor Statistics basically just removed 2 million jobs that were supposedly created over the last three years of the Biden administration. They were just vanished. They were downward revised out saying, "Oops, sorry. Uh those two million jobs, they never actually happened." Uh so I mean the if that the Fed looks at the labor market, they're going to probably they're probably being pressured by President Trump and Scott Ascent to cut rates so the US could start refinancing more of its uh debt long-term at lower interest rates. But then you have uh Eric Trump. I think he's been hinting that basically they want to restart QE as soon as possible once they get in more Fed governors that vote for QE. So they're trying to bring in new Fed governors. What with with Lisa Cook and then Scott Bent and President Trump have basically said over the last like 6 months that they want a weaker dollar even though Trump did a speech what yesterday I think on Tuesday. He was talking about oh the dollar is going to remain strong but just look at the dollar index how weak it's been. I know it's rallying but it it's been very very weak for 2025. Yeah, it's had one of its worst years ever and we'll see how it ends the year, but it certainly has been a run for the history books and a lot of that has to do with reducing confidence. What I heard from Powell was that they want to end QT and it kind of makes sense. The run rate of QT is so low right now. I think the cap is 5 billion a month on treasuries. It it almost doesn't even matter that it's happening. It's it's such a low rate. like they're out there at the auctions buying treasuries to replenish their balance sheet because so much is running off at a faster rate than is allowed by that cap that it's inconsequential. Some folks look at those auctions where the Fed is participating and say, "Oh, look, they're doing QE again." But I would caution to say they're just maintaining their balance sheet. And now they're going to say, well, now that we're, you know, we've shrunk our balance sheet by um not an immaterial amount, but what just about one and a half$2 trillion over the last several years, they're saying we're going to keep it around this really enormously large level and just kind of replenish our holdings and treasuries and mortgage back securities, which I would say it's neutral in the impact it has on the market. In that same speech that Powell hinted that they want to end QT, he said that the Fed should have stopped QE before November of 2021. And I would agree. I think they ran their QE program both too hot and too long back in the, you know, postcoid crash era where there was this bazooka from central banks around the world and government spending just aimed into the global financial system and it just stimulated things too much. It created asset bubbles. that help to drive inflation. And so I would agree with his sort of hindsight is 2020, but he was also the guy that told us inflation is transitory. And he's kind of told us that again about all that's going on right now. Now, obviously, he could be right. There are signs the economy is slowing. Um there's, you know, increasing signs the labor market is having some serious difficulties. You mentioned all those jobs that never existed. And it's funny because we've been talking about this I feel like on and off on mic off mic for years that things in the labor market weren't what they appeared to be. And that there were a lot of signs that Jolts was full of ghost listings and that there was a lot of nonsense going on at the time in tech in late 2021 and early 2022 before that whole software developer job bubble popped. And we saw a lot of fallout across tech companies laying people off pretty aggressively from 2022 on even into the present in some of these companies including some of the large ones. And it feels like that was a disconnect between the reality and and really what was going on in the economy that's similar to the disconnect we have now where we know there's something that's different between how these lower tier and mid-tier earners are experiencing life versus wealthy earners. We know there's a difference between small and medium-sized enterprises and the experience they're having in the economy versus the large companies. There's so much macro data to illustrate that, not just here, but in many other countries around the world. So, it's kind of a similar thing to the labor market. You've got this feeling something's kind of rotten underneath the surface. And over time, we'll probably have some exposure to that. But in the here and now, I think we have to kind of look forward to there's still another inning or two left in this game in all likelihood because we haven't seen, I think, the peak phases of mania. And that's not to say that we can't have some pain along the way, but I don't necessarily think that the hype plateau has been realized because you still have the plans for such aggressive AI capex. And we've talked about this for a while. I thought they'd pull back a bit earlier. It would have made sense to pull back a bit earlier with the compute efficiency gains that have been made with how you can have really well structured smaller total parameter models outperform the larger ones where quantization to 4-bit precision is allowing less compute and less VRAM to perform similar to 8bit on some of the higher uh I guess the the more sophisticated quants that are coming out from companies like Nvidia and Unsloth and otherwise. Point being, you need a fraction of what you used to and you can do more now. I can even do more on my computer than what I was able to do on the cloud in many cases in the first quarter of this year. So, there's just so much progress happening with how you can use less to get more from AI. But then there's also the glass ceiling overhead, which is that despite all of this progress, we're still having trouble seeing constructive use cases at scale that resemble anything even close to the scale of capex so far into AI. >> Okay. Before we talk about uh big tech and AI spending and learning language models and how those might be sputtering out that the efficiency gains, productivity that you talked about that the uh return on investment for companies spending and just throwing billions of dollars at uh chips and data centers and computing like the the productivity gains may be sputtering out. Uh Scopascent supposedly it's being reported in the mainstream media that Scottent is interviewing candidates for Federal Reserve Bank Governor if they can get Lisa Cook out and he's asking them like about quantitative easing and restarting. So, it seems to be the plan. So, while President Trump is talking about, oh, we want a strong dollar, I mean, a lot of the actions seem to indicate otherwise. >> I I would not doubt that that's something they're going to push for. And I think Powell will push back until the economy weakens materially, and that would be the wise decision to make because if you start to add even more heat, you could really rekindle some level of inflation again, and you could start to have some serious structural economic problems as a result. So, I think that one of the things they're also doing is basically making sure they have a person in the Fed that's kind of willing to pull the lever when they want to do it so that the Fed can become less independent and more under the direction of the White House and the Treasury Department, which it seems that's the trajectory that we're getting from this administration that they really want to tell the Fed what to do and when to do it. Now, Powell's got until May of 2026, and unless they're able to find a way to fire him before then, he's still ultimately going to have a lot of influence as to what decisions are made there. But once that time period passes, uh, assuming again he's not somehow gotten rid of before then by this administration, I think that it's going to become a lot more open for the Fed to be a bit more of a marionette of the administration because they're going to start doing what they can to exert greater and greater influence and push against any of the people at the Fed that don't want to vote on what they want. So, it seems all but inevitable. I just don't know if we're there yet. >> Yeah, I think S&P if that does happen and they announce yield curve control, the Fed's balance sheet expanding again by another trillion or two trillion pretty quick, some type of stimulus package or some other program that they announced. I I think we could see they've already secretly bailed out Intel. There was a mix of like private sector funds with in uh with uh Nvidia and then a government backing with that and shenanigans with with Intel core with a secret bailout with that. But I think we could see, you know, gold prices way above 5,000. We could see uh silver prices enormous, S&P 500 up. I mean, we could see like huge asset price inflation if they do announce like an official QE program. >> Yeah. And you also have to wonder what's on Secretary Bessant's mind that he's thinking about this right now. Because if you look around the corner and we talk about those three pegs again, there are signs that any of the three could be under some level of duress. For example, if we start to look at uh the impacts of any kind of reduction in the amount of spending on AI capex that could happen over the next year, that could send shock waves through the economy. That could be something that could get a Fed, especially one that's more under the influence of the White House, to start to try to turn about and they could call it preemptive easing. If you have a uh situation where the government remains shut down for a very long time, this is also becoming a growing risk for a different reason because this is becoming a reason to sort of punish the federal workforce. And while you know anyone can make a reasonably sound argument that we should shrink the government, I'm not so sure that this is the best way to go about doing that. And it seems like it's going to be proving to be increasingly disruptive the longer it goes on and the more people that are fired because you start to get into situations where some difficult questions need to be asked. One of which is this is unlike a lot of the other shutdowns in the past. The office of budget management had always given some kind of contingency plan about how the furled employees might receive back pay when the government reopens and otherwise. And this time around, not only is there no discussion about how that might work outside of the troops and some of the law enforcement, but there's also this propensity to lay off thousands of government workers. And while those cases work their way through various federal courts as to the legality of these firings and otherwise, there's this potential for some degree of disruption. And you already see things like auto loan delinquencies going up, credit card delinquencies going up. So >> student loan debt, too. Yeah, there's Houston going on now. >> Yeah, that's been an issue especially because, you know, now the the sort of exemption is gone. So, everyone is starting to face some pressure there. But the one area that you really have to watch if you want to start looking for signs that there's real problems in the economy is going to be the last form of debt that people want to default on, and that's their home. If you start to see mortgage delinquency really rise, that's when you have a problem. And so I think we're not there yet. You do have a lot of yellow flags along the road in the economy, things that are worth monitoring. And I think that those things could put some pressure on where we are right now and maybe slow the economy, but we're not yet in red alert territory over the next quarter. Now, looking ahead into next year, it's a lot more unclear because there's a lot more moving parts on this road ahead. And one of the bigger road uh you know moving parts on the road ahead going into the year after into 2027 to 2028 is going to be the outcome of the midterms and how that might affect future federal spending. >> Well, you also have more aggressive and more extreme negotiating tactics from the Republicans and the Democrats. So the Democrats basically have in the past they might not have uh shut down the government at all because they wanted the government employees to make sure that they were getting paid and you know regulation because that's part of their um their policy ideas. But then President Trump has also threatened that if the government shutdown continues that some of the government employees not only will they be fired, he says some of them won't get back pay. So I mean there it just seems like both sides are just lobbing like you know Molotov cocktails about uh negotiating tactics here more and more extreme. >> Yeah, there's really been no official plans on backay other than what they're now talking about from military and potentially law enforcement. There's been no official document about that, no official discussion. And with every other shutdown, there was always some kind of contingency plan. So, it does feel a little bit like flying by the seat of our pants as a country right now in terms of some of the things that are going on. Uh the US China trade situation also seems like, you know, tensions flared up. I don't see this as a huge concern unless we get to, you know, November 2nd and all of a sudden we've got 100% tariffs and this time it's actually real and it lasts for a period of months then it's going to start to have an impact and especially if rare earths are not being imported into this country. But my thought on that Jason and then I want to hear what you think is that there's going to be cooler heads that prevail before that deadline and we don't see those tariffs or those restrictions in in full play. Uh I mean China always threatens the rare earth stuff. I've been I wrote a research report. I spent over 80 hours uh researching the rarest market. I think back in like 2011 and I was co interviewing a lot of rare earth expert over the years. So from like 2011 when Molly Corp went IPO before they went bankrupt I think in like 2014 or 2015. I was covering the industry extensively. No one has talked about yet um mayhem. No one has brought up the company. I see all the talking heads. It's all the miners. No one brings up the Canadian company that has like locations in the US and other countries, Neoperformance Materials. They actually make the magnets. So, if I was President Trump, I was the Trump administration, I'd be calling up Canada, calling up Neoperformance Materials, and I would say, "We'll give Canada tariff relief. We're going to have we're going to fund Neoperformance Materials to go build a magnet facility here in the United States." I haven't heard it yet. I don't know what's going on over there. >> I think that that's an area where we need to invest, though. We really need to have our own rare earth supply chains that are diversified away from countries that would even threaten them. It's too critical for us not to take that seriously. So I hope to see more investment. I know we're starting to get into it. >> Well, the rare earths themselves aren't actually that rare. The only one that's actually like rare that I've studied is actually dprosium and it doesn't occur in in any economic concentration in really any rare earth deposit. So that you need the dprosium to uh increase the the magnets so they don't deal with the they can deal with the heat about 30% better according to a research report. So you need the dprosium in order to make the rare earth permanent magnets the neodymium iron boron magnets. Um it's just very difficult to to find the dprosium in any of the rarest mines. So all the rarest mines that have been found over the last like 15 or 20 years they have almost zero dprosium in them. Well, maybe technology will find a way to, you know, make another material that can similarly augment magnets for better thermal resistance, but I think ultimately we need to diversify and and that will help us, you know, to secure better trade relationships. We probably shouldn't have gone in guns blaring without having laid some of the frameworks for these specific areas that could be used as pressure points, but we are where we are and so we just need to monitor the situation. But ultimately, I think that the reaction last Friday was way overdone. I especially in what happened to crypto that was its worst dollar amount liquidation ever and one of its worst percentage liquidations. And you know, with what it is with so many people employing leverage, even firms, there were a lot of portfolios that went to zero that day. So, you know, obviously it's a study on risk management and complacency and trading psychology on the one hand, but on the other hand, it's also kind of a sign that when this market starts to get scared, the first thing that people want to sell is the thing they can't value and the thing that has the highest implied beta and that ends up being crypto almost every time. >> Well, as you know, that's not President Trump's negotiating style. So, he's bombastic and he's kind of outrageous and he he goes to the extreme and then he tries to talk talk things down. So that's more of his negotiating tactics. >> Yeah. The art of the deal, as they say. I I don't know how well it's doing with our trade situation. But, you know, I understand that it's an element that we as market participants and economic participants have to contend with. So we have to kind of focus on what are the after impacts, what are the probabilities and like you said it's mostly just threaten big and then come back and try to make a deal where everyone can walk away looking like they did something constructive as particularly Trump and his team want to be able to have some talking points they can come away with to say well we did this and that and that and you know everything's on the up and up and we got this done now we're winning for x and yz reasons and that's fine that's kind of where I see this situation playing out it's sort of more the same. But I think also China is going to be more and more tempted to start to increase the amount of economic stimulus that they're going to be using in their economy as time goes on because the real estate crisis is what's keeping their middle class and wealthy class from spending. And if they were spending enough, that could offset some pain on the export side. So if they ever want to become a little more independent from being the world's manufacturing house, they have to repair their own domestic economy. And as time goes on, foaming the runway to kind of prevent a better a greater crisis as it were is not enough. >> Well, China is spending on infrastructure. So they are building out what the uh power components the um infrastructure, the nuclear power plants, the solar plants, the wind turbines for the electricity demands for their modernized city, but also for the data centers for their artificial intelligence programs for their next generation of technologies. >> Yeah, they have and they have a lot of slack in their power uh grid unlike what we have here. So that is one area where they have a leg up. It looks like, you know, we're dealing with some serious scaling pains where you've got people dealing with power bills that are increasing by 1 2 300% in some areas where they're near data centers and the cost per kilowatt has gone up so significantly. And that's creating a lot of consternation. And that could be if any of these capex projections are even vaguely accurate, that could be just the beginning of the cost increase in power prices that are realized by people near these data centers. Which again, you know, this this goes to the whole K-shaped distribution of socioeconomic outcomes, but it also comes to this sort of transfer. You've got the AI companies kind of benefiting from a lot of what they're able to realize with their scale of economy with the amount of lowcost money that they're having invested in them and you have a lot of the working class and middle class on the wrong side of that. Now, their portfolios could be going up in price. Perhaps their homes are going up in price, but ultimately everything that they buy is going up in price, especially things like power and also water because these data centers need to be cooled by massive volumes of water and a lot of that stuff is competing with municipal sources or using municipal sources to augment that cooling demand. Are these uh six-f figureure computer programmers, these software programmers that have been working at big tech for a long time, they've gotten promotions, their middle management, maybe their upper management, are they starting to get fired now as um and then the tech companies are saying excuse, well, our artificial intelligence software, we uh we have employees that can do the work of a couple employees or productivity gains. I is that justified or do you think that um it's more that like the AI investments and the software being used as an excuse so these tech companies can maybe reduce their employee salary overhead and go bring in HB1 uh visa employees. >> That's that's a great question and there's a lot to discuss about it. So let's take it uh step by step. So, first one thing that I see that's going on with the hiring and firing trends is junior programmers and junior uh developers and data analysts and otherwise are just having a really hard time getting jobs. If you're on the very very new end, just out of college or just out of your first job, getting that next job or that first job is really hard. So, that's one area. Another area that I've seen particularly in the larger companies is they're firing a lot of middle management because they just deem it to be and perhaps accurately so inefficient. You know, you don't need managers managing managers managing other, you know, it just kind of gets to be a little bit of a bottleneck. So, I can see why they're doing that. Um, but in terms of your question about AI and the role it plays here, I think the message is overstating the impact, but I do think there's an impact. My interpretation having used generative AI for years and working even with local inference for some some of the pretty complex openw weight models these days both in training and fine-tuning and inference really if we're looking at um kind of some of the use cases that I have. I would say that everything I've learned is that AI is a force multiplier. It helps people get more done when applied correctly with reasonable expectations with properly specific models that are trained for the kind of workload that they're going to be doing and with guard rails where everything is kind of being human supervised and validated. There's in those situations the potential to save you know hours a day or a week and become more productive. Now is AI just replacing jobs? M the results there are varied at best and I think it's one of those areas where you kind of almost need to queue up the curb your enthusiasm theme song. >> Would you say it's an excuse? Sorry to interrupt you. Would you say then it's an excuse? So these companies then say to protect their profit margins to cut costs. Okay. AI software we've invested in it. Okay. We're going to start firing managers. So you think these S&P 500 the big tech companies are using the AI software and the investments more as an excuse to fire employees? I think that it's a blend. So I think that if you start seeing employees improve productivity, you can start trimming some fat and you should. And so there is some of that going on. But there's also still this overhang from all the overhiring that happened in tech from the you know 2020 through early 2022 period where there was talent hoarding. And so companies are still doing attrition programs like hey everyone's got to come back to the office. Why do they really do that? They really do that not to make team building more cohesive. Most of the time it's about getting people to quit. And when you get them to quit, you know, you don't have to pay them the the same kind of unemployment and the same kind of, you know, severance and etc., etc. So, it's like you're kind of going to save some money doing attrition that way. And on the other side of it, you know, when employees say, "Well, how am I going to be able to afford the commute and all this other stuff if I'm not close to the office?" and they'll say, "Oh, well, you just have to figure out how to be a team player." There usually aren't stipens given to people that are having to return to office at a lot of these companies. So, it's really just them saying, "Ah, either you're willing to eat your time and kind of your pay with the increase in cost for fuel and maintenance and otherwise, or you got to go look for a job elsewhere." Those are straight up attrition programs, I think, 90% of the time. And then the AI component here is another convenient excuse because you can say, "Hey, we laid off a bunch of people in sales or customer service or HR because we replaced them with AI." The truth might be more that a we've made productivity gains with AI and other technology too. There's a lot of other things that are going on in technology that allow people to be more productive and so we don't need as many people in our workforce. But I also think that some of these companies are looking at the economy realizing how concentrated it is and derisking a little bit from what they thought might be the growth trajectory to what they may now think is a more realistic trajectory going back to the three legs of that stool. I'm not the only person that looks at the economy that sees that. There's there's a lot of other people that use much more sophisticated models in the seauite particularly the CFO of these companies and say you know maybe we should pull back and be just a little more cautious in what we do and we can continue to be bombastic and aggressive in our plans and you know and be very kind of um enthusiastic about where we can spend and and and just kind of push back on any criticism. But we can push those plans further and further into the future, you know, and you see the circular deals that are happening with Open AI and AMD and Nvidia and Oracle and Soft Bank and the government and all these other companies and it kind of starts to paint a picture of well this all sounds grand on paper but where's the money for one and two with all the efficiency gains that we're making and let me go back to say again frontier corporate proprietary cloud AI models in the for first quarter of this year. Things you could only run on data center scale hardware that would cost five or six digits. You can now run on a decent computer that costs mid to high four digits. That's an incredible change that changes a lot of the dynamics as to how AI can take shape for business use cases, for personal use cases. And on the other side of it, you've got Open AI wanting to have people give them their IDs to validate whether they're not be able to be treated like an adult on the platform. They're going to be randomly I mean it's become a trend now just banning accounts for different situations, some of which are not warranted. So when you get banned, whether it's on Sora 2 or Open AI's chat GPT, you're banned from the whole platform and all of your data is forever gone unless you can appeal and win that appeal. So, do you want to give everything to platforms that are willing to just lock you out and in some cases you get no customer service and that's all of your history and maybe it's even company stuff you need or do you want to do that with a little bit more upfront investment but much better quality control and the ability to really have full continuity of data privacy and security in that as well. it just it's becoming an easier choice to say I'm going to look at the options that exist local if indeed I need this technology but going back to the curb your enthusiasm theme for one moment only about 5% of firms that have invested aggressively in generative AI are seeing productivity growth that is at a scale that makes those investments worthwhile. The majority see modest or negligible improvements. So productivity gains at a macro level as a result of AI are not the same as what's being advertised is what it can do. And that's the disconnect because the narrative is it can replace people. The reality is it can force multiply when used properly. And so you see companies, for example, Taco Bell. You would think replacing someone who takes orders at a drive-thru would be trivial, right? You just have something that recognizes their voice, can take an order, process it, and then, you know, be able to make sure that everyone in the kitchen gets all the stuff they need. So by the time the guy drives up, they're able to pick up their food and pay. Well, there's been situations where those systems have broken down just randomly. But there's also been situations where customers have been able to exploit them. One gentleman ordered over 8,000 bottles of water and created a whole fiasco for that ordering system. And again, the problem comes back to what is generative AI. And generative AI is not really artificial intelligence. It's more a sophisticated token prediction engine. The output is tokens. The input are treated as tokens. So when I tell an AI, you know, hey, I'd like you to give me X, Y, and Z information, it goes into its training data and it looks at it probabilistically and it connects the little bits of that puzzle to say, oh, well, this bit of data that the user said looks like it matches this bit of data with a high probability, so I'm going to return data that includes that. And this is a very simple version of how it generates a response. So what that tells you is it doesn't necessarily understand what you're telling it, nor does it understand what it's telling you. This then explains why we have problems like hallucination which Jensen Hong told us in the first quarter of 2024 would be solved by retrieval augmented generation. It wasn't and the hallucination problem is still in play. And so what happens is because AI doesn't necessarily in this current form generation you know this sort of generative AI these large language models um that use transformer technology this basically causes an issue where if you ask an AI let's say something it's not trained on or something where the training data doesn't create enough highquality probabilistic inferred relationships that it can answer with confidence, then it just will make something up. A lot of them don't have any kind of guard rails that are in place. And these would be external guardrails. They're not built into the model. They'd be built into their tools and other things that are monitoring your chat to be able to know in real time what it's telling you and whether it's even vaguely accurate. So, this creates a problem because it was kind of told to us, you don't need Google. You just ask the AI, but if it doesn't know, it just makes up an answer. >> Yeah. And that's bad especially if you have like a legal research one and it just makes up a fake case law. So if you're trying to replace parillegals or associate at a law firm and you're using like a new AI program, which is what a lot of the money is going into to try to replace human legal research, it could just create a case out of thin air that actually never existed. So I've seen instances in the last couple months where uh case research they printed up and brought it and the judge said, "What case is this? It doesn't exist." Yeah, exactly. So, there was a a lawyer I tweeted this out the other day. There was a lawyer caught using AI while explaining to the court why he used AI. So, the attorney not only submitted AI generated fake citations in a brief for his clients, but also included multiple new AI hallucinated citations and quotations in the process of opposing a motion for sanctions. I mean, it just it's it's incredible where it's used and why it's used and how vastly inappropriate those use cases are because an AI model isn't trained on case law unless it's specifically trained on case law. Like it maybe some of this stuff makes it in on the margins in the vast swath of training data that's fed to it. But unless this model is really just using some kind of tool to query case law or its model is trained on case law and in addition to that it has a tool to further query to get more information. It's just going to return anything and it has no idea what it's saying or if it's even vaguely realistic and related to what's actually happened in the court system. So that's one example. Medical >> so those are the guardrails you're talking about that need to be in place. There needs to be guardrails in place saying if you have to go and double check to make sure there's an actual case and otherwise you cannot you know type up that there's a case here that does you shouldn't be doing that then >> I mean you really shouldn't because the problem is you don't have so for example if you go to chat GPT or you go to perplexity or you go to uh you know anthropics cloud or any of these if they're not searching externally and perplexity to its credit does more of that external searching so it's generally better overall, but if you're not explicitly searching data sets that contain the information you want and just using the AI to take that information, what they do is they embed it. Right? So there's there's different types of models. One type of model is an embedding model. And what that does is it goes out takes a large volume of information and it optimizes it so that an AI um that's more meant for generative kind of interaction can read that data coherently and and really be able to pull from it so that you can query it. This doesn't completely fix hallucination by the way but it's a step up in the quality of the responses you can get. But the problem is that the vast majority of the models that exist out there in the commercial cloud are not really specialized for these use cases. They don't have these external tools plugged in and they don't make it clear to the end user that this isn't really just a general purpose ask it anything and it knows everything that's the compendium of humanity's knowledge. It just doesn't. And that's the problem. That's the disconnect. Again, the hype versus the marketing. The reality is if the model is a general purpose model, it might be good at creative writing. It might be good at certain kinds of reasoning and logic. If it's trained on math and science, it might be good at that. But once you get more out in the into the fringes and things that are very specific that it's not aware of implicitly in its training, if it's not looking at data and doing so with the right attention to detail, it's very likely to return inaccurate, if not very misleading results. Okay. So, there's a lot of bugs. Now, the um big tech companies, ma'am, they spent what $300 billion approximately. At least that was a projected capex budget, capital expenditure budgets for if you add up like the five or six largest tech companies. So, like Amazon, Google, Meta, Microsoft, um and a few of the others around $300 billion. That's not counting Europe, the European tech companies, the uh Chinese ones. What are they spending the money on? Is is it mostly on data centers? where where is the majority of the capital expenditure going that they're claiming has been spent already? >> It's going into a lot of these investments. I mean, I don't think that they're being inherently dishonest. I just don't think it's going to prove to have the level of return and growth that is being hoped for because there's this massive disconnect between the people that are making decisions and the people that are marketing these tools and where they're actually at in their evolution of uh technology. and it goes to the Gartner um sort of plateau of elevated expectations if you look at their hype cycle. That's where we are with generative AI right now in that we're kind of cresting at the top and starting to see it come down as a lot of the enterprises that have invested in generative AI aren't seeing the ROI. they're really kind of starting to pull back more like Taco Bell now all that order taking is supervised right and that attorney that's one of as you mentioned there's other cases there's been dozens and dozens of these cases around the country these examples exist throughout all kinds of different industries and it just kind of gives you a sense that yes there is some really cool stuff going on don't get me wrong there's if you have the right model and you know how to use it and you're able to feed it the data and the prompts that it needs and you supervise it properly, you can get more done. No doubt about it. That's true. The problem is that the way this stuff is marketed is you can just automate people out of their job or you can you know create these pipelines of agentic automation that allow you to automate customer service and billing and HR and all and it's you know can you make the jobs of people's that are working in those departments can you make those jobs easier for those people? Yes. Can you replace them writ large? No. So you're basically saying if I could summarize that these uh the AI the current version of the software is more of a productivity increasing software where you're getting substantially depending upon the industry or how uh buggy the software is you can get substantial productivity increases from your human employees but to say that you know they can fire all their uh customer service agents they can fire all their accountants they can fire fire all their computer programmers or parallegals we're way uh far away from there. >> Yeah. And there's been studies that have said that a lot of the firings that have been attributed to AI were not really because of AI. So, you know, it's a force multiplier. It's a valuable thing to look at for certain industries, but that's the key that you mentioned. It's very specific as to, you know, where it's going to be a force multiplier and where it's not. Like if your employees are m paid minimum wage like someone who takes orders at Taco Bell and a data center GPU for just like one of your locations costs about $8 to$10 an hour. Are you really saving that much money? Probably not. And then what about the scale of errors that can happen? But if you're say a $200, $300 an hour consultancy and you can force multiply with deep research and document analysis and analytics and other things, you know, then you can add more value for your clients and maybe for your employee and for the firm as a whole. So, it's really just it's specific to what you're using it for, how you're using it, and making sure your expectations are properly grounded and that you start realizing there is a learning curve here. You're not just going to, you know, ride right in and be able to get a 50% productivity gain. It's going to take some effort unless you're using tools that are already built to be easy to use, industry specific. Like you look at some of the tools that lawyers are getting now from uh Lex Machina and Lexus Nexus and otherwise, they're definitely adding value. They're probably allowing them to downsize some of the administrative work and parallegal work. But is it going to replace all those people as as a whole? Like you said, no. It's we're just not at that point. And I think we're further from that point than even some of the skeptics might realize because it's not generative AI from transformer large learning models that is going to become artificial general intelligence or now the new term it seems like AGI's almost been dropped super intelligence. This is not that. It's not it's stepping stone to that to be sure but it's not what's going to get us there. And if you're a graphic designer or you're good at making animation, you work in Hollywood for CGI, CGI shop, you're going to have huge productivity increases. You don't need as many human employees. The amount of rendering time it's cut like enormously. Look at the stuff that's come out from Google Vo3 and Sord 2. Yes, there's bugs. Yes, they're violating copyrights and IP, but the quality of the videos, I mean, like the the animation, the defa it's unbelievable. Go and look at some of the Pokemon videos, you know, with Pikachu cooking or or the turtle ones, the Squirtle. Uh some of those like are just insane in in full 8K. >> Yeah, it's it's remarkable. And I think for video, image, and to an extent music generation, generative AI has come a lot farther, but it has everything to what you it really has everything to do with what you just said. It's trained on other people's IP and without their consent and under the guise of fair use. Fair use very clearly stipulates that what you're doing is not for commercial purposes. And yet all of these models with the exception of the openweight models that are given away for free all these commercial models that are sold at subscriptions or by API query or tokens etc. those are for profit. So you cannot claim that siphoning up everyone's IP is fair use. And we're seeing lawsuits some of them successful some of them not so much. But it is a legitimate query and it's a legitimate quandry for everyone that was in the business of content creation in any form because now at any point a model could just vomit up exactly what you did and you may now have to challenge that company and whoever used it to have any vague resemblance of intellectual property rights that you thought you once enjoyed. And that may not prove to be anything more than expensive and fruitless litigation that sets you back on time and also, you know, obviously on money. So, it's like it it's presenting some >> ins too for Nintendo for a public relations nightmare if they go and sue a bunch of teenagers and college kids over, you know, taking like their old their old characters, their IP. So, Mario, the Mario Brothers, you know, flying the Millennium Falcon. if someone made like a, you know, minute video of the Mario Brothers like uh with Sor 2 flying the Millennium Falcon or Pokemon, you know, doing funny stuff or I think there was I think there was one a Mario Kart I think he he drove the Mario Kart away from a cop. >> Uh yeah, I know Nintendo is one of the most aggressive IP litigants in the entire video game industry. They are easily really one of the just the worst in that sense. And this is sad to say because I have so many fond memories of playing Nintendo various iterations of the systems over the years, but they do just sue teenagers and and older people and anyone. I mean, anyone that even like has something that even looks something like a Nintendo product or >> I mean it's viral marketing though, mayhem. They should instead is viral. It's free advertising for them. They should be they should be like promoting some of the best videos. they should be saying, you know what, we're going to use this for marketing of our new video game, our new video game system. We're going to resell the old video games. They're going to they should embrace it as free viral marketing. >> So, I don't disagree. And the problem comes down to sort of you know case law and expectations on IP law and the the basic understanding or principle is if these things haven't been authorized and these people are doing this stuff you know sort of uh outside of your company's purview then you have to litigate to protect what you have your trademark your patent your copyright or otherwise now is there a very powerful case to be made for modernizing IP law so it isn't so draconian uh that that you know because we haven't even really talked about digital copyright law for 25 years in a serious way in Congress since the digital millennium copyright act which is an abomination. So on the other side of it the law as it exists now is too aggressive but then you have companies that are just ignoring the law and just basically engaging in the largest pirating operation in human history. somewhere in the middle we have to meet, you know, and I think that's kind of where uh things need to go in terms of this discussion because it's not fair that open AI or Meta or Google or anyone else can just steal what I might have written or what I may have made on a video or otherwise and just use it and then have a model potentially put out the exact same thing and someone else could claim credit for it. That's kind of infuriating. But on the other side of it, where copyright law is right now where, like you said, like you know, Nintendo could just sue a teenager for making something that's purely innocent because they're showing appreciation for what they grew up on or what they enjoy. That's also wrong. So, we need to reconcile that. And I feel like with the extremes on either side, lawlessness in AI and overreach in the way the DMCA exists on paper, there's a meeting ground in the middle here somewhere. Yeah, the laws are complicated of the federal government, the government court system. Uh it's a mess. Uh the court system and laws right now. So I hope that uh the the teenagers and the college kids are not sued, but I think you said that they might. Uh I think if I was in Nintendo, I would embrace this and I would use this. Hey, that this video got millions of views for free. We're going to use this to promote our product and we're not going to sue the pants off uh you know, a college kid. That's good. So, I I will say this also though, Open AI at first when they released Sora 2, it was a free-for-all. And I think they did it on purpose without proper copyright safeguards to create the virality and the sort of videos that would come with it. That would be things that would be shocking or be seen as, oh, hey, look at there's an icon or like a cartoon or some famous person doing this, that, or the other. That's all been pulled back. And so now, if you even try to do those types of videos like we're talking about earlier, you can get banned. And if your Sora 2 account gets banned, so does your OpenAI account and your whole history. And they're not very good at customer response from what I've seen in some of the people that have gotten banned, many who claim they're completely innocent or did very mild infringement of the rules. And when you lose that access, you can't go back onto Open AI. If they recognize you're the same person, they might ban your other account, too. So, you know, you got to kind of uh look at it, I think, with that in mind that a lot of these companies are starting to get way more strict about, hey, we'll train on your IP, but we'll try not to replicate it one for one. Can they promise that? No. Because of how generative AI works. They can only add guard rails and hope that people aren't able to break them or inadvertently work around them. So, I want to ask you over the next couple years, we're starting to see announcements for initial public offerings for Unitry Robotics, which is China's top robotics manufacturer. Our listeners can go out and look at the videos of the robots. They're working at human factories already in China. There was an article that came out in the last couple days about Western senior executives, publicly traded companies, CEOs and senior executives at US and European companies, uh, going over to China and seeing like the humanoid robots and just shocked how how good and efficient they are. And that's with Unitry Robotics about to go IPO soon. Do you think that these robot companies are we going to see huge progress from the artificial intelligence software? Is that like an operating system from drones? Are we going to start to see profitability and enormous productivity increases and growth from the other the hardware side then of for robots and for drones and stuff because of the improvements in the software? >> I think I think there is room for that discussion. So if we come back to some of the stuff we've discussed with regards to generative AI, we've talked about it being a little less capable in some of the edge cases that we've seen people apply it to where it's not really what the model's created for or when they try to use general purpose models for things that are not really within its scope. But if you talk about robotics and you talk about how much compute power has involved, how much better battery efficiency has become and how much more articulate a lot of these robotic components have become and also much cheaper to produce at scale, it's all but inevitable that we're going to have robotics play a much bigger role in the global economy in the years ahead. And I think what unitry and what figure and Boston dynamics and others are doing is actually very impressive. And the role that generative AI can play in that and then future more sophisticated versions of AI which I believe are in the pipeline is going to be really in some of these jobs that are very repetitive and very predictable and patternistic. That is where there's a use case to replace humans at a larger scale. And the more articulate and reliable that robots are, the more likely it is you're going to see larger and larger adoption trends across industries with repetitive manual labor. That's going to be big in manufacturing. It could be big in agriculture. There's a number of industries, even eventually construction, where you could see this play a bigger role. And I think that's something that is a lot more of a practical thing to pay attention to in terms of job replacement than where generative AI is for the services industry where it's not really tangible and where it's not a lot of very patternistic and predictable workflow. >> Yeah. And then to add to your points there, Figure Robotics, which also has AI software programming, they have been running humanoid robots in a BMW factory, I think every day for five straight months already. And I don't think there's been that many problems with it. So they're they are making huge progress. And I think the Chinese robotics companies are way ahead of Figure. So Figure and Boston Dynamics are ahead with the advancements because they have robots that are very dex dexterous. They can do martial arts. They can do back flips. Uh the Tesla Optimus robots, I mean that test that came out in the last week or two, uh it was delayed with the responses. It was very very delayed. I mean, it had trouble even answering like basic questions and then it tried to do a martial arts or something and it was very stiff uh compared to the Chinese robots and um Figure and Boston Dynamics ones here in the United States. I think that this is an area where again the hype versus reality is a problem and it comes down to Tesla's I think overconfidence and their ability to scale when they don't have the core tech and we've seen this in the company at their self-driving and their robotics and to some degree their batteries as well where they're trying to use this as scaling up battery tech but primarily self-driving and robotics. It feels like they are falling behind and it's really disconcerting because you know they would be a company that could scale this stuff if they had the tech. They talked about their dojo chips making semis inhouse for some of their uh AI and they've abandoned that as well. Another area where a lot of innovation was required and they've kind of walked away. I don't know that they're going to walk away from full self-driving or robotics, but I think they could be relegated to obscurity if they don't start taking these innovation trends seriously. And I know this isn't going to be a popular point of view for ardent Tesla fans, but it may serve as a healthy reality check because like you said, if you look at the ability for some of these American and Chinese and other robotic companies that we're seeing, I think I've seen some pretty impressive stuff out of Poland as well. If we look at where they are versus where the uh Tesla robots are, it's just night and day. And so I think that's a big issue. And I think that it's something that, you know, if Optimus is going to catch up, they need to invest a lot more aggressively in the AI and in the the motors and the whole set of components that power their robots. And it just doesn't seem like even though they've come a way from where they were, it doesn't seem like they're evolving at the pace that we see. And the same is true of the full self-driving. We talked off mic about Whimo and this has been a company where this is like one that Google's invested aggressively in. and I believe it's a part of Google at this point. Um, and Whimo has LAR and cameras and other sensors. They've got a full stack that they've built on their own that shows much better safety and much better regulatory approval trends as a result. You know, you can use cameras that can only see the visible light spectrum that humans can see, but you're going to have situations where they don't work. And I know Musk has said, "Well, humans can drive with just their eyes. Why do we need LAR?" Well, I think that's because there are situations where cameras may not work, where they may not be able to see properly because of weather conditions like heavy snow, fog, heavy rain, uh, sun blinding the cameras and otherwise. And if you had LAR as a backup, you can see right through all of those conditions and make better decisions that may save the life of the driver, pedestrians, and other people in the cars nearby. >> And Whimo is more expensive right now. So the cost to manufacture these these cars or electric vehicles with Whimo integrated into them, but what they're running in five or six cities already. I think they're rolling out to Washington DC metro area next year in 2026 and a few other major cities on the east coast. Well, the thing about technology, Jason, that we both know and can appreciate over watching all this play out over the decades is as it scales, it becomes better, it becomes cheaper, and it becomes more efficient in terms of size, energy usage, you know, everything. And so I think if we look at what Whimo is right now as a prototype and consider that it's something that could be much more effectively integrated into a car if it was manufactured with it and then we scale that technology to say a 100,000 units. I believe that the cost the overhead it'll still be higher than what Tesla's hardware stack is because it is more sophisticated but the cost is going to be far less appreciable than what it is now. Add to it, LiDAR as a technology has gone down in price and improved in its reliability and quality marketkedly just over the last 5 years. Yeah, I agree. I think we'll see. I I was listening to interviews with what Mark Andre who's a big investor into all this stuff and he was talking about the the uh hardware components and a lot of the other components going down for humanoid robots. He's like the quality is improving enormously. The costs are coming down enormously. I think you said the costs are down almost 90% over the last 10 or 15 years because of uh efficiency and productivity gains with Chinese manufacturing. So, China has had a huge advantage with the humanoid robots and manufacturing. So, there's going to be beneficiaries here. I think the beneficiaries are companies in other industries that use the software, the humanoid robots. It's not necessarily going to be the open AIS, artificial intelligence software companies that are going to be the beneficiaries per se. I think it's going to be the other companies and other industries that adopt, you know, the drones, the humanoid robots, the software, and learn to become more efficient to lower their overhead costs and improve profit margins. >> I think so. And I think that really it's going to be a efficiency evolution across the board with all this new frontier tech. What we see time and time again through subsequent, you know, iterations of these improvements is that the technology scales. it becomes more costefficient. It becomes more well adapted and then subsequent generations improve on that and they become even less costly and even better at what they're doing. I think we'll see that in AI, in robotics, in full self-driving, in uh drone technology, in compute as a whole. And so it does call into question the scale of a lot of the proposed investment and what the ROI is going to be because either you're going to have the most massive data center operations on earth that are able to do what they're hoping to eventually be AGI or super intelligence or you're going to have one of the biggest busts with that much spend and really not the revolution to show for it. It also calls into question something else and that is is the general purpose GPU the right chip for the job because you have Google with their own TPUs. You have OpenAI starting to plan their own chips. You've got Microsoft a quarter or two away from making their own AI chips. You've got Broadcom and uh Qualcomm making their own AI chips. You've got Cerebras and other companies making large wafer AI chips. I mean, eventually you're probably going to have chips that are specialized to whatever workload they do, training or inference, vision models versus, you know, media generation versus text generation. And I think that's going to change the game a bit as well because right now it's not that costefficient to invest as a data center or an individual, although it's getting much more efficient. No doubt about it. probably a change by about, you know, a reduction in 80 or 90% over the last eight or nine months for what you can buy and what you get out of it versus where we started the year. But it's still not necessarily as practical as it could and should be. But if we know anything about technology, this is the best thing about it. It's inherently deflationary. It gets better and cheaper over time. And so I think >> for the hardware and the software, but I see the constraints, the bottleneck still being like, okay, we get way more efficient with the software mayhem. Okay, the hardware goes down in price. It improves for drones and robots, but then I see us overall using way more electricity. I see see real world power constraints still for electricity, for cooling systems. I just see that like that's going to be the huge bottleneck. >> I agree 100%. In the United States of America, if we don't start aggressively investing and upscaling our power infrastructure by any and all means necessary, we're going to run into some bottlenecks that could have economic consequences which would be very unpleasant because it could price a lot of people out from having the amount of power that they want to live comfortably and uh you know that could lead to social unrest that could lead to some degree of economic upheaval. It's just it's ultimately it's counterproductive where it is right now. The hype versus reality is a little bit dangerous if these investments are realized and if the power demand is realized as expected in the time frames that are being discussed. However, if it's realized over a much longer period of time with a more realistically smaller scale of investment and we grow our power infrastructure aggressively and I know you could stop me because I'm making too much sense and the world around us often doesn't. If those things happen, then I think we could realize a future where a lot of this is much more sustainable. >> Well, people are playing around with what chatbt on their phones. Like the average person is messing around with artificial intelligence software on their phone has even like a lot of senior citizens have have at least one software program AI on their phone. My parents do and they don't even know that much about it. But the average person doesn't understand that that on the back end the servers they're using way more electricity. So over time, we're going to have huge uh bottlenecks here in power constraints. We're going to have to fix this problem here. Otherwise, um we're not going to be able to compete with uh China, which has uh you know, can build what 9 or 10 nuclear power plants a year. They're installing enormous amounts of solar panels. It's just easier in China to build a lot of the uh infrastructure and a lot of the next generation uh electricity. You know, I think I saw a chart that speaks to that in a way uh that the internet took something like 20 plus years to reach 800 million users and chat GPT took just several years to reach 800 million users. So there is an adoption rate here that is breathtaking for even just general usage and a lot of people are using this as like their digital therapist or you know their counselor or you know kinds of use cases where quite frankly it's dangerous and if you're listening to this and using it for that right now I would just gently encourage you to please don't do that um because it's not going to understand what it's being told it's not going to understand what it's telling you and it's certainly not advice you can rely on. So, you know, th this stuff is very interesting. I think it's going to continue to get better and more efficient. And I think that the offset for the generative AI I spend as predicted is likely to be a combination of economic realities, limitations of current power and infrastructure in other places such as water and bandwidth. Um, and also it's going to be efficiency and how much more efficient this stuff is going to get just over the next year or two and how that may change the plans in scale and time frame. >> Well, we have uh XAI has built what a couple Colossus data center locations, the ones in Tennessee, the one uh new one in Mississippi, and then uh Stargate just the phase one of it in Abene, Texas. that's online now and they're scaling that up and they're announcing additional Stargate facilities in the Midwest. I think basically any country that has cheap natural gas is very very interested in building a large data center next to it. >> Well, yeah. I mean, I would say they should build a lot more in Texas because they're just flaring that stuff off because they got too much of it. I mean, there's times where Henry Hub goes negative in price. you should be parking as much uh you know AI and Bitcoin mining and whatever else you could use because that stuff's getting burnt one way or another so it doesn't really make a difference but at the end of the day yeah we we have a lot of energy capacity if we start using it wisely and we start building on it we should be in pretty good shape to scale whatever it is we want to invest in the question is going to become more and more if it doesn't show this ROI from companies that are investing in generative AI is do we need to build this? Is there the demand? And are the use cases realistic? And I think again there's a big mismatch here. So we're going to have to reconcile where the technology is on its own evolutionary curve versus where expectations are. And expectations are going to have to become more and more modest as the technology progresses, but not at the rate that people expect or have been told to believe. So, if you're warning people, and I agree with a lot of the warnings that you just said, and then Mark Zuckerberg, this founder and CEO of Meta, he's doing podcast interviews, and he's talking about basically like he's going to ignore his board of directors that that are giving him warnings and he wants to commit $600 billion, I think, of capital expenditure in addition to what they've already spent. He said by the end of 2028, what do you think he's going to actually use the money on? because I I know Meta is working on what these virtual reality integrated with AI for Rayban and and military usage. They have a a joint venture with Andrew uh for for military mass and then they have the the um commercial retail one that are sunglasses with AI and recording and social media uploads built in. What do you think he's going to if he does spend the 600 billion? What do you think's he going to spend it on then? >> Well, I think Meta has kind of a checkered history of innovation. They've had a lot of products that they've tried to build inhouse that haven't done well. They had a portal product for video calls that didn't really go anywhere. Their earlier generation of these meta glasses really didn't go anywhere. The current generation, we'll see. They did make that contract with Anderall and the US government recently to be a part of making those augmented reality glasses uh for military usage. We'll see, you know, how that tech goes. But I think that Meta is much better of a company that either copies or buys technology than builds in house. So, you know, in this case, it's been well, they just basically aqua hired and and uh and poached as many AI experts as they could from various different companies in their pursuit of super intelligence. Now, you might ask yourself, how well is that going? Well, I guess, you know, with Meta partnering with another company to make generative AI images and videos and things like that, not doing it in-house, that might give you a sense as to how well it's going. It's not going that well if they're having to partner with other companies to do things that you know Grock can do, ChatGpt can do, Google can do, Alibaba's Quen which is much smaller company than Meta can do. Um there's so much happening that Meta is behind on in AI. It's actually staggering. In fact, their openw weight models were so disappointing in their progress from the Llama 3 series on that they stopped releasing them entirely. And Mark talked about how this is more of just them turning around and focusing more on their in-house. It really was a sign that they had plateaued at that point in time and their ability to innovate. again reaching a familiar ceiling for a company that tends to succeed more from copying or buying than building. And so I will be really curious what they try to spend that money on. They've talked about building all these massive data centers, buying all this compute from Nvidia and AMD and may maybe even making their own chips as well. And I just am I'm very reticent to believe that that's going to be realized. We also remember not that long ago he was caught on hot mic with the president and what he said before that was when he was when he was asked by Trump how much are you going to spend on AI he said around 600 billion then on hot mic he said to the president I didn't know how much you wanted me to say and so I wonder how much of this is also just part of shaping the psychology of this AI boom narrative if we just keep talking about future dates and large numbers. How long can that be a part of driving the hype cycle? And obviously there's real world tangible investment being made, but you'll notice a kind of a bit of a cycle here and it's familiar. The biggest promises are made at dates that can easily be pushed forward over and over again. And that has been a phenomenon we'd seen from Elon Musk and Tesla that's becoming more of a theme. And it was a theme back in the metaverse days. It was a theme when virtual reality and augmented reality were all the hype in 2021. And it's sort of that same hype cycle now where there's all this talk about investment and we're going to build all this exciting stuff, but hype is so far exceeded reality that these pie in the sky projections may be no more realistic than what we're being told generative AI can do versus what it can really do. >> Okay. So you if I could summarize what you said over the last couple of minutes, you expect then that in the next couple quarters maybe that the companies will start to push back into future years the uh amount of capital expenditure. So instead of saying like a hundred billion is going to be spent the next year or two, it'll be over 5 years. So, they're going to start um spreading it out and then or you'll start to see announcements maybe where um saying they're going to spend $80 billion cuz a lot of these tech companies for our listeners out there, they were saying that they were going to spend between 60 billion and like $80 billion a year going forward that we're going to start to see those budgets, those capital expenditure announcements slashed enormously going forward in the not too distant future. I think it's becoming increasingly likely, especially if we start running into economic hiccups that threaten government deficit spend, the affluent consumer, the overall trends of markets and asset prices. You know, all of this stuff right where it is now is perfect to just keep making projections and spending some, but not necessarily doing it all in the here and now. kind of pushing that forward but keeping the hype cycle going, keeping the narrative going. All the circular deals that keep the bid under a lot of these stocks. But, you know, those circular deals aren't much more realistic than the AI capex if you look at what's behind their structures and how long of a time frame they have to realize them. And if you recall from the first Trump administration, there were enormous deals like Foxcon coming to the US and building these large factories and other big announcements from tech companies that never materialized. So, how much of this is part of the hype cycle to kind of keep us positive on the economy and on investing and how much is actually real? And I would argue it's a blend, but I think that the more aggressive tone that we've heard about AI spend is more hype than substance unless we see some groundbreaking technological improvements that can ameliorate the core problems that stop companies from adopting it at scale. reliability, ease of use, being able to use these things securely, having some confidence about security and privacy, any of these kinds of things, they're still yet to be addressed by any of the corporate cloud-driven providers where all this investment is happening. So, I think there's there's still a big disconnect. I also think what's happening in China merits our attention. China is now making more openweight AI models than any other country. They are on the bleeding edge. If you look at ZAI, this is a relatively unknown company, especially months ago. I've been watching them for, you know, since they started their GLM model series at version 4, and it was an extraordinarily impressive model. Now they're up to 4.6, six and they're running neck andneck with Claude, Gemini Pro, GPT5 and this is an openweight model that you can download and run if you have the resources or you can use their inference which only costs about or starts at about $3 a month fraction of any of the US companies. Now, you want to tr, you know, trust a big Chinese company or midsize Chinese company with your data that might have ties to the government. Probably not. But you can run this on different clouds, too. You're not tied to their infrastructure. You could run this on your own private infrastructure or you could run it on shared infrastructure on a US cloud like a coreweave or an Amazon or a Microsoft or whatever you wanted. The point being the idea that only US companies can produce superior AI models is not durable. So the other risk to this narrative in the capex is that the you know genie is out of the bottle. There's all kinds of very smart people all over the world but particularly in China and from China. If you look at the US AI expert composition as well that are leaders in this space and there's no way to make this something that stays proprietary behind a subscription wall etc. It's going to be a lot like Linux or otherwise where there's going to be a fair amount of usage that's on these openw weight or even open- source models that you either do inference locally or in other cloud providers that aren't necessarily part of the big tech narrative or if they are they aren't necessarily making the same margins because it's not part of you know the subscription models where there's better margins. You're just doing bare metal inference building your own stack. So the point of this is we've got to temper our expectations. What's been projected versus what's realistic has a lot of left tail risks that have yet to really enter the narrative meaningfully. And the last time they really did was back when deepseek came out in the first quarter of this year and rattled markets because investors started to realize that maybe the west doesn't have a durable edge. That is something we still need to reconcile. And what DeepSseek was open source or mostly open source. So Perplexity, you know, like a week or two, they had the DeepSseek version on their own servers for protected data. So you didn't have to go to the Chinese tech company and deal with them there. They had what the open- source version of it on the Perplexity one, the American software company. But overall, what this is deflationary, I see this as deflationary. I don't see the software companies necessarily being the winners. If there is Jeban's paradox and productivity gains and the prices do fall, we're going to end up still using a lot more electricity, a lot more cooling systems. So I see the infrastructure companies, the electricity companies, the uranium producers, the nuclear power, the utility companies. I see a lot of the natural gas, the cheaper producers, they can negotiate what a data center deal around there. So I see those being potentially the beneficiaries in all this. I can't I'm not a tech expert like you. I can't tell you which tech company is going to win in the software or the hardware for the robots or the drones, but I could tell you that, you know, if these things do play out uh huge improvements, we're going to see an enormous amount more electricity and energy usage for the electricity for what the drones, the the robots, the data centers, and then also for uh the cooling systems. >> Yeah, I think it'll it'll be a mixed bag. I think it'll go steadier and slower than what's projected. But, you know, we'll also see big efficiency gains, some pairing back, some pushing into the future of the capex and and also I think we'll see a lot more emphasis on building out power infrastructure of different kinds. I think you're right, the power usage is only going to go up from here after decades of increasing efficiency and flatlining going down. We've turned the corner and now we're going to see that go the other way. But thankfully, we're also sort of the Saudi Arabia of natural gas. And since that's our largest source of power generation, we at least have something we can tap into that gives us a bit of an edge there. So, you know, it's a mixed picture, but overall, I think long-term, the future in this stuff is very bright. And I think there's a lot of transcendent trends that are taking shape that will play out over the decades to come. I just think that near-term we're a little too excited about the progress we've made and talking about it as if we've made uh milestones that don't yet exist or haven't been achieved. And that's sort of what's being advertised and that's what's creating a lot of buzz and excitement that again it's it is exciting. What is happening right now is extremely exciting. It's just not the sort of transformative version of AI that people think it is. Well, over the last couple of months, I think the Trump administration uh President Trump and and the other people that go on the business TV and the other talk uh the political talk shows uh political news, all the different cable channels, they basically when Trump got into the White House administration there uh after January in February and March for the first, I don't know, 6 weeks, 8 weeks, they basically said that they don't care about the stock market. The tone has changed a lot. I mean, he talks up the stock market now. He talks about investing a a lot and honestly like the gold prices just outperformed. I mean, you didn't have to do that much over the last 10, 15, 20 years. If you just steadily accumulated gold, you would have at least kept up with the S&P 500 outperformed it now. So, it's just funny that like Wall Street is talking about, oh, we're good at picking stocks. And look, the gold price has just outperformed a lot of the companies over the long term. >> Absolutely. And some of the companies have outperformed the price of gold as well. talked about Rio2 over the last couple of interviews we've done is one of the junior miners that I've liked a lot and it's already up 3x from February. So, there's a lot of opportunities that continue to emerge. I do think gold it would be healthier to see it consolidate here. It's going to do whatever it wants to do, but it's it would be healthier for its trend structure to consolidate some of these gains so it can build some support for any, you know, potential corrections to come. anytime precious metals go parabolic, it does not have a very good ending. We're not 100% of the way there yet with positioning and otherwise, but I think we're closer to that point. So, I'd love to see just a few months a sideways trading, maybe even a corrective impulse and then a continuation of the move higher so that it could be something that's a little bit more sustainable versus the past kind of parabolic rallies where you just you keep, you know, just rocketing higher and then you have a really sharp bloody knife on the way down. Well, I don't know if we're going to see huge corrections in the gold price um that last a long time at this point because I think the governments are whether it's Japan, the European Union, with the French government debt, there's just so many uh problems with government finances that I could see a lot of these governments wanting to, you know, debase their currency at an even faster pace and monetize even more government debt. We're basically kind of at the end of the road where the math just doesn't work for a lot of these government finances. And the way that they're going to get out, they think that they're going to get out, is with some type of currency debasement or debt monetization program. >> Yeah. And I think over the intermediate to long term, that's 100% going to be the case. I think the risk for gold is outside of it getting overheated and maybe correcting a bit is also that if you have an actual global economic recession and people start deleveraging, they're going to be selling gold, too. And that happens every time in the initial part of the recession, the first period, you know, whether it's months or or weeks or even, you know, a quarter, you typically have gold as a beneficiary of that increasing uncertainty. But as it drags on, should it drag on, typically gold gets smashed pretty hard when you start to see uh deleveraging across multiple asset classes. Now, to me, any real material weakness in gold is a buying opportunity. So to be clear, I still have a very optimistic view on where gold can go over the intermediate to long term. And I think dips should be used as opportunities to buy, particularly more meaningful ones. But I do think that we just have to be cognizant that it's still a financial asset. And so it's still to some degree when there's these really material events where people are deleveraging, there is some risk there. And there's also risks if it just, you know, you just see it go all the way up to 5,000 by the end of the month or like early next month. A move like that is probably going to lead to a 10 or 20% move down following. >> Well, I mean, the the selling of gold that's already I would argue that's occurred over the last 5 years here in the United States. A lot of uh retail investors, middle class people, they've been huge net sellers um of cash for gold. pawn shop brokers, if you talk to them, there was a lot of selling of of gold and silver, coins, bars, jewelry over the last 5 years. Americans sold it because they lost their job or a financial emergency. There was a lot of that. So, uh they have not been huge uh buyers. A lot of Americans, not until really recently, but over the last 3 to 5 years, mayhem, there just has not been a lot of buying for American retail investors for physical gold and silver. They've been net sellers. Well, also if you just look at the price of gold and silver from their top in 2011 to 2013 to where they went before say 2020 2021, there was no reason to want to hold it for a lot of people. They bought it thought, you know, back then a lot of people were buying gold and silver at or near their tops and then they, you know, they bought it, the price got absolutely destroyed. It didn't recover for, you know, basically a decade. So there was a lot of people selling because they just got their souls crushed from the amount of non-performance while they watched stocks go up and up while they like you said experienced economic hardships while Bitcoin added competition. There are a lot of things that pulled capital away from those areas. So I agree but I would say on the other side of it a lot of the price movement that happens in the futures and in um you know what sets the spot price of gold in London which is the biggest area of the world where gold trades doesn't really have to do with the whims of retail. It has a lot to do with these global money flows, larger uh trading desks, some of the sovereigns, and really just sometimes you have forced deleveraging. So, it's not I'm not bearish at all on gold to be clear. I just feel like if you do have an economic disruption and things get a little deflationary across asset prices, everything kind of gets to a correlation of one at some point. So, it's kind of a margin call and you think that gold could be one of the negative uh not a beneficiary of the potential global margin call. Say if like some of these banks like Jeffre Bank or some of these other banks start to fail with some of their bad loans or something and then there's a global margin call, not quite Leman Brothers necessarily, but um then that gold would be one of the things sold in in that scenario though. I think like some of the gold companies that have enormous free cash flow, huge profit margins. I think there would be acquisitions cuz some of the the mid-tier precious metal royalty and streaming companies like Sandstorm Gold, that deal is about to go through next week. I think uh Royal Gold, like the acquisition will go through. So, they'll be part of Royal Gold. But the other two mid-tier ones, so like Triple Flag Royalty and O Cisco Gold Royalty, they just said on their conference calls that like they're being contacted and they're open to either a merger uh like a relatively 50/50 merger or some type of that scenario or being acquired. So, I mean, we are starting to see I think some of the larger companies are just waiting for that dip that you talked about. They have the cash, they don't have debt, they have record profit margins and free cash flow. They want to go buy stuff now. >> Yeah, I agree. I think you'll see a lot more M&A in that space especially if you see some de-risking. >> But uh as we wrap up here I think the the biggest contrarian play and you know a lot of the stuff we talked about with tech I think the underlying theme here is energy. So we've seen a huge move when in the uranium uh companies like Kamico and um you know the investments for the electrical utility companies they're planning like huge um investments into infrastructure. The the sector that's the most out of favor with valuations right now is oil and natural gas. A lot of the oil producers have an enormous amount of natural gas. I think that sector for riskreward is a contrarian value play because if those things play out with the tech, I think we're going to use a lot more natural gas. Uh I don't think the market has the right projections for how much natural gas we could be using once those data centers are built and up online. Yeah, I think that eventually there's going to be a lot more demand for energy across the margins because of a lot of the trends that are going to expand usage that are going to become more durable as time goes on. I mean eventually technology is going to play a much greater role in our lives than it does even today and technology consumes energy and even if efficiency trends are going all in the right direction to reduce that usage, the the other side of it is we're still using more and more and more technology. So ultimately, yes, I think energy will be a beneficiary. I just think it's going to take some time for that to materialize. There is still some slack here. And I would really uh ultimately like to to be more confident, like to see, you know, some of the the little guys, the the working class and the middle class, you know, have a fairer shake in the economy so that they can start to realize they're part of the American dream, particularly the younger generations that really haven't had a chance because that's the other side of the ledger. you know, if Gen Z doesn't want to have a family, doesn't want to buy a home, you know, that's one of the largest contingents of energy demand is from family formation and, you know, buying a home and having all that square footage to heat and cool and light and, you know, and entertain yourself in. If that just ends up being people living in like, you know, apartment-ized spaces and not having families and otherwise, that reduces energy usage in aggregate. So, I'm not saying that offsets all the tech, but it's a trend we have to consider because there are alarming economic implications if it becomes a multigenerational theme. >> Yeah. I mean, the people that are the tech company CEOs, the manufacturing companies, I mean, uh, Trump, the administration on the TV interviews, they're talking about basically like the capital's coming back, the investment's coming back when he announces a deal. But when they announce the factories, they're not saying that it's going to hire a lot of humans. They're basically saying that a lot of the factories, they even admit this, are going to be automated. So, we're going to have like those robots in there. We're going to be using a lot more energy and electricity. So, not necessarily good for a lot of uh highpaying jobs for humans. But that that's I mean if you're going to bring manufacturing back to a country where regulatory bodies are even with some of the reduction that's happened are still going to be more critical of their supervision more I would say like impose more overhead on companies um margins then that's going to be one facet. the other facet, you know, and you've got things like OSHA. You've got other kinds of supervision that just increase overhead. They're great for safety. They're not great for profit margins. And then you have just higher in aggregate labor costs than other countries where you want to bring manufacturing from like China or Malaysia or Vietnam or whatever. That means that the only option you really have is to use robots because there's no regulator that's going to be as concerned about supervising the safety or the health of robots or looking at repetitive stress or injuries on the job. They're a robot. You just repair them and then you know the labor cost is basically you buy them or rent them and then you do whatever maintenance is involved and that's it. And in aggregate that ends up being cheaper than having a lot of humans. So it it's just businesses doing what businesses do. We outsource for the same reason. I love this whole thing. We're getting so mad at China and I get why to an extent, but where did the outsourcing revolution really come from? It came from investors wanting better returns and seuite saying, "Well, the best thing we can do is just start pushing our manufacturing overseas to give our investors the returns that they want and pay ourselves more." So, we'll fire a lot of Americans. we'll shift these jobs to Mexico or to China or to any other country where we can do the work more cheaply. And that really talks about the disparity between what consumers expected. They wanted cheaper products. You know, they wanted products that didn't rise at the rate of inflation, at the rate of debasement. So outsourcing became a way to induce disinflation, but also was investors wanting better returns. And so we kind of did this to ourselves. It wasn't just China, you know, kind of taking advantage of us. It was what the country in many cases without really understanding what the consequences would be what they wanted. The investing class wanted returns, consumers wanted affordable goods. The only choice ultimately came to moving them overseas. And similarly, when we move them back, the only choice is going to be to do a lot of that with robots. >> Well, I agree that um we're that the uh decisions that were made were hurt a lot of the American workers. But I think what a lot of people didn't see was that um when you they moved their factories there that a lot of the uh good products so whether that's an Instapot the electric pressure cooker. So if you the people on the Shark Tank are talking about this all the Shark Tank investors that like uh there was all these good business ideas and they invested in them. They moved the manufacturing to China and within a couple years if if the product takes off and sells well in Amazon or in retail stores, the Chinese have what half a dozen or a dozen copycats of it at a fraction of the price and basically the company is uh defunct then potentially bankrupt. >> Yeah. >> So that's that's like the major issue there that I think a lot of people didn't see. So they saw the stuff you said with the cheaper labor and that like look, we don't want to pay the United Auto Workers cuz the United Auto Workers don't do a good job and they demand higher wages all the time. Uh my friend opened up the General Motors factory in Beijing. He's moved back here in the United States cuz he's American even though he speaks fluent Mandarin. And he basically said that like the United Auto Workers do not want those humanoid robots there. They're going to be fighting it every single way possible. You're seeing the same thing what with the Screen Actors Guild in Hollywood. They want like anti-AI stuff put into legislation. >> Yeah. And the other risk of going to China was all the IP transfer that was forced by companies where if they manufactured there for 5 or 10 years, China ends up inheriting all that IP. So they don't even need the company and they make copycat versions of their products and you know can sell them at a fraction of the price and the companies were just assuming they would have subsequent iterations of those products that would make that a negligible hit. But the reality is that the trend of innovation at a lot of those companies wasn't such that that that was material enough to offset the risk of IP transfer. So there's a lot of things wrong with outsourcing. I'm not a huge fan of it myself, but I always get a little bit amused by the arguments that we have about it with some folks that kind of try to blame it on a certain country or a certain population because it's like, well, there are a lot of forces behind this and a lot of them were at home. Like the vast majority of the reasons these things happened were were because of decisions that were made for short-term economic exigency for certain groups of people. So, you know, we can we can try to reconcile and we should we should become more self uh sufficient. we should still engage with the global economy but do so in a more uh self- sustainable kind of way and maybe not think that just in time supply chains are something we should rely a whole lot on when it's a global supply chain either but that's a whole other conversation I think ultimately though in the near term cooler heads will prevail she's under pressure in China the next plenary may even threaten his reign there uh there are a lot of rumors circulating about what could be the result and I think ultimately he's not in a position to basically tank his economy which is export reliant plant right now. So, he's going to have to turn them out. And Trump, similarly, the United States isn't going to be able to withstand 100% tariffs on Chinese products for any long duration of time. So, I think cooler heads will prevail there. I think the government shutdown will probably be resolved in the next two or three weeks. So, I think some of the the things that are near-term aren't as big. I think it's the more existential risks that exist in these narratives that are driving the hype versus reality disconnect. >> Yeah. And I think the 100% tariffs, I think that's a, you know, negotiating tactic for him to to work down from the 100% down to something else to bring them to the table faster. I mean, if the US had had better planning, we would have already negotiated a deal with the Canadian magnet company, Neo Neo, uh, Neoperformance Materials to get a uh, magnet factory here in the United States. That should have already been done. But unfortunately, you know, a lot of these things in the US are not done smartly or efficiently. >> Unfortunately, not. And you know when you have the four quarter a year reporting cycle that can create a lot of foolish decisions to be more exigent for the current quarter than for the you know trend of innovation at the company or the decision-m or planning. Similarly when you have a two-year political cycle it creates a lot of that kind of problem as well in the government in its planning. And so as sad as it is to say, you know, having a situation like China where it's basically a uni party, they can make a plan and stick to it. And it's not always the best plan. In some cases, it's the worst plan you could ever imagine, like the one child policy. But ultimately, if you do make plans like building infrastructure, you can stick to them for longer periods, it'd be more consistent. It isn't this back and forth shuffle between, you know, two sides of the same coin, granted, as it is in the United States, uh, where both parties largely serve the same financial elites, but there's enough of a disconnect that it creates a lot of kick the can, short-term oriented decision-m in our political system and and unfortunately in the seauite of publicly traded companies. So, you know, there's a lot to unpack with that. I think that, you know, it it requires a bigger conversation about how we govern our country and our companies, but that's something that's going to require, I think, some more study and some more um just creative approach to what now has been a system that's probably a source of more of the issues that we're experiencing than the solutions we need. >> And it's going to be tough to fix things properly for a smoother long term when the US national debt is rapidly growing. We're down last year from $1 trillion added every hundred days down to I think about 80 days now for every trillion. It's just getting out of control >> completely. >> Well, uh, thank you so much for your time today, Mayhem. Please tell my listeners more about Macrovisor and Trader Aid. >> Yeah, absolutely. Thanks for the chance. So, at Macrovisor, we cover the global markets where we find trends of market performance that meets the macro that we're analyzing. So, we look for the combination of the two to identify longerterm trade and investment setups. And then at Trader Aid, we help traders sharpen their trading edge with our Discord community that has exclusive tools for analyzing options flow, our S&P options visualizer that you can plug into Bookmap and Motive Wave to analyze flows and levels that are moving the markets, as well as my one-on-one coaching services where I can help you become a better trader or investor. You can learn more about Macrovisis at macrovisisor.com and Trader at traderade.com.

Mayhem: Big Tech Is Over Hyping AI? AI, Robots, Drones & Data Center Boom As Gold Prices Gap Up

Summary

Transcript

Join for free and get access to the buyside

Get insights from the world's most influential hedge fund investors direct to your inbox

Follow Us On: