Transcription
All right, welcome back to everyone who's back from the holidays. Uh, it's a good time to be uh, watching this week. There's a lot going on and I definitely have a uh a lot of slides to go through this week relative to what I expect to be both the uh investment side for this year, but also as you can see on on number five, I'm very focused on there being a, you know, a solid 20% correction at some point.
Even if it's not for the broader market, um I definitely think we're going to have uh episodes of some risk this year that will be outside what everyone's looking for. It is not related to an AI bubble, but it is related to delays that I'm expecting in the data center side that are actually real. Uh it won't slow down the AI compute and the AI adoption. So you're going to get a transition in the market that to me is is a combination of a a change from momentum over the last three years relative to the chat GPT uh launch, but at the same time a 15-year shift related to software to hardware.
So uh if I'm right, there's a lot of opportunities to make money. There's also a lot of opportunities this year to lose money and to get caught in positions that are just going against you continuously. Uh start of the year just to get going. For people that are bearish, uh I fully expect there to be 15% earnings growth in the S&P 500. I fully expect the S&P to at least be up 15%. On the back of that, I expect the economy to continue to show the productivity gains that showed up in the last two quarters. uh which means this is all about the AI productivity gains with inside the market. The replacement of people not firings but no more hiring relative to nominal GDP growing relative to revenues growing which means profit margins go but this is the year of the adopters. This is a broadening out. This is a PMI sensitive uh story where PMIs are rising. Uh if you're going to call me every week and say the PMI is not above 50 yet, uh you're wrong. go for it.
Um, making money in the markets is not about telling me what's news. It's about telling me what's going to change over time. I'll go through a lot of those things, but everything has inflected since, for me, July, when I wrote my first piece on the PMIs. Uh, I use the markets as a tell for what's happening. And all you can see here is in the beginning part of the year, we've got green across the board, whether it's in local currency or in foreign currencies, every single part of the world is up to start the year, which finishes what happened last year. When you go through the sectors that have led, the S&P is up uh a little under 2% so far. It's all the things sensitive to the economy. Materials, consumer discretionary, industrials, energy leading the way. Uh the safety names towards the bottom and tech unchanged in the S&P 500. That will be a major theme for this year. That is part of the rotation. Small caps leading the way. Small cap, small caps are going to be a dominant theme for me this year. Midcaps, I expect the Russell 2000 to be up over 50% this year. Uh with the S&P up at least 15%. And again, basic materials, industrials, you've already got double digits there. Tech led by semiconductors, X Nvidia, X Broadcom. Again, part of this dramatic shift away from the last three years. And consumer discretionary as well. MCI World X the US. Again, we broke out last year effectively of a John Rog 18-year basing uh formation. uh a big base and we broke out last year. This is not ending right now. And when you go through the sectors of the world, every single sector in the large cap midcap is up as well. Banks leading the way. You don't have bare markets with banks making all-time highs. Transports breaking out finally out of a base part of the PMI special. I've been talking about transports. You're going to see this a lot this year in terms of the transport side. Everyone trying to pick a bottom in software. Don't waste your time. You might be able to catch a little bounce in it. But here's the white line. Equal weight software and equal weight semis on top and the S&P there. This is a hardware situation. This is the end of software. Uh I'll say it again and again. Software is now ubiquitous. The competition will be never ending. I can build apps. I have I will show you. I've showed you last week. I will show you more.
Um, here's IWM. Here is a 50% year coming out of the last time PMIs were depressed for a long period of time and the last time before we got a commodity cycle that I believe this is the first true PMI business cycle based on massive demand for things as opposed to re uh rein uh bringing inventories back up coming out of fears which was the story from the great financial crisis all the way through COVID. Um, everything was a rebuild. There was never an expansion of any kind of uh breath. This one will be as we have to put intelligence in everything. Uh I'm going to keep showing this until the pay wall is launched and then the people who are paying for it get it. But I built an immune system which is a turbulence model has a 100 assets in it dominated by AI looking for signs of stress with inside the market measuring it through correlations and volatility. So a coariance matrix nothing. It's completely healthy right now. um we did get kind of one line where we had during so far this um thing you'll get episodic like thin lines like this. What you're looking for is multiple days of seeing correlations and volatility change and then for bottoming is when you kind of get the pukes of these big red lines. So the turbulence model is saying everything is fine. Nothing to worry about right there in the immune system of the market.
Uh Goldman Sachs put this out again. you have to make your calls. But if you're just going to use the numbers and you go, "What happens to multiples in the S&P? If we've got a stable GDP environment, which is fully what I expect, I actually think GDP will continue to surprise the upside." I think inflation, although it'll bottom this year in terms of somewhere a little lower than where it is right right now in the first quarter, it'll start to head higher, but only after energy prices start to move higher. If they don't move higher, we're not going to see that. Um, and then also in an environment where the Fed is cutting rates, you have both of those going there. They don't happen too often, but that's when you get the best side of multiple expansion.
Uh, in terms of the payroll numbers for the week, just look at this red area. So, instead of just looking at the number which came in at 50,000, we had 56 last month, we had minus 173 in October. When you add up just the last four month, I mean, sorry, the last five months, you have zero. You go back and do this. We've created in the last 8 months less than 100,000 jobs. Now, the only part that's green is effectively education, healthcare, and leisure and hospitality. If you strip these out, you're dealing with negative numbers for a long time. Now, this is ex those two sectors, we are at minus 71,000 on a five-month average. So, we are in recession territory for the bulk of the economy from a jobs perspective. But again, in my opinion, this is all related to AI. This is related to AI driving the stock market higher, allowing more people to leave the labor force part labor force completely. Uh whether it's a two household, someone just gives in or they turn 50, they lose their job or they're getting pressured and they take a package. It doesn't really matter. They've created enough wealth uh via their house and via this the benefit of QE. But this is really the benefit of having AI an exponential innovation that started after the great financial crisis. So ADP confirmed the fact that we have no job creation. So it's not just that. And if you look at the jolts numbers, we have finally now have 685,000 more unemployed people than job openings in the US. This is going to be a continuation because this is the year that adoption picks up, meaning digital employees pick up dramatically. I'll talk more about this, but this is the main story is that you were at the bottom of adoption. So in the same way that when Chad GBT came out, we were at just starting the data center buildout, we are at a new investment theme that will forever only go one direction. Productivity will only go one direction from this point forward because we're not just dealing with AI agents. It eventually gets to humanoids. And as you go through this, you're replacing human beings that are very expensive with computers that work 24 hours a day. So per hour on everything. And if you go through productivity numbers, you're going to see per hour. When you get the GD, the employment numbers, what we normally care about is the wage component and the amount of jobs. We don't spend time on the hours worked. But if all of a sudden we go from 34 hours work to just multiply 7 days a week, 24 hours, you've got a massive productivity boom, which is coming, it's not a guess. It is a certainty that we will have productivity gains at the bottom that will continue. So for people trying to make the argument that this isn't going to get better, this is the productivity over the last two uh quarters. And again, you can see per hour, 24-hour workers per hour. How much work is going to get done when people are not working a 40-hour work week, but they're working a 100 change hour work week. So, this is all going to happen and the cost of those people goes down. So, productivity is going to only go up from here regardless of what your views may be. If you haven't thought about it in those simplistic terms, you will.
Here's what's happened in productivity or at least nominal GDP. nominal GDP which absolutely does not purely capture uh the intangibles of AI. This is going to be a bigger issue as the velocity of money increases with stable coins and with tokenization. I'm not going to go through that today, but if you haven't thought about that, give 22V a call. I'm happy to come in and do what I'm doing for many many places at this point, which is preparing them not just for the AI productivity side and showing them how to incorporate in their business, but also the changes that are coming into measurements that are no longer relevant like GDP. So you can sit there and pay attention, but the reality is nominal GDP just grew to 8% in an annualized basis in the prior quarter. Look at how many times that happened this century outside of the postcoavid bounce. You're dealing with the largest annualized nominal GDP. U Atlanta Fed GDP now and I should remind you all that is during a time period where everyone was worried about the tariffs causing a recession. So, not only is the data saying something different is going on with AI, intelligent people are having a really hard time thinking about how AI is incorporating itself into the market. You cannot capture it. go back and read the Allen Gr the Greenspan pieces back around 1999 on why we can't cover intangibles in this measurement which was not created for it and why this will continue to be a situation where profit margins grow stocks reflect the productivity GDP will start to reflect it now because of what I showed which is there is a relationship between how many hours worked for the job for the work that you're getting and the transactions that are happening Atlanta Fed GDP now for Q4 is currently at 5%. That has some noise in it because of the trade situation, but the reality is currently we're five. We'll see where it ends up. We got the ISM services number. The only reason I want you to look at this again, if you look at the granular level, the new orders component went up to the highest level since 24. So, you're starting to get the new orders coming in. On the PMI side, we didn't get the ISM PMI to break above 50, but again, we've been above 50 for almost a year now on the global one. The ISM survey, which is more dependent on the foreign side and what the tariffs did, uh people, it seems like corporations still haven't adjusted. That'll all change as we move forward. So again, if you're focused on the fact that the PMIs are not there, what matters is what the market sees. The market always gets ahead.
This was my outlook piece for this year, the physical world upgrade. Why AI's next chapter demands a hardware renaissance. There are many, many trade ideas. If you guys did not get it, I'm going to try to do what I did on and I'll I'll reference this as we go on. Uh, but the PMI story is a massive story because this is not a one quarter thing. This is not a one-year thing. We will be building more cars, more phones, more appliances, more computers that all have intelligence in it. I'll talk more about that. But if you've read my pieces on edge devices, on NPUs, the Nvidia Grock deal, if you're not doing your homework on this and you're just sitting there reading research reports from cellside research people talking about what the economy is going to be without understanding AI, it's going to be a very challenging year to answer the question as to what's going on. I want to remind people, I wrote this piece in July 21st. The reason I'm bringing it up now in terms of PMI, paranoia motiv motivates investment. This is the most critical part because this was the best week to use as an example of this paranoia and why this drives the need for a race for commodities, a race for hardware, a race for chips, a race for countries. I believe the US manufacturing PMIS is poised to move significantly higher driven not by traditional demand cycles around real estate and autos but by a new and powerful force paranoia. governments and corporations. And that's the key thing. The environment of uncertainty and strategic anxiety is now driving investment at a scale and intensity that I believe will be showing up in the manufacturing data through a long overdue rise in PMIs. And soon, this paranoia won't just be confined to the boardrooms or defense departments. It will spread to the workforce as employees begin to grasp how rapidly AI is reshaping the nature of work and the security of their jobs. That is exactly what's going on later in the piece. The underlying shift in PMI should not be interpreted as merely whatever. Begin to recognize that this is not about a near-term stimulus or post-pandemic normalization. That's the inventory rebuild story that I'm saying, but instead a longduration investment cycle tied to AI, military modernization. We saw Trump say he's going to increase the military budget. We've seen this go on in Europe in terms of the breakdown of NATO. And the path to AGI will force corporations to continue to spend money on data centers regardless of any bottlenecks that appear. And regardless of the price that happens for the inputs, which I will talk about, this is the beginning of a secular shift. As Mark Andre put it, this shift is about soft isn't about software eating the world. It's about machines eating the world. Could not agree more. I have a big part on Mark Andre, an interview we did this week.
Um, so I go through this. You got to look for different structural winners. Commodities usually have a huge impact on what goes on. The compute is real. The applications are scaling. The infrastructure is being stretched. And financing is coming due to paranoia around military supremacy at the country level and obsolescence from the hyperscalers. The world is not simply prepared from an investment standpoint for the electricity, hardware, and commodity needs of the AI revolution. This will be a return to physicality. This is what the future is. The past is about software. Here is the proof. Venezuela, Greenland, let's go give money to defense names as well. These all have something to do with what I said. They're all part of the bigger picture of commodities, rare earths, China, Russia. This is a strategic change because the military situation of the world is in a race. And if you don't believe that, you're not paying attention to what happened with the tariffs last year when I started saying very early on. Rare earth will be the deciding factor. We cannot put tariffs on China. The tariff situation is over. The only way it was going to be a big thing is if it was China and China had something above them. We're still playing the game of rare earth and commodities. Guys, you can see it in silver. Prices are soaring just as China moves to restrict exports, prompting fresh worries about supply of metal. This is not good. Elon Musk talks about how critical in this article silver is for AI. Big copper shortage to pose systemic risk to global economies, warns S&P. This is in the FT this week. I'm throwing this in there just because for the people who keep fading Trump's decision to make sure that the manufacturing side and that people can afford to buy houses. He's doing everything he possibly can because right now is an 80cent probability in prediction markets that the Dems win the house. So, whatever he's going to throw at this, it's credit card uh limiting credit card uh interest rates this morning. It doesn't matter. If you continue to fade this and think that shorting these things are going against the administration is a good idea, I think you're going to suffer.
Here is a rate of change of silver going back to 1990, the last 100 days. Only time it was bigger was during again a supply shortage issue. Uh in terms of uh what happened coming out of COVID copper same chart 100 days you're getting rises in copper DRAM prices insanity you stay between two and a half and five for basically the entire time until when? September. Go look when the silver broke out. Look when I wrote the piece in July. Basically, the PMIs, the market started discounting that PMIs were going higher in September. What else blew out in September, something I wrote about last year. I am not a bottoms up semiconductor person, but I pitched and pounded the table on Micron all year. Made a lot of money on Micron personally. Here's what Micron has done. If you don't believe in Micron on this, here's Western Digital. You don't see charts like this with software companies unless it's a bubble. These have to do with shortages. These are all about shortages.
Alon Musk, most important interview I've heard for this year in terms of an outlook piece. No one will listen to it because most people don't sit there and believe Alon Musk is someone you want to listen to when he's the guy that actually knows everything that's happening from an industrial and a power perspective. He was on moonshots for a three-hour podcast. I'm not going to read this whole thing. I'm just going to read you the front first line and a couple points in here. He frames the current mo moment as a near-term transition problem, not a long run scarcity problem. This is really the issue, and I talk about this all the time. The next five years is a dramatic issue for the world. We underinvested in power. We underinvested in hardware things like gas turbines, transformers. Our grid is old and and decaying. We are not ready for how quickly the needs for AI have come. And that's going to cause issues. the labor market who went to school who thinks looking for answers and doesn't know how to use AI and is resistant to use it. Companies whose structure does not allow them to thrive with AI, lack the curiosity, lack the people to do it. We're not ready for this. We're not ready to have an economy where people are going to lose their jobs because not because they're they're not capable of getting a job. They refuse to learn artificial intelligence. They refuse to read anything that says they should be using it. Instead, they just read what they want to read, which is it's a bubble. It's going to end. I don't need to to know anything. That's what the last 3 years has been about. And as someone who has to sit there in meetings and listen to people, usually above the age of 50, where I am too. Start the argument of it hallucinates. It does this. You guys are all wrong. You're falling behind by the second. Reach out to 22V. I'm happy to do presentations. I'm doing more and more of them. And I'm actually doing them now, which I'll show you later on how to get from point A to point C. And if you don't want to do it, you should be hiring people, young people, hoarding them. The biggest thing that I would be doing running a business right now is looking for all of the talent that are having trouble getting jobs out of universities that don't have the best grades. They have the highest curiosity and they're AI native already. you can make a lot of money because they can pick this stuff up in minutes, not days, not weeks, in minutes.
So, all of this stuff on here leads to a violent regime shift, short-term social unrest, long-term abundance, and a narrow window where infrastructure, not ideology, der determines outcomes. You've got to generate the electricity. You need transformers. You've got to convert that voltage to something that computers can digest. You've got to cool the computers. Electricity generation and cooling are the limiting factors of AI. This is the most important point of the three-hour podcast. We are at the point where all of these bottlenecks are now going to impact investors because we will have delays in data centers. We will not have delays in progress in AI and I'll go through the reasons why. But if you've invested in the data centers and you're long GE Vernova and you're long I'm not going to name all of them because you guys are long all of them. Everybody is. It's not that these companies are not going to make money. The question is over the next two years are they going to be able to surprise on the upside. And the reason that becomes important if your multiple is now 30 40 50 and you've built in the next three years that's fine except for the fact that for the next two years there's going to be a lot of headwinds to actually get the data centers built to actually find the power and can get it in there because of these issues. This is a major major issue because the grid can't give everybody what it needs as quickly as people want. It's an issue not just because of the bottlenecks here. Labor is a bottleneck. The governments are becoming a bottleneck at the state level. All of these things will cause delays and as I'll highlight and as you know whenever we've had delay fears in data centers or efficiency gains on chips which are accelerating dramatically based on what Nvidia said this week, you will get pullbacks in the marketplace. They will be painful. This year is about finding adopters. This year is about finding commodity stocks. Those things have not built in the next three years of demand. What has built in the next three years of demand already is not ready for bottlenecks, which I think are going to come. So just be prepared as I go through this. Look at your portfolio. I'm going to show you charts at the very end that highlight that these things are already starting to underperform. Why is Nvidia unchanged and and and Broadcom unchanged while many semi-names are up 25%. I put a list out with 22V that I only showed two names on last week. I'm going to show you guys the entire list this week just because the payw wall is not done yet. And for those of you who keep reaching out, the payw wall will be done this month. I want to make sure and so does 22V that everything is functioning and everything is good. There's a lot of stuff to go up on it and I want to make sure it works so we don't have to put it up, pull it down, put it up, put it down. So unfortunately it hasn't been finished yet.
These are the main points he makes in there. And what I do is I have a prompt that just says go through and give me the single ex the signal extraction. Eliminate the noise and just give me the most important things that he mentioned. White collar cognitive labor is the first category to be displaced. Absolutely the case. But if you want to train your white collar labor, it can be done. Progress is capped by power generation, power conversion, and cooling, not model design. This becomes important if you made money on Bloom Energy this year. We're entering a new stage. I'll go through the the the bring your own or bring your own generation, bring your own power, whatever you want to call it. Uh it's a major theme going forward. I'll probably write something on it next week. Prices, not wages, collapse in an AI economy. You're going to see massive deflationary pressures. If your business gets hurt by deflation, anything built on code gets hurt by deflation. Get ready. Data centers, grids, transformers, and cooling systems are the choke points. investors underestimate the inflection point and this is where he brings up two important points. Number one, China will exceed US electricity output by multiples by 2026. So if energy efficiency or if energy and power is the main point, China will exceed electricity. I actually don't think I'm not in agreement completely on this. I think the compute side will be okay and I'll go through why. But his point, chip over supply can coexist with an inability to deploy. This is the most important thing. Cyclicality appears artificial. It's caused by power, not demand. AI demand is infinite. I'll keep saying that. Every single month, every single day now that I use it, it's better than it was the day before. It's no longer every release. It's every day. Every single day, I do something differently. I learned something new that it couldn't do the day before, two weeks before, because a lot of these models, particularly Grock, they're updated regularly just like your iPhone would, except it's happening without needing to do a big update. It's from the cloud. It's I'm doing this most of this stuff, not on my uh application. I'm doing it on Chrome, as you can see.
All right. The t the mismatch decides the winners. Semiconductor supply is no longer the dominant limiter. AI development cycles are now measured in months while the physical systems required to support them power generation grid are measured in years. This is the mismatch. Chips without power are stranded assets. He makes the point that at some point this year within 12 months we will have a bunch of GPUs just sitting around because we can't get the power to turn them on. Think of having a billion light bulbs but not having enough sockets to plug the light bulbs in or enough power to get them going. He's saying we're going to be there this year. That's how big the problem is in terms of the uh power. So he was arguing that when asked the question, is Taiwan semi holding things up? He said they might be worried about the past in terms of cyclicality and overbuilding compute, but the reality is they're right. We're going to run out of power. Uh we're not going to be able to plug them in. They're going to be stranded. You're going to purchase them, but they're not going to be doing anything. So just think about that when you're going through. That is a problem that Elon Musk thinks will happen this year. He's already seeing the bottleneck. He goes through these components again that we've talked about. Compute is already power constrained at the margin. This is why hyperscalers are building off-rid. So this is where we start getting into Alon Mus saying you have to bring your own generation. You can't depend on the grid to do that. You obviously have to find a way to build it out yourself which is getting harder and harder. Um the real problem is the infrastructure lag. compute will will appear cyclical on paper while demand is infinite in reality. You're going to get GPUs uh this year into next year sitting idling waiting for transformers, training clusters limited by cooling not chips. Capital markets are misreading underutilization as over supply and that's what will happen this year. there will be effect uh uh at least 70% of the people purely based on the AI bubble um Deutsche Bank thing that I showed last week that the number one risk for this year was that there's a bubble in AI. It was over 50% and the next thing below it I think was 20 or something along those lines. AI growth versus power availability a perfect storm in 2026. The reason this is important and the reason Alon Musk is the single person to know he has already built the largest data center. He's on his way to it being 1 gawatts. There are no other one gigawatt data centers. He's structure or he's his goal is to for it to be two gigawatts soon enough. He's done a cluster that's already over 100,000. I think the largest cluster of GPUs out there currently is still only 50. We don't have Stargate online yet. We don't have any of these other ones. Elon Musk did a Colossus, the first version in 122 days. He is the person that has the only experience of the difference between doing a 50 megawatt data center and a 1 gigawatt until you've done it which nobody thought you could have clusters that big. He's the one that started this whole thing. So he should be the one since he now can talk about batteries became operational and mandatory not optional. That was based on his exposure. And the reason is there's all kinds of instability that are going on because power demand is not linear. So it's very volatile. He exposed how AI training creates highly volatile nonlinear power demand. Vertical integration was forced by execution speed. They did the whole thing. Colossus reframed cyclicality is stranded compute risk. Again, he's the expert on this. If you want to go spend time, spend time with Alon Musk on what it was. He's not Satcha Nadella. He's not Sundar Pachai. He's not. Go through the list of them. Demis Sabis, all of these guys, they're all really, really good technology people. Elon Musk is an engineer. He is a mass scale person. That's what we're at now. That's what PMIs are about. That's why Tesla, in my opinion, will be the best performing one. But it's not just Tesla. his entire business, XAI, SpaceX, Tesla, they're all interconnected into his long-term vision as a systems thinker. The winners in the AI arms race will not be the firms with the most advanced chips, but those who can energize and deploy them fastest. Power, not silicon, is now the pacing item and execution. We are at the inflection point for where software did not need energy. We got fracking. We got software that knocked oil prices down. That knocked natural gas down. We are at the early stages of a new commodity boom because commodities are not a big cost in this situation. And unlike when gas at the pump would triple and oil would triple and we'd see the economy go down, this is not an oil sensitive thing. The energy power is being bottlenecked by the things going on. And we have enough energy in the world. We have enough enough natural gas. So this is a very unique situation. But we do not have enough copper. We do not have enough silver. If you want to read something good on the uh bring your own power semi analysis did a very very long piece. It's free. It's in Substack. How AI labs are solving the power crisis. Highly recommend the semi analysis guys during this. I use them all the time. Um, remember when Musk bought a power plant? Semi analysis was the first to say they just bought a power plant from overseas and are shipping it for the US because they couldn't get a new one built in time. They're doing all this crazy to get the compute. So again, very different time period. Is Colossus the largest? I'm just going through this so you guys can get the numbers on it and just see the scale and the size of this of what he's doing. Does he mention what can be done to speed up power generation? I'm only bringing this up because again, if you want to find investments, this is where you want to spend time on. Okay, I'll eventually put together a list as I've done before.
Um, this is what I included. BYOG and behind the meter power. If you didn't read this week, the physical uh upgrade, I go through a bunch of places in there where you guys should be focusing your attention. edge devices for the computer and and phone and auto upgrade cycle on premise is going to be a big theme this year because you can't depend on the cloud. So every major big company is going to have to have the AI with inside their walls to deal with the amount of bandwidth that's going to happen but also cyber and everything else. This is the paper I'm working on for next week just to show that I'm focusing my attention on where how you can make money on bring your own generation and just look this is what happened again. And look at when this started. I remember having conversations in over the summertime about Bloom Energy. And at that point, most energy people that I talked to thought it was kind of a scam and magic and, you know, Jack and the Beanstalk beans stuff. Again, I think the only reason people thought that is because they didn't recognize the AI problem and the fact that you needed to have your own generation. And this is not a h ah, you know, we'll do we'll wait for the data center to come on. That's the whole point. you need. The reason the US is being so aggressive in terms of doing things around the glo geopolitically is because there's no time left every day in AI is like a year. And that's the part you have to start getting in your mind is the reason the charts look like this is because the demand is insatiable and the need is insatiable and the speed is insatiable. Just another one gigawatt. You're going to have more of these um these stories out there. I have these set up on ChatgPT Pulse, so I get every single one sent to me. If you haven't used Pulse before, I would start using it and just telling it what you care about.
Chevron, I wrote a piece about Chevron. Um, people reached out and said they didn't see it. And I realized I got to make sure I time these things when the actual care goes because everyone on the hedge fund world is very much about what's going to make my make me money the next month. Uh, Chevron's going to be a great stock this year. I'm not going to read you all the stuff, but let's just say this is the critical point. AI companies are going faster and faster to deploy this. They're building bigger and bigger data centers. If you're going to do a 1 gigawatt data center, you need the help of massive companies. You cannot go farm this out to Bloom Energy with that rally. Before that rally, it was less than a $10 billion market cap company. You can go through, you know, they can't do that much. Well, when you start getting into one gigawatt, 2 gawatt, 3 gawatt big data center things, who are you going to bring in? Well, the energy sector is not that big. And that means Exxon and Chevron are going to be part of this. And plus, you need the natural gas. They own the natural gas. So, again, whether you've done the work on this, whether you think of them as an old school company, commodities are going higher. We're at the beginning of a commodity cycle. Go look at all the energy names. Energy and basic materials will be the two perform best performing sectors in my opinion this year. Chevron been consolidating for a while, but more importantly, since fracking became a thing, we went sideways. We did nothing. Now we've got another again. I'm telling you, this is the beginning of something different. We are short power in the world and however we have to get it, generators, whatever. We are not waiting and we are price insensitive. Price insensitive. That's the thing you have to go. That has never been the case before. But this is controlled by a few people and their cost. They're selling something distributed to 8 billion people. It's a odd lot in the cost. I'll show how this is for silver as we go. Exxon just about to make new all-time highs. Very close at this point. The XLE consolidating here. These are John Rog specialties. If he's not already positive on them right now for the breakout, it will come.
Copper and silver play critical roles in AI data center infrastructure, but they represent small percentages of total costs. typically low single digits for copper and negligible for silver. The reason this is important is whether this price is 12,000 or 100,000 it is not going to change the copper and silver needs. That's the thing you have to go and that's why it's price inelastic. We haven't been in this situation. They need what they can get. Now the metals are leading energy. So here's the CRB raw industrials. Look at how good this looks. I've shown this thing many times. I've shown John Ro showing other versions of it in terms of the metals. So this is the raw industrial commodities. Just nice, beautiful rounding bottom. When these things tend to go, they go. Here's the overlay with the six contract of crude. Obviously, we've been breaking away over the course of the last two years. And again, they were all lined up together. And then chat GBT comes out. We have the metals kind of sitting here. This goes down. Everyone knows that right now there's a shortage in metals relative to the needs. And so there's hoarding going on.
Gavin Baker put this out on December 2nd. The next 12 months might be different as the GB300 which is Blackwell finally ramps and then Reuben. I want you to pay attention to this. This is critical. If you don't understand what this means, this is really important. If we don't have any more data centers built this year other than the ones that are near finished, like the Colossus Next Stage, the Blackwells will be ramping. Right now, hoppers make up about 80% of the compute from the data centers. Now, Hopper is a far less efficient chip than Blackwell. So, these are different generations of chips from Nvidia. So, if you remember DeepS last year, what we're saying is these are ramping now, which are about a 50%. So if you have equivalent of 10 gigawatts with hoppers in them, you're now going to basically be equivalent to 5 gawatts of use. After that, it takes it down again and you're down to about 1 gawatt with Reuben. That's how we're going to be able to continue to have the AI progress as the chip efficiency comes. So the last three years or two years while we waited for Blackwell, it didn't come from chip uh improvements. It came from algorithmic improvements like reinforcement learning, all the things we heard from Deep Seek. Well, now we're getting into the chip side. This is why the US can compete with China regardless of the fears that are happening because we're getting efficiency gains. And this week at CES, we heard that the Reuben is going to be the second half of the year. Models trained and then inferenced on these GPUs are likely to show a dramatic leap in capability. I think we see first significant AI scientific breakthrough and the first economically useful agents. That's the thing you have to realize. You if if you're waiting to get from from New York to uh uh to Boston on a train in an hour, the only way you can do that is to have an improvement in the tracks and improvement on all. That's what we're getting now. We're finally going to be able to move faster, which means we'll be able to have AI agents roll out, which will speed up the adoption side. This is what he talked about. And it's cheaper. So inference gets cheaper and it's efficient. It's a mega news story. It's really important. it continues to get through the fact that the next generation will allow us to do more. You'll have more breakthroughs. And that's what he talks about. The breakthroughs across healthcare, climate science, robotics, embodied intelligence, and autonomous driving are going to accelerate. This is why when you hear recursive self-improvement, you hear AGI. We can't get any of this without the models getting better. The models can get better through more power with older chips, which is what Elon Musk has been doing. But then when you say more power meaning more data centers, but then you also get the chips now you're really accelerating things and you can see where the compounding gets. So, I just went to saying that hoppers 10 gawatts now 50 megawws in terms of what it would be with Reuben. And then if you guys haven't done this and you want to go do the numbers in AI and then you're like, "Hey, take these numbers. Is this the right number?" Bring it to um Gemini. That's what this is. Gemini is my academic. I find it to be reminding me of school. I don't like to use it for general questions. I love to use it for correcting stuff or fact-checking.
Um DeepSeek comes out, market freaks out. Microsoft drops some AI data center. TD Cowan market freaks out. Microsoft abandons data center. This was in April. Freaks out. Oracle delays some data. Remember all these things last year? Every time they happened, there was a fall-off in that's going to happen all next year. This this year, this is going to start being a real thing. You're going to have delays. You're going to have all these things show up. And you had Baird come out. And I don't I don't know anything about the person. All I know is that they started coverage, I think, in 24. They raised their price target up to 816 not too long ago and now they just cut it and again growing investor worries about potential power capacity over supplies. It's not an over supply issue but that fear is going to be there.
Mark Andre 2026 outlook cannot highlight enough why this one is also what you need to do. I broke this down very simply. AI is already monetizing faster than any in history. If you guys are worried about the revenues coming in, listen to who you're reading and then listen to Mark Andre. This new wave of AI companies is growing revenue like actual customer revenue. Real demand translated into dollars showing up in banking accounts at an absolutely unprecedented takeoff rate. Again, I I this is why I find it ridiculous when people regurgitate to me that there'll be no revenues. there is going to be trouble for the hyperscalers to get enough revenues relative to the capex in the time period because the capex is going to come first. The bottlenecks are there. If they can't get the revenues because they can't build the capacity like Google and Amazon for the cloud, they're just not going to be able to get the revenues as fast as what people are going to want. It's not a collapse in the stock, but it is probably multiple compression. It is probably continued fears. We're still early in the AI cycle. AI is bigger than the internet. Think electricity or microprocessor. He says it's the biggest investment opportunity of his of his career. The price of AI will collapse and demand will explode. This is the Jevans paradox thing. This is what we're seeing and this is the case. Once AI capability is proven, others catch up shockingly fast. Once somebody proves something is capable, it doesn't seem to be that hard for other people to catch up. This is the no chance of a moat. You can do it with fewer people. This is my whole argument as to why eventually AI cannibalizes everything and capitalism is in a very late stage. Public fear, actual behavior is different. AI is early, monetizing faster than any prior tech wave, deflating in price, expanding in demand, geopolitically competitive, and structurally favors portfolios over single single monopolies.
Um, on the payw wall, this is something else I'm going to do because I've had requests from people. Every time I show one of these, it kind of forces you guys to just watch the video, pause it, take things out. What I am going to do is put on the payw wall the name of the podcast that I think are the most important, the link if you want to go to it, a sum, a three paragraph summary, and then the timestamps underneath. So that way you guys can just go to that site, you can get them all once it's launched. Everything that I show in the video for that week, plus somes that that I don't go, you'll be able to do. Uh you guys realize that all software is about to be free, right? and software is
Presently, most valuable capital asset. So, two points in this. I completely agree, and I still debate people that Adobe, Salesforce.com, they have moats. They're blah, guys. I'm, I'm telling you, I, I just, I don't agree. I think you're, you're trying to pick a bottom in something that could bounce for three months. It could be a shift. You could get short covering as people deleverage. But in terms of businesses and what they're fighting against in software, I, I can build stuff so quickly. It's just not, it's just not there.
And then on this point, and software, software is arguably 30 to 40% of all the benchmarks that you're invested in. That is a massive problem. So, again, I think the S&P is going to be up 15%. I think the software companies, let's assume they're flat. 15% growth in the S&P while the top 40% is flat, just as a general thing, means the bottom 60 have to be up a lot. So, just keep that in mind because if we get that kind of a dispersion when everyone is positioned in this and short the other names or underweight the other names, it is a massive change. And if you're a mutual fund, what's the size of your growth funds? What's the size of your value funds? What's the size of your commodity fund? What's your size of your tech fund? This is going to be a massive rotation in my opinion.
And Elon Musk seems to agree. I wrote this this week. Um, I enjoyed writing this one. I wasn't planning on it. Opus 4.5 inflection point. So, I showed last week what I was able to do with it. I highlighted in here between December 26th and January 3rd, the span of a single holiday week, while most people were not paying attention to the markets, an XAI co-founder, an Anthropic AGI researcher, a Google principal engineer, a former Google Gemini lead, and the founder of Midjourney, one of the, the, the image places, all independently concluded that Opus 4 and 5 plus Claude Code had fundamentally changed what was possible. This wasn't coordinated marketing and orchestrated hype. It was organic recognition from technical leaders who suddenly found themselves experiencing the capabilities they hadn't anticipated arriving for months or years. It was a massive moment. I felt it. I've been talking about how I've seen everything change since the summer, but this was breathtaking.
This came out, uh, yesterday. Claude Code is changing how people hire and work. The transformation Claude Code is making cannot be understated. This isn't a fun tool for devs. This is an entirely new way to use computers, do work, market, sell, and so much more. AI agents, Claude Code is a game changer. Here's mine. For those of you who've ever walked by a desk of one of your data scientists or one of your coders and you see all this, this stuff on all of the screens, I just want you to look. This is me and this is the output I built in terms of it's 800 lines of code. It took me to get this took me seven minutes. And I'm not kidding you. Seven minutes for me to verbally say what I wanted and then go through a few more iterations. And it took me a total of three hours to get to the point where I had this, something I wanted to build for, as I wrote in my paper this week, a decade. This is my turbulence model. I wanted an immune system to warn me, to give me something that basically was like a card counting thing of just like, okay, the cards are no longer in your favor. Doesn't mean you're not going to win the next three hands. It doesn't mean the market's going to go, not going to go up, but it does mean you should be paying attention for other metrics. So, you want a warning system that kind of says, "Hey, there's an increased probability of something happening." And the way the model is set up is it only gives these signals when we're trending above the 50-day moving average. So, you could add other things into it. You can use it yourself off the, the paywall. Uh, but I just wanted to bring that up.
Elon Musk said it best. Singularity has already arrived. It's the year of singularity. Singularity basically means that AI agents, recursive learning, AGI, it's already here. That means the world has changed. Now, a visual for people that have had trouble keeping up with everything that I went through here. AI model demand just continues to increase exponentially. Chip efficiency and adoption continues to go. The problem is we've got issues here. So, the hyperscaler expansion in data center upgrades, they need to increase the data centers to be able to allow more users. We don't have that. We've got a power and grid problem. We have a commodity supply chain problem. We have a memory supply problem. We're seeing it with SK Hynix, with all. We have issues that won't allow us. The demand is there. It's the underinvestment in this side that's going to take time. Memory is going to take a long time to catch up. I had this discussion with people. I said repeatedly in the summer that said we were going to have a glut of memory. Micron is now up from 150 to 350 during that time period. I showed you Western Digital. There are tremendous opportunities if you can find the places where demand is massively above supply. Why now in terms of energy? Because AI demand is accelerating because adoption is finally beginning. It will only increase every single day from here. Computer is no longer the bottleneck. Okay, good.
So, we have the chips. Efficiency gains paradoxically increase total energy demand. This is Jevons paradox. We're making faster chips, cheaper. It allows people to do things. It increases the energy demand because you get more and more users. AI agents are dramatically more. Ask yourself, how many hours a day do you use artificial intelligence? I would say the average person is still less than an hour a day. An AI agent will be working 24 hours, 7 days a week. So, every digital employee and AI agent that gets hired this year, of which there will be millions, is going to be working far more than an individual. Gets back to the productivity side, gets back to the GDP side. Data centers are no longer 20 to 50 megawatt facilities. This is a critical one too because the infrastructure has entered the scale phase. We need bigger ones. We proved the point. That was not the case two years ago. We now need multi-gigawatt AI factories. This change in both size and the speed requirements means energy is now an issue. The grid has run out of slack again. We've been using excess capacity. Now we're running out. Energy has become the binding and persistent constraint. This forces a structural shift in energy sourcing. Hyperscalers can no longer rely solely on the grid. Bring your own generation moves from optional to mandatory. Pulling AI directly into the domain of industrial scale energy production, fuel, logistics, and infrastructure buildout. Go through the entire, uh, chain. We need more energy and we will forever need energy until we can do this all off the sun, which isn't coming anytime soon. The long power problem, a structural investment regime. We're at the beginning of a multi-decade power constraint. Efficiency gains are not reducing demand. They're accelerating it. Again, think about it. In between power, which normally historically if power prices went higher, we would have an issue and demand would go down. We're getting efficiency gains and we're getting a reduction in price on the things that are driving the demand. This is a very unique situation. You have to pay attention to. You have to think outside the box. And I understand no energy person understands this. I also understand no silo-based person understands this. So, it just is what it is. You can spend time with me if you want going through it, but I try to make these as easy as possible. AI adoption is still early. I would say in the 1%. We're in the 1% out of 100% eventually getting there. Right now, power becomes the persistent bottleneck. It's not a transient one. It is forever. For portfolios this year, you want to be long AI adopters. You want to be long healthcare companies. You want to be long insurance companies. You want to be long financial companies. You want to be long CH Robinson. Basically, you can find individual companies that are adopting Goldman Sachs, Morgan Stanley, all of the big banks, not just the enablers, not the ones giving it anymore. Be long power providers and infrastructure. Be long commodities tied to the physical upgrade. Be cautious or short. Businesses built primarily on code. Do your homework on anything you're invested on the software side that can be replicated very simply with what I showed you, which now takes seconds. Because if something takes seven minutes for me to build, it's seconds. For 2026, be wary of high multiple data center buildouts. Delays will cause multiple compression.
Now, this is the bubble. So, the bubble again gets back to what I've just talked about. The top 10 companies have the highest weight in the index since 1929 relative to the 75th percentile stock. I want you to pay attention to these points in time. So, this is just before 2000 and this is just before 1972. The reason I showed you those, there, those inflection points of when we had concentration, it started to change. One of them lines up here. We've seen correlation as the concentration has risen, meaning the ability to make money has been driven by less and less things. The correlation with inside a hypothetical multi-asset portfolio, you're either a winner or a loser and that's it. There's not enough breadth going on. There's basically been a one-trick pony. The one-trick pony since 2009 was exponential innovation. It was software. Software dominated. We are at the end of software. You're going to see a change in multi-asset portfolio. If I'm right and correlations go down, I think you're going to see a dramatic shift in leverage across the system. That's my big theme for this year is that leverage relative to the liquidity that's available for something like the Russell 2000. The reason I think it'll be up 60% is because every single portfolio manager on the planet's underweight or short it, even if it's just not through ETFs, it's through small cap names, it's through everything. These points in time are back to that graph of concentration. We were highly concentrated here. We were highly concentrated here. This is the logarithmic scale of the Goldman Sachs commodity index or what was the Goldman Sachs commodity index. When commodities rise, it's a problem for concentration because historically when there's no movement in energy or commodities, technology wins. The ability for a very few amount of companies to win wins. But when energy becomes the dominant force again, a lot of a lot of places are needed. This is the broadening out of the market. This is the PMI. This is the MAG7 relative to the Russell 2000. We've stayed above the 200-day moving average for a while. We held it there in Liberation Day. It was a great buy. Every freakout we've had, I would watch this chart closely this year. It peaked right around the time early September when we started seeing Micron change, Silver Ch, all of the PMI related things started to change and all of a sudden small caps started outperforming the MAG7. I think this is going down a lot. Look where it started right around ChatGPT. That was that second thing. Here's another ChatGPT one. This is the Goldman Sachs power, uh, customized basket. If you guys are long this, which I know a lot of you are, this is it relative to the S&P. Again, very similar since 2022. You wanted to be in the power names. This is going to have all of the names you guys have probably been in. Another way to play it, Nvidia versus the XSD, equal weight semiconductors. So, I want to be long equal weight semis. That's why I show you equal weight semis all the time. That's why I show you equal weight software. I don't think Nvidia is going down. Just like I'd be surprised if, you know, the MAG7 got hit very hard. I don't think they're going down. I think Nvidia is cheap, but I do think Nvidia relative to the other names is an exposure because since ChatGPT, this was about the data center buildout. Now we're talking about putting this in edge devices. I want to be long semis in cars, semis in phones, semis in, uh, in computers. I just recently got myself a new Mac. That's what I'm working on. I got myself a new iPhone for the first time in a long time. Game changer for me in terms of its ability to do LLMs faster than what I do on my ThinkPad, souped-up ThinkPad. Uh, I bring up the Nvidia versus, uh, XSD because I showed this last week. This is what I wrote at the end of the year. Why this deal mattered and how it basically says the architectural revolution has is now required over Moore's Law. It's no longer about the compute itself. It's about the packaging. So here are the names that have happened so far this year. Nvidia unchanged, Broadcom unchanged. All of these names are up double digits and you guys can go through. This was the complete list of the names that I went that would benefit from the advanced packaging side. The names that I would want to be long this year, which didn't include Nvidia. It does have, uh, Broadcom because they benefit from this part. But as you go through it, this is the way I'll be doing things. And again, John Rog, we'll put the technical score if you know his work, 0 to four, four being the best. One of the names that I love this year that is a bad one is Synopsis. I'm not going to go through details why. You guys can do your own work. The commentary score, what the earnings commentary have been. I have an algorithm or prompt that will basically go through the last three, uh, earnings reports and basically look for progressions and go through it related to the theme that I'm doing. And then estimate revisions. Because obviously all the estimate revisions are going higher almost across the board right now. But again, if you go through the names, you can go through and figure them out now and then go look and ask, "Hey, if I were you guys, just take a snapshot of this, put it into your LLM. How do these names benefit from advanced packaging?" You'll get details on all of them. You don't need me to go through it. But if you guys want to figure out how I can do that, go from a podcast to those names. All the prompts and the algorithms. Call up 22V and you guys can figure out in terms of how I can do that presentation for you.
Um, I wrote this on Bitcoin. I'm gonna finish up this week with Bitcoin. Again, for those of you frustrated by Bitcoin and going through it, I will continue to say Bitcoin, in my opinion, based on everything I showed you today, based on what I showed last week, will be the best performing asset. It is currently held down by the technical levels. Technicals matter a lot for it until we start seeing it break out, which it did finally accomplish something. Thing I highlighted last week, we had closed below the 50-day moving average for a record time period or one of the longest in terms of the last five years. The last one we saw in a similar one was here and then we eventually broke out. We did have a crossover point in terms of the 20-day moving average getting above the 50. We've continued to make higher highs and higher lows. As long as we can keep this up, we're going to get up here. Then we'll spend some more time and then I would think we would eventually break out once we get back above this trend line and back above the 200-day moving average. They're probably going to meet before that happens. Uh, I think this will soar for the rest of the year. We have MACD weeks that are down here. Only reason I'm showing this again, doesn't mean we're at a breakout point yet. The technical chart is still brutal. When the technical chart is brutal, I found in, in Bitcoin and all crypto, there's a lot of traders. They're good traders. They play from both sides. They're shorting it. For people who only buy it, that's great. They're waiting for this to go higher, but the dominant force right now is people selling into strength. The OGs are still selling. There seems to be buying underneath. But when we get the breakout and we get back above 100,000 and I've said, and someone called me this week, I've said on Pump, I've said it on here, three daily closes above 92,000 in a row. And I believe the bottom is in. We did two. A bunch of people called up and said, "I hope we're going to get three." We didn't get three. I still think we're going to get there and go through it. I just want to highlight that one of the reasons that I think this will be the best performing stock is because right now it is being impacted by the Nvidia, the MAG7 stuff, all the things I mentioned because people still view this. I hear this even from people who are focused and do weekly macro podcasts that involve Bitcoin where they're like, "Yeah, well, it's just the NASDAQ." It is not just the NASDAQ, guys. I can't believe I listened to this. But at some point, people have to understand why I wrote the piece that I wrote, that I just showed you guys, and why I've said repeatedly, Bitcoin is the only pure AI trade. You guys want it to be relative to now. Well, what you're going to see this year is the AI trade is power. The AI trade has nothing to do with technology. AI is not technology. A pure AI trade means when does AI start destroying all the tech companies and what actually doesn't need a moat in its business? Bitcoin doesn't need a moat. This will be a story by the end of this year and I promise you if you've never purchased Bitcoin and you've been doubting it, by the end of the year, you will start doing research. So, I'm just going to try and help you get through it because you'll be staring at tokenization every single year, stable coins every single year. The banks offering it every single year. The next time Bitcoin is up at all-time highs at the end of the year, every single person that has doubted it will be questioning themselves. They'll never admit they bought it. They won't tell anyone, but they will.
Here's the chart that is the most positive right now. Ethereum relative to Bitcoin. If we, if I was worried right now, if I thought we were going to go significantly lower, I would think this would break down. I think this is like NASDAQ over Qs. I'm sorry, Qs over S&P, Qs over SPY. This is the growth thing for it. This is the thing that matters. As long as Ethereum keeps hanging in there relative to Bitcoin, then for me, I think eventually these will all go up together on the PMI cycle. Final thing I just want to remind people because we're at this stage. The money is not the problem. AI is the new global arms race and capex will eventually be funded by governments. If you want to know why gold, silver, Bitcoin is soaring, it's the debasement to fund the AI arms race. But you can't print energy. Elon Musk, that is why Bitcoin is based on energy. You can't issue fake fiat currency and every government in history has done so. But it is impossible to fake energy. Elon Musk has forever equated Bitcoin to energy. That is the reason why you need to do your homework because everything that was up here was on October 13th, three days before we had an event where Trump came out said this. We ended up losing market makers in crypto and crypto's been kind of in a bare market phase since then, which people thought was some other magical thing. The reality is the next time we go higher and people realize that Bitcoin is energy. It's based on energy and energy is the new currency, I think the future currency will essentially just be wattage. That's it for me this week. Hopefully I provoked some thought. Reach out to 22V. I promise you I will get the paywall up as soon as possible. Um, I'm going to keep creating things. We're going to keep working through this. But the main thing for this year is if you've been spending the time putting AI in a bubble, start reaching out. Start finding ways to get me more involved in how you're doing things. Even if it's just a one-time show. It's a big year for AI and if you don't do something this year, I think your business, your children, the whole thing, you're going to fall way behind because it's accelerating too fast. It is time to spend the time on it. I'll also release my HRV post this week. And just as a reminder for those of you in Miami or those of you in Florida who've been reaching out to me, Pomp and I will be doing a live broadcast from Real Vision. I'll be down there for at least four days. If you guys are covered by 22V down there and you want to see me, I'm going to be in the Miami area for a while. You can reach out to the sales force at 22V. Happy to spend some time with people who want to get to know me. All the best, guys. See you.