Semis vs Software Is Breaking the Market (Scarcity vs Abundance)

Um, we got a lot to cover today. Uh, so I'm not going to read through all this, but let's just say, uh, most of the story is going to be about scarcity verse abundance, about what's going on with semis and software, and then some of the releases on chat GPT health.

But to start, let's go through basically the week. Uh, S&P small change down slightly for the week. Q's down a little bit more. They continue to underperform. Mag 7 underperforming worse than that down for the year. IWM's the outperformer we keep seeing small caps uh is basically about 8% uh sorry 7% in the last uh uh two weeks and when you look around the globe uh it's very easy to see that the worst performing market uh basically falling again the way it was last year so far is uh the US market.

Uh, this is a theme that I think is going to persist not just this year, not just from last year, but going forward. Uh, I'm a big believer in foreign markets, emerging markets, uh, and small caps and midcaps. I think this continues and again it's all about software. Uh, that may not be the way people are thinking about it, but I have equated uh in almost all of my presentations over the years uh looking back to 1993 as when Netscape came out and the US dominated coding uh that effectively the dollar and the capital surplus, capital account surplus has grown since uh Silicon Valley basically dominated the world and now with AI that dominance is over uh and the disruption has started.

So the NASDAQ, you're not going to see that too many times with a fully green board and the only market it is uh outperforming right now is the Swiss market. Uh, when you go through it's this is not just a small cap midcap story and this is something I want to make sure is clear. This is uh about concentration. This is why it gets back to coding. It gets back to anything based on code that has worked over the last whatever years including things like Mastercard.

So, uh, if you go through, this is the top names in the S&P 500. You can see how many names have actually outperformed the index. Out of the top, you know, this is, uh, 18 names, uh, because Google's counted twice in here. You have about three that have outperformed, sorry, five out of the 18 that have outperformed the actual S&P so far. This is the S&P 1500. This is always a gauge that I've used to get a sense on concentration. much better than small cap uh midcap uh it includes it so it's impacted it but this story resonates across the S&P 500 as well so within large cap you're seeing the same thing which is this year so far has been about a broadening out and a lack of concentration this takes the S&P 1500 all the names so far this year the S&P 1500 is up 1.75% and you have to go all the way down to name number 1 so 107 out of the 1500 have outperformed the index. Uh, which means the concentration you're getting u a broadening out. That's why you're seeing equal weight do well. And this is a dramatic shift from what we've seen over the last five years.

This is the total of the last five years. So forget last forget um, you know, the prior years with chatbt. This is just taking the last five years. Uh, the S&P 1500 was up 79%. All you had to do is go down to uh number 448. uh 448 out of the 1500 outperformed the index. So again, we have uh, we've seen a change going on. And that's not the only change. This is uh an overlay between the NDX, that's the orange line, still sitting up near the highs. The other line is the Russell 1000 growth index relative to the S&P. So what's happening is growth or software. Take your pick. Whatever this stuff is, growth is getting hit. This is another major theme for me, but it all lines up with the same thing, which is to get in the growth side, to have the stuff that's been growing. We're running into a big disruptive phase right now where the commodity side, the scarcity side is starting to dominate the index.

One part of the scarcity side, I've shown this before. I'm going to keep showing it because people are continuing to call up saying you got to buy software. Uh, I didn't get many AI bubble uh talk themes showing up. No trending topics. So instead of AI bubbles, everybody who was an AI bubble seems to be trying to pick the bottom in software. I think you have to look in the mirror and say, am I bullish software because I don't believe in AI or am I bullish software for some other reason?

Um, this is the equal weight semi out of the 15 S&P 1500 making new highs and now up 65% since the lows back in May of last year. Software made new lows. It's now only uh, it's now down 6% or 5% sorry since that period. So we're continuing to see that divergence. Uh, you're going to have periods where you get some kind of mean reversion. It could happen around earnings time could happen occasionally. Uh, I just think this is a trend that is pretty much fading. Uh, the prior technology in the decade disruptions were which were really the two big ones. Fracking which eventually disrupted and especially disrupted oil. Uh, but the energy names ended up getting sucked up completely into it. And then you also had the mall names getting sucked up by uh by Amazon.

Uh, another week of just you can't I mean it it it's amazing. We have so many stories every week in terms of out of the White House and just things that people uh get bearish on and they end up fading away because there's so many of them. Um, we had the PAL subpoena to start the week. Then we got the credit card proposal. Then we got the electricity bills geared towards the hyperscalers. Um, I'm I'm bringing this one up twice because I've highlighted this for six months that the tech companies because electricity is just not a big portion of the cost and because they're going to be focused on building behind the meter will have to cover the costs on the grid. Uh, I think now that is a given. It wasn't just Microsoft which was the first story uh that said they were coming in but now he's basically saying they're all going to have to deal with it. uh and that'll help on this front.

Inflation continues to be a non-story. Uh, if you're talking about inflation, it's more about what's going to happen in the future. Uh, I'm sure we're not going to get down to 2% overall. At least that's my gut. Uh, but I still believe that with house housing costs falling, wages uh under pressure, and the jobs market remaining weak. At the same time that we don't have oil going higher, which I think will change, but I don't think we're going to have a violent move higher. I think the inflation picture is going to stay uh basically in the 2 and a half to 3% 3 and a quarter range. We're at 270. This is sticky inflation X shelter versus the core CPI. They're both in line. So, that's there. And then you have true inflation uh year-over-year core, which is actually making new lows and heading back down. So, take your pick. Inflation not a story.

The other data out this week, we got IP out um the diffusion side, which should be the most correlated to the PMI. This is the industrial production three-month diffusion, five-month average related to ISM PMIs, which still remain low, and again, I talked about that last week. The global PMIs are actually up close to that level. Uh, so, don't get caught thinking about the ISM PMIs. Eventually, that survey will go higher. Uh, but where we are right now is that red line there is where the current reading is. This is the five-month average. So the industrial production side continues to get strong. This was out from the global macro investor team Raul and the gang. I will be down in Miami speaking uh down there and doing a live uh broadcast with Anthony Pompiano. So if anyone's down in Miami uh stop by and see me.

Uh, this is based on financial conditions. this is their own proxy. Again, highlighting PMI should go higher. And then we finally started to see the bounce. We'll see how the rest of the regionals do, but this is a very strong sign to start off the year. You have the Empire Manufacturing beating in a big way, coming up sharply higher than last month. Same thing for Philly Fed. Uh, so the first two regionals to come out are highlighting this.

I mentioned the scarcity set. I just want to kind of go through what I think the themes are that are secular for this year. Meaning I think these are the ones that are going to happen all year. It's not just a small cap midcap story. I want to be long scarcity. Um, it's one of the reasons why whether it's gold, silver, uh, copper, semiconductors, memory, um, all of this is related to an underinvestment on the industrial side. Power fits into that. Short abundance is very simple. Uh, that fits in with the growth story. Anything that has been isolated. Think about abundance as a broadening out. The opposite of concentration. uh, we are running into a problem for anything built on abundance because AI is disrupting anything in terms of I want to be long bloated AI adopters healthcare companies financials health care companies I'll go through healthcare later and then short the high multiple AI infrastructure I still believe the data center buildout will happen I think there'll be at least two fear trades this year or one big fear trade related to bottlenecks slowing down the buildouts uh like we saw in Oracle style selloff but in a bigger way all of those names teams that uh have already built in the next couple years of the buildout that that are going to suffer. If there's any slowdowns, I'd be very wary of that. Uh, short non AI native knowledge workers, long AI native knowledge workers.

This is a theme for me personally. I'm spending a lot more time doing training sessions for asset managers and in particular, I'm starting to help a lot more college kids. If you guys are interested in uh, I'm going to show some of the stuff today and obviously the payw wall go up but this is a big story. You have to get on top of the uh AI usage. You don't have time to wait and remember bottlenecks matter more than narratives this year.

I released this this week on the power constraint. For those of you who got it um, you got to see it. If you didn't get the names on it, reach out to the 22V crowd. I think I gave 23 24 names that are best associated with it. Some of them fit in with those bottleneck crews, but there's a lot on there that I've already kind of fallen off where you can uh find some names. The driving factor behind that was the uh this build out. I just wanted to show this and just make sure people realize what Alon Musk is doing with the build out of his data center. You can see the scale and size of this and his play on Microsoft macro hard. Uh, nobody else can build as fast as Elon Musk. And for those of you who keep wondering when I'm going to say Elon correctly, uh, I apologize. A good friend of mine used to work for me in Brazil named Elon. And I keep saying Elon for Elon. So, I will try to correct myself and be on top of it for those of you reminding me.

Uh, Oracle, again, I just want to go through this fall that happened. I think you're going to see these in many of the names. If they miss, you're going to end up in this situation where they're still going to produce good returns for the year. I think they'll underperform, but I think it's a much harder trade. You're swimming upstream because of the crowdedness. When there's a broadening out, by definition, what it means is I got more choices. So, if you've got more alpha choices and that that se the stat I showed you of how few names uh under outperform the prior five years, a lot more names outperforming now. So, if you're in the crowded names, the the coding names, the software names, and you're trying to pick a bottom, I think you're at at risk.

Um, in terms of the power names, I just wanted to give you guys a again, this is something I put out at the end of December. Uh, this remains a big story, but it became a bigger story this week. So, as a reminder, the thing that I'm kind of spending with people on, this is the process that I go through to get to the names, all of this in here is very important and part of what I'm starting to tell people. This human element here, which cannot be replaced, this human element here. So, in terms of picking the ideas and then getting to the selection or the filtering of the names, the rest of this is what you have to incorporate in there. This is how I go through my process.

Um, basically to start off, I have a skill that's created. It does all of these things and you end up with at the end this. So, this is the stuff for 22V. I've got Qualcomm highlighted here because it was from last week, but there's a reason why it's also highlighted this week. But these are the names with inside the advanced packaging which fits into the stories that's really floating around on data management. I think it's important for you if you haven't spent the time to go understand why these names are doing what they're doing that these are the lists of the names. I created an index called package for advanced packaging. So all of those names here this is an index equal weight of all of those. So far it's outperformed the market significantly. It's up 12%. These are the semi-name mainly semi-names that are in there, but it's primarily a semiconductor story. There's some packaging things in there as well for AMC core and uh along the lines ASML. But if you look back over the last four years, this is not done what Nvidia and Broadcom have done. This is more like the PMIs. This is more like the XLE. This is more like the things that have been consolidating related to PMIs. This is the very very early innings of what I put out. If you want more details on it, again, reach out to 22V. Happy to go through this.

Qualcomm did have a very good CES. It has underperformed significantly so far this year. It's down about 6% year to date. That name fits in with the group. It also fits in in a big way with the enterprise uh that's going to have to build out on premise. It will benefit from uh the Apple launch in terms of the iPhone uh AI. I've I've mentioned I got a new iPhone. Phenomenal uh uh increase for me in terms of productivity, because of the battery, and because of the speed. Uh, but also it's going to benefit from the computer side as we do the roll out. I think companies are going to want their employees once they have the agentic side set up which is going they're going to want them to have iPhones. So I think the enterprise uh upgrade cycle will happen this year even if it's in the second half.

We did get earnings from TSMC. I'm not going to read all the details but basically we believe the AI is real. Not only real starting to grow into our daily life. He revealed the AI accelerator revenue accounted for a high teens percent. Remember TSM has been one of the ones holding back uh supply out of fear. He discussed this. This is the CEO. The bubble commentary when asked about an AI bubble. He stated careless investment would be a disaster for us given that we're spending over 50 billion. He emphasized that Taiwan Semi does not build capacity based on forecast but based on committed demand. So basically was saying he already has the demand. They're still being very conservative. Uh, he noted higher capex which again they're increasing their capex significantly from 42 to 52 to 56. Uh, insatiable demand for AI related computing power. Just remember how many times you've heard people say AI is a bubble. Uh, if it's a bubble, everyone is still saying the same thing. And as I go through with Goldman Sachs earnings commentary and JP Morgan, uh, not only are the uh companies that are making AI or producing the chips for the intelligence, you've got the power needs that are happening, you also starting to see the AI adoption growing rapidly. Uh, he briefly touched on riffs. He acknowledged that high bandwidth memory shortages could constrain the overall server market, potentially capping how many logic chips it can ship if they can't be packaged with memory. The memory shortage is a massive story and I hear more people fading it. As someone who has been on Micron for a long time now below a hundred and continuing to sit in the exact same position, it's because I see nothing on the horizon. And out of every 10 people I talk to, at least six, if not seven of them either don't believe it or it's too late and they can't get in. Just a summary on everything that's gone on. So the Micron story is still real.

Mark Newton, one of the guys I follow, just to look for some interesting commentary. I thought the Lag seven and the software names remaining in consolidation was a big deal. Until we get more stabilization from the big boys, meaning as he's calling them, the lag seven, it might prove tough to show real acceleration on any rally. But make no mistake, with transports, IWM, midcaps at new highs, and equal weight S&Ps, this is exactly what I've talked about, and this is exactly what I've shown so far. This market is stronger than what the cues may imply in the short run. Just many aren't used to seeing these names, transport names, chemical names ripping while they sit with their palunteer waiting. Adapt or die. Could not agree more. This is a different year. And I think you have to get used to this for the foreseeable future. I think this is the official end of the dominance that happened with coding. And that's why I said this has got to be understood by people from a macro basis. Uh, and I don't think a lot of people have thought about equating everything including the capital surplus account to coding.

Nvidia Reuben, I brought this up last week. This is from a data center thing this week. The shift will reshape how data centers get built this decade. Again, Reuben was a major announcement in the first week or last week in CES. It gets right back in to this which had nothing to do with Reubin. This just had to do with the Nvidia Grock deal and what Nvidia was sending as a signal and what was supposed to happen from the growth in NPUs and edge devices. This accelerated it and that's basically data movement. So, I'm not going to read all this, but if you want to take these, take a snapshot, upload it into your LLM, just go in there and say, "What is the the change from GPUs into data movement for edge devices?" Uh, and the new architecture for Reuben matter for new companies. And I'll basically go through this, but five bullet points. Computer is no longer the dominant cost. Idle GPUs are the new tax. Locality beats raw speed. This is going to be really really important as we kind of get into on premise uh east west traffic becomes critical. This goes with some of the names that are working. The reason I'm bringing this up is not to overwhelm you with something that's a little technical. It's to make sure you realize that we ended the GPU side in 2025. It probably happened midway through the year. That was the end of the GPUs being the only story for AI. Now we're transitioning for every GPU story which is Nvidia. All of those packaging names I showed you did not include Nvidia. You're dealing with a lot of winners now and at the expense of the one winner. It's not that I think Nvidia is going down. I don't think any of those, the lag seven, any of them in general as a group are going to go down. I just think they're going to underperform. And if they finish down 5% and the S&P finishes up 15%, the bottom 60% are going to do what they're doing, which is carry the index and be up over 25% to make the math work. And then the Russell 2000 will be up 50 plus percent to make the math work.

Um, just a simple line here to make sure that you understand the transition from the GPUs to data management. So we also got a deepseek paper out this week where basically they're confirming that a memory augmented architecture for smaller more efficient AI is going to be the story going forward. So Reuben and Deep See confirm that memory is the key to performance per watt. memory becomes the marginal dollar that improves efficiency. Again, these are th this is not old news at this point. And I remind everyone something I've said multiple times. It was the summer of last year that I was hearing from investors that I was visiting that there was going to be a glut in memory. So, you have to think about the fact that that was only six months ago, less than six months ago, and I was hearing that from multiple people because the cellside research was going around that way. So you have Micron that have gone from 150 to 360 as of Friday. But it's not just those names. So again, direct beneficiaries from data movement bottlenecks. Here you go. There's some names. Most of those are on the list. I did SanDisk got a lot of love this week. And the they came out and said this is incremental new nan demand that is not previously in their current forecast. So again, they're admitting that this is news. So what has their stock done? Well, everyone thinks it's a bubble because it's gone from 225 up to 387 already this month. But it did it in a basian way. It did it on new news and it readjusted in price and I'm telling you news that was not expected because this is a big change for right now and for what Reubin's going to need. And everyone's going to want Reuben as I showed you the efficiency gains and everyone starts realizing that the bottleneck is power. So the efficiency chips are more important than they've ever been. Ruben shifts a large portion of inference memory from transient DRAM usage to persistent local storage. That one architectural decision explodes nan demand. So again, just names that are fitting in there.

Uh, another one I spent a lot of time talking to people about this week, Pure Storage, which is a great name. It's done well, but it's also corrected over the course of the last 18 months. Uh, more on this on there. City came out and talked about it. Um, you've got uh people talking on the DDR side and just going through it. So you've got Micron, Samsung, SKHEX, all of them showing up. So when you look at this chart, you're like, this is a bubble. At the same time, you go look at the earnings growth that's happening. And then you go look at the pees. So again, when I first started pitching this stock, the PE was about six. So now off next year's earnings, it's currently at nine. So when you go through Expost, you'll see the inevitable bears saying this is a normal cycle and eventually you're going to have peak prices because this is a semiconductor. Again, I'm going to say this over and over again for the next five years. And this is not going to be for every part. The DRAM thing may peak next year. I doubt it's going to peak with the multiples here when we're still in a serious supply shortage. And when Taiwan semi-chairman Mark Leu buys 8 million of Micron shares near all-time highs, he's on the board of Micron. Uh, I don't think this would have flied necessarily in the US the same way, but whatever. Uh, you've got more stories on Micron. I'm I'm giving you guys just you can go do this research on your own. Uh, it anticipated will only be able to meet half to twothirds of the demand from its key customers. It has devoured the global memory supply. We are in this situation for one reason and one reason only. As of a year ago, software was the only game in town when hardware had been through so many down cycles. So many where every time things go higher, they collapse because if memory memory bandwidth got to this situation with phones and the phone costs doubled, it would be an issue. In this case, what's happening is the driving factor for this is is not a big cost, input cost for artificial intelligence, and you're in a race for all of it from places that have massive amounts of money. A memory bottleneck dubbed the memory wall that now limits AI scalability and is expected to drive up prices for both enterprise and consumer devices. AI memory is sold out, causing an unprecedented surge in prices. Micron just broke ground on a $100 billion memory manufacturing complex in history. The largest semiconductor facility in the US. They just broke ground. For those of you asking what are they breaking ground on, there's a lot of stuff that's happening this year. Again, belong PMIs. At the same time, SKHEX to invest 13 billion in new plant mount shortage. These are all this week. Everything that I've showed you is this week. So again, supply shortages. What's going on? One analyst estimates the DRAM industry can only support 15 gawatts of AI data center capacity over the next two years without cannibalizing other markets. Yet plans call for 40 to 50 gawatt. Shortages could cap deployments leading to underutilized delayed projects.

So let's go now to the hopper. Three-year-old chips as Matt Seagull put out. Um, and again you guys can watch this in Bloomberg. uh rental prices where Jim Chanos has been focusing on just are up 14% from the November lows. Even though these chips are uh Reuben is 90% uh be or significantly better. We're talking uh 20 times better than Hopper in terms of uh power, but you're saving efficiency in terms of the power needs by 90% relative to to Hopper. Uh, Intel almost sold out. Apple and Qualcomm fret over strained supplies of Japan's glass cloth. Again, I'm not going to show copper this week. I'm not going to show uh silver. You get the point. Scarcity, guys. Invest in scarcity. Learn how to invest during a commodity super cycle, which is what we're in. There will be sharp falls during the year because when something like Micron goes up as fast as it does, eventually it'll hit a level. And we may have hit it on Friday. We'll hit a level. It will fall back $50 to $70 uh just in a correction. And I assume that's going to happen with the Mag 7. I assume that's going to happen with fears over the buildout or fears that their revenues haven't come in.

So, let's now jump to software, the gift that keeps on giving for now. Uh, every time there's positive news, CRM stock pews, is it an inverse AI ETF? Now, this gets back into my theme on coding. Gavin Baker this week coowork which was which came out which I'll talk about more but remember when Microsoft co-pilot uh the stock ripped higher because they did the math and people the analysts started saying how much Microsoft would get in terms of revenues from co-pilot that was a long time ago I I believe it was 2023 uh that's a problem to have Microsoft co-pilot as a story in the summer or not even the summer of 2023 so almost 3 years ago and Claude Co-work comes out and it was built in 10 days without a human being involved except for the idea. The idea came from the users. If you haven't read the story on it, Claude code opus 4.5 claude co-work. We are in the point of recursive self-learning and building stuff. So when you think about Microsoft Copilot, I want you to think Claude Co-work. I'm using it already. I have refused to use Microsoft Copilot even though it has improved because I find it to be useless compared to the LLMs. I'm a power user on every LLM. I'm on it all day long on my phone and my laptop no matter where I am all the time. You cannot pay me to use Microsoft Copilot instead of the LLMs. Even if I was in my I I think this is a major problem for software companies including Microsoft unless something changes because these things can be built quickly. I installed Cloork on my personal laptop. Since then it has boom. Morgan Stanley says anthropics code plus co-work is dominating investor chat and adding pressure on software.

Uh, again, I don't, you know, Elon Musk has been all over the place saying this is the end of software. Eric Schmidt has said this. Uh, Chamath has said this. Uh, there's a I mean, I'm going to go through Sequoia Partners. I I there's a lot going on. This is not just me, but I think ma most investors who are, let's say, mean reverting, of which a lot of X's, they tend to be the bears on AI. This is a AI bubble trade now to try and pick the bottom in software. I highly highly doubt that that's going to end up working out. There will be a couple that can adapt, but as I get through this, just remember incumbents have a really hard time changing the culture and changing the business and AI agents are moving way too fast. And I and the costs are going up for the models. I just don't know how these guys are going to pull it off. We'll see. Uh, but for the time being, I'll give you some trades towards the end. Uh, Ralph was released, too. It's an autonomous AI coding loop. you can go find out about it. These are the stories that make them the the the most important.

Joe Weisenthal, who I've listened to a lot of the OddLots uh podcast this year, uh over the last year, I would say generally been bearish or skeptical of of the AI trade and and had a lot of bubble people on the data center thing, overbuild. That's gen been the general thesis. Well, he posted this this year. The AI productivity potential has never been more obvious in caps. In today's OddLot newsletter, I wrote part one of my takeaways from the weekend playing with claude code and building. You should really give it a read um and just see what someone who's not a coder who sat down and did the same thing I did with my turbulence model, which is just try it. You'd be shocked at how easy it is to build things. So later in the week, three million lines written over a week of continuous AI agent time with chatbt5.2. This is from Greg Rockman at OpenAI. Extending the amount of time it can run on its own from the cloud one. Again, they just keep leaprogging. If one releases one, then the next one releases a model that's better than that one. This will continue. And all of these models, I will reiterate, are not on Blackwell. Blackwell will come out, especially for Gro 5 in Q1. The acceleration in these whatever you're seeing right now in AI agents will be significantly stronger in the next three months and in the next six months and in the next nine months big businesses especially big companies where you see their buildings across the country like Salesforce.com or Microsoft very very difficult to navigate getting your entire organization to change when both of them Microsoft Copilot and Agent Force for Salesforce this has been part of their theme. Maybe they're able to turn this around. But as I go through this and I show you some of the papers I'm writing, I just find it hard to believe that you're going to fight against a tide that is just massive.

Aaron Levy, another person to respect from uh Box. The capability overhang right now in AI is pretty massive. Most of the world thinks of AI as chat bots and that will answer a question on demand, but yet do not yet do real work for them. this concept of asking a question. If you do that, if you basically do Google search with it, you're not using it yet. It is not made for that. It is not made for giving you an answer. It is meant for you giving it context and getting a probability on the question that you're answering. If you're an if you're looking for an answer, you're more likely to get a hallucination. If you're looking for something to include on your thought process, because I've heard multiple people say, "I asked to do this. It gave me five different answers." Don't ask it a question for an answer. Don't ask it, "My knee hurts on the inside. What is that?" "Oh, you have tendonitis. You have this." It's trying to please you. If you go through and say it and you say, "What is this most likely?" Giving it the context, it'll give you a higher probability answer based on it. You have to learn how to use the context. That's the human element. We have become stupid due to school and due to Google search. We've become siloed because of corporations. You must think broader and you must think like a polymath and ask it the right questions.

AWG from Moonshots is now posting awesome exposts every day. He only has 26,000 followers. I will single-handedly try to drive that up just because I think all of you guys will will be grateful in the end. He every day puts the news that's going out. I go in there because I'm looking for the best places to get it. His stuff is genius. And recursive self-improvement has graduated from a safety paper to a shipping manifest. Anthropic has confirmed that Claude Code wrote the entire new Claude Cowwork desktop in just one and a half weeks. Think about that. No humans. McKenzie 25,000 AI agents is now counting their workforce, including the digital employees. Get used to it. Uh, Sundar Pachai talking about AI agents and announcing the universe commerce protocol AI agents for shopping. Goldman Sachs on their call about AI. With AI, you will be able to break that paradox and scale output with the option of adding people or not. Again, getting into the efficiency and productivity gains. We just announced the launch of one Goldman Sachs 3.0, know our new operating model propelled by AI starting with six work streams we identified as ripe for disruption. It creates an ability for us in the next 5 years to accelerate the pace of operating leverage creating capacity to grow revenues. This on premise release especially by the financial companies is going to be the gift that keeps on giving. Remember the paper I wrote on artificial intelligence is a faster and better version of what QE did for profit for profits and for stocks back during the prior decade. This will be a golden age for profit margins for companies that are bloated like the banks like the insurance companies. Uh JP Morgan massive investment to create autonomous systems that prevent headcount from growing with revenue. JP Morgan was aggressively moving beyond simple chat chat bots with agents. These agents have goal-oriented re behavior. They confirm their LLM suite is already handling autonomous trade settlements. Again, very very small in terms of what's happening there. So far, I talked to enough people at JP Morgan who still complain about their ability to use it. But for the workflow AI agent side, that's there. What is going to take a lot more time is breaking down the silos of these organizations. I will write a paper on how you make money on breaking down these silos. One of the ways is pure storage but I want Palunteer and data bricks and um snowflake all of these companies will be a part of it but I was involved as managing director at Morgan Stanley and I know the uh silos and the amount of just the intense amount of of software in every single silo. It's another reason why the software comp the enterprise software companies are screwed. When these silos come down, the bloat comes down. When the bloat comes down, the dollar spent outside on software will go down. That's what you're going to be replacing. People and software budgets. They guided for 105 billion in expenses driven largely by AI infrastructure and talent. So again, I t T T T T T T T T T T T T T T T T T T T T S T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T Semi spending uh more in terms of uh infrastructure and capex. The banks doing the same thing. Here's that article again for those signed up for 22V here. when the payw wall goes up, just let me know. You guys can get this as well. This goes through how artificial intelligence will flood the economy with just an enormous amount of money seen in the form of productivity and we're already seeing it with creating uh 4% GDP, no inflation, and no job hiring. Uh again, I've written these two papers. Vibe coding is the chat GPT moment for code. Why the great rerating began? When I wrote this, I got multiple people re reaching out saying, "This has already happened." Again, trying to pick a bottom in this stuff. I can't see a bottom. I I I I just can't. I think if you're a one-trick pony in software or you've got a large institution, uh, can Microsoft come out of this? I Yeah, they're a big organization. They've got our hands in everything and they certainly can, but I I think it's I think it's up for debate on all these. And I think it's very difficult. The revenues will speak. Uh, but I think more importantly Claude Co-work just came out this week and it just highlights that we're only at the very beginning stages where the code is now good enough for it can doing the entire app building which takes things from taking months if not years into days and that's the critical part that took 10 days to build and as Gavin Baker said Microsoft Copilot has had three years to try and do what co-work did in 10 days think about that uh the Opus 4.5 inflection point so this is not something I haven't been on top of and next week I will release a paper titled why buying cheap software is the new AI bubble trade comparing it to people who faded the ompic side. So we were fighting obesity and the staples processed food companies suffered. I think software companies are basically had their osmpic moment and that ompic moment is basically co-work. Uh, again, if you want to go listen to Sonia Yang, partner at Sequoa, speak about the software side. You don't have to take my word uh for this. You can go listen to someone, but this is the highlights of it in terms of what's going on. Um, again, they're a VC place, so they're focused on startups, which are competing, but every single person, I mean, Mark Andre, all of them are saying the exact same story in terms of the the difficulties it'll be to challenge uh startup businesses and software and AI native places. Nate Jones did one on it this week in terms of uh Claude Co-work. He even gave a demo on it. I'm gonna highlight him a lot.

I did want to just highlight I took the transcript on this and I want to show you one of the things that I'm training people. So I do have a hedge fund analyst skill that takes any information. It doesn't have to be from a podcast. So it doesn't have to be a transcript like Nate Jones. It can be something that I read on X. It can be a report that I got from Goldman Sachs or Morgan Stanley sent to me from someone else or Susuana. Uh, it could be a conference that I'm attending and I hear something and when I come back I've taken seven bullet points. I take those seven bullet points and I basically paste them in here and I say run the hedge fund analyst skill on that. This is what it does. It's creating a it's a skill I've created which basically creates a PM research memo output with all of these different components. Again, highlighting the risk side, the reward side, the second and third tier effects, looking for the supply chains, every single part of it to keep players, all the stories involved. It does that in a matter of seconds. This is a matter of seconds. So, the output ends up being 10word pages. I take that and then I go through my equity hedge thing that I showed you before into step two. Uh, again, I'm showing this as a live demo now for people, I will try to finish the videos on it so they can go up on the payw wall uh to help uh younger people and anyone interested in learning this. If you're on the institutional side uh and you want to just watch me do it and kind of ask questions if your organization's looking for help on this on how you can utilize it, I will guarantee you that one hour with me, you will learn things that you haven't seen before on it and it will speed up the process of getting going. This is something that you will never get a button to push. AI agents are not made to solve the investment process. They are made to solve the part that is the grunt work. The human element is where context comes in. Context is important. At the end of the year when the last football game is playing, the context that matters is does the game matter and is the starting quarterback playing? Obviously, well, the starting quarterback playing would immediately change the odds and I think AI would be pretty good at that. But knowing that the game doesn't matter or that a lot of the players are only going to play for the first quarter, it's not going to have the context. So, think sports, think everything. That's what we do for a living is the people that are great in this business, they put into context, positioning, sentiment. Uh, is this a story I've heard before the relative nature of the sentiment? All of those are context.

That was the moonshot for this week. Uh, I'm not going to go through all of them, but they did say this is an inflection year with rapid acceleration and it feels like the year of singularity. AGI's here is effectively being a story that I'm hearing repeatedly. Elon Musk, um, they spent a lot of time talking about the interview they did with them. Some great stuff, including the humanoid side, which I'll get into here. uh they went through a lot on opus 4.5 and acknowledged the fact like I said this is happening before Blackwell that this is AGI. They also went into a topic that I've spoken about here with Pomp uh and really since 2013 when I said it is now officially not a relevant statistic. It will continue to screw with macro people. It will continue to screw with economists and strategists. GDP is an irrelevant statistic. It's never been less relevant than it is today because of how fast exponential innovation. The reason 2013 is because that's when exponential innovation hit kind of the elbow point in my world for disrupting GDP. They talk about this in there and for the first time on Moonshots and really the first time I've heard anyone go through it. They said it can't work. It it it's not built for this and it's especially not built for the fact that AI is structurally deflationary, the most deflationary force we will ever feel. um output decouples from lab decouples from labor. I've talked about that relentlessly. The better AI works, the worse GDP becomes as a proxy for progress. Could not agree more.

They talk about it. This was really interesting because when you have someone like Dave Blondon who has been a naysayer on the humanoid deployment and saying repeatedly that we just don't have the scale to get this stuff out. So maybe by 2035 we could get to and I forget what the number is. Let's just say a million uh humanoids. He just upgraded everything significantly. And the reason was after they were done or while they were at the Gigafactory doing the interview with Elon, hopefully I'm doing a good job with his name today, guys. Uh, he basically highlighted that he saw robots building robots. So he's getting what the Ford CEO saw in the dark factories in China, which is, hey, we're already at the point where robots are building robots. And that's what allows it to speed up. Because if you have that capability where you don't need humans to be there, then you can work 24/7. Everything changes 365 days a year. And you get back to what I talked about last week, which is we as humans work 34.2 hours a week according to the most recent payroll numbers. Let's take that number down to about 20. When you strip out the cigarette break, the lunch break, the talking, the chatting, the flirting, the whatever else goes in an office that people do, you end up with a situation of 20.

hours. Robots are going to work 24/7. Let's get to 168 hours a day. You're dealing with about 8 and a half more productivity just by that. This is going to be a dramatic beginning. This is just the beginning.

Uh, they talked about data centers and how they're transitioning, and I agree, they're becoming factories. I will highlight some of that. Uh, and you have to think about AI factories, not buildings anymore with servers. The change is happening now.

So, best proxy for you guys this year. I've talked about it, but we finally got a massive breakout, which says to me the scarcity trade is on. Long scarcity, short abundance proxy: long Chevron, short Salesforce.com, both around the same market cap. And if you think this is crowded, oh, you're wrong.

Here's the trade over the last 15 years. Think of this as, oh my gosh, this is when the capital account surplus really started to take off. The US outperformed the rest of the world. The smartphone came into existence, and we had a software dominance with no hardware. Oil collapse. China stopped growing, and now we're getting the renaissance. This is the beginning of the unwind of this. And maybe it only goes to here. I don't really care. The main point is from '72 to even here, which might happen this year, that's a big man. Long Chevron, short Salesforce.com as a simple proxy for you for scarcity over abundance.

Now, ChatGPT Health came out. This is a major deal. Not a minor deal. This is a major deal. We're almost up to 20% of all spending, nominal GDP, is now healthcare. I, you cannot even think of the importance of this for everyone out there. And I'll go through some of the reasons.

As someone who is very focused on anti-aging and very focused on this, and already does effectively what this thing is, which is, I already upload my blood work, which I get done every three months. I already upload any information that I have, all my historical data that goes back to 1993, I believe. All there so that I can have conversations. And then when anything happens where I have a knee pain or anything, I update it every day, almost like a journal on what's going on, so that it literally is my doctor on anything that's happening. I use it already. It introduced ChatGPT Health, a consumer-facing tool aimed at helping patients. It signals AI and healthcare is moving from experimentation to infrastructure. It's not to replace clinicians, but it is absolutely to make it easier for the consumer and easier for the healthcare provider, uh, and for the insurance company. It has, it says 230 million users ask about health every week. This is a huge deal, guys. I meet dozens of AI health startups every week, and you can tell this is a big deal. Most of them will become redundant once this gets adopted. It will be disruptive. But for the bigger health companies, which are bloated, it's going to help for sure.

Now, they not only launched ChatGPT Health for the consumer, they also opened OpenAI for healthcare. The platform is rolling out at major health systems, including Advent, Memorial, Sloan Kettering, and Cedars-Sinai. It aims to streamline clinical workflows, reduce administrative burden. You get the picture. OpenAI, in the same week, acquires Torch, a startup focused on the healthcare sector. It's to enhance its capabilities within healthcare and its health platform. Anthropic, seeing this immediately, jumps into it, launches Claude for healthcare, challenging GPT Health. The AI show talked about ChatGPT Health and AGI co-uh, AGI, uh, code, sorry, Claude Code. The Claude Code moment signals the tipping point for AGI-level usefulness. ADI doesn't need AGI to transform work. This is really critical, guys. If you're waiting for a push-button for this, again, you have to get some training on this. You have to do something. ChatGPT Health launches with deep medical integration. If you want to hear them talk about these, go listen to it.

If you remember a couple months ago when I highlighted pharmaceuticals and said Eli Lilly, which will be one of my favorite, one of my top five names for this year, especially on a risk-adjusted basis, uh, meaning V-adjusted. Uh, Nvidia and Eli Lilly announced a co-innovation lab. Now, they announced a, one, a building a data factory back in November. May have been October, but I think it was November. Now they're doing a co-innovation lab. So, one of them is, think of it as their own LLM, meaning Eli Lilly's got all this data in all these silos for all these different diseases. That wants to use all of the data they have, all of the experiments they've done, all of the tests they've done, all of the failures, all the successes, and they want to use LLMs and AI to basically be able to compare all that data. But the data is all in silos. It's in different structured scenarios, probably in different software, who knows? Well, they need first the factory. This is going to be about the innovation lab where they're going to work together to reinvent them, to discover, to develop, and to manufacture. So, you're starting to see the acceleration that's happening. And I think this is going on again. And the reason I love the pharmaceutical names because AI drug discovery is right here, right now. And there's going to be major announcements, I'm sure, this year on this, based on what I've seen from Google DeepMind and Silico and Eli Lilly and other places as well. So, this is going through what happened. Yeah, it was October, November. Um, one was a data center. This one is a co-innovation lab. Shopify's CEO made his own MRI using uh Claude Code. China is involved in this too. So you're going to have advancements around the globe. They develop AI platform that achieves millionfold increase in drug screening speeds. A new AI tool could dramatically speed up the discovery of life-saving medicines. This is why this is so important.

Here's pharma relative to the S&P 1500 all the way back to 2001. Do you think this is uh, we had a big move up? I highlighted it was the largest move I think since around 2003. Uh, sorry, since around this period. Uh, this is the beginning of a long-term move. Pay attention to pharma.

But it's not just pharma, guys. Look what's happened to healthcare. This is the S&P 1500 healthcare sector relative to the S&P 1500 since ChatGPT launch. Again, this is a relative thing. We don't want to be long any of this. So, biotech underperformed. Now, you're getting a bottom here. I would suspect that healthcare is going to outperform a lot of the sectors. It's one of my favorite ones.

I finally launched the first uh, or my HRV Substack. So, the second Substack I have for those of you interested in HRV, interested in learning about it. I do believe it is a metric for aging. I am going to spend every week for many years just highlighting all the different things I did. At the end of all of that will be a prompt that is there for people to learn on their own. Uh, this is the prompt. So, this is a normal prompt for me. This is the prompt that is attached at the end of what will be the first publication next week. You can copy it from there and then you can go paste it. It will take you on a journey. The first week is about the importance of the microbiome for this.

If you guys watch Limitless with Chris Hemsworth and Peter Attia, uh, basically that whole thing was about HRV. They just barely said HRV, but they talked about the importance of breathing. They talked about the importance of strength. They tal all of these different things that all equate back to HRV. The problem is Chris thought he was healthy. He's Thor. He wasn't. They didn't show his HRV at the beginning, but I'm telling you, as someone who has gone through this. This is my Aura ring data. I don't have two. Well, I basically had the same thing. I told you I've gone from the 20s. This is where I am so far this year. Continuing what I was, I'd shown you that I'd been up near 70 even this year. Uh, I'm still at 70. Uh, but I've been rising higher. So, I'm going to go through in that how I used AI to get there. It'll teach you about AI. It'll also, if you're interested in health, go through it. You can find it at my Substack.

And then finally, well, not finally, but we're at the AI trending topics. I just wanted to highlight this is what I went through this week. I ran this AI agents, Claude Code, AI and healthcare, efficient models. I highlighted these because basically, guys, when I do these videos, I'm on top of everything that's trending. I'm on top of everything that's going on. The only way you can do that is to be listening to podcasts, be on X. You take that information, you put it into my format, and then you can generate alpha. As Gavin Baker said, investing is a search for hidden truths. You generate alpha by finding a truth before others see it. There is tons of information to find right now, real-time, for analysts and PMs and traders that are out there. All you got to do is this. Find some new alpha that's interesting that you haven't heard before. Do what I did in terms of the skills thing. You come up with an output. You put it into here. Once you get that, you generate reports that go into a deep research report. Once you get that, all the deep research reports come in here and you search for ideas, and then you put them through a screener. That is the whole point of what I built. It's what I'm going to show. This is what each of them does. If you want to blow it up and read it, go ahead. Uh, that's why I'm spending my time on this. But it's so much fun for me to teach people how to use AI, and I've been having success with it, uh, that anyone that wants to get involved, again, call 22V, wait for the paywall, whatever the case is. You can do this with fantasy football, you can do it with finding a job. The whole theme of what I said with Annie Duke, which is AI will help you make better decisions. You just have to get the context right, and it'll lead to a better decision.

Finally, on crypto, the banks are not giving up yet in fighting crypto. So, Coinbase CEO Brian Armstrong said they're trying to kill the competition. The bill is still being held up. Despite that, Bitcoin did have a breakout week. We broke out of the channel. So, basically, we broke this trend line. We broke the 50-day moving average, 20-day moving average. We're still below the 200-day moving average. That is the final line in the sand. And as I said last week, this line here is right around there, $100,000. So, I think we're going to have trouble getting out of this. We get the Clarity Act, we break above, it's there, but we're building a bottom here. In my opinion, I've been heavily involved in terms of the last two weeks. We did do three-day closes above uh, 92, which was my trigger point for me in terms of thinking that everything would go. I love the way the chart looks. It's kind of reverse head and shoulders. The retest makes sense. We'll see what happens. If we fail and go back into the range, it's just going to take a little bit longer, but we do have the MACD weekly ready to cross. So, if this is a bottom and we get a cross down here, I fully expect not only new highs, but then confirmation as we get up towards the new highs of it being the best performing asset. It has already outperformed the MAG 7 this year by 10%. So, already this year, in terms of Bitcoin's move, very quickly, if it got up to the all-time highs, just so people hear, sh, it will have outperformed the S&P this year and last year. So, we'll see what happens. Ethereum also ready for a MACD crossing.

All right, guys. That's it for this week. Uh, hope you enjoyed it. Keep reaching out. Hit the subscribe button. It helps me out, trust me, more than you know. Keeps me going on this. Go to my HRV Substack, and the paywall will be up. Hopefully, I connected on the Elon side. And then I hope to see a bunch of you. I know I will, down in Miami with Raul and the gang. So, have a great uh, extended weekend, the holiday. And I'll see you next.