Ice cold, zen-like investing with Alex King

0:20Alex King:Thank you.

0:30Alex King:analysis about what's happening in these markets. We've had Tesla earnings. We have a government shutdown. We have quantum and government and so many things to pick at and decipher. What are you thinking about most of all? What are you looking at at these markets? Yeah, mostly I'm focused on maintaining a Zen light car. Good luck with that. Yeah. But I think that's the thing right now. I mean, it's really easy to get caught up in the noise. There's a huge amount of noise. If you spend any time at all, on social media, particularly X, you get buried. And it's really hard to focus on things that matter.

1:08It's hard to get a fix on, you know, is the market an uptrend or is it likely to roll over soon? It's really easy to get distracted by politics because, you know, whether you are vehemently pro, vehemently against the current administration, that tends to colour your own assessment of investment opportunities. And of course, that's not where you want to be. You want to be cold and distanced. So truthfully, you know, my own focus is on that, which is just try and keep out of the foray and focus on, you know, price, volume and where appropriate fundamentals and just keep it really, really as cold as ice.

1:43I think that is the that's the trick right now.

1:46Alex King:What would you say about the tech space right now that you're looking at in terms of the obvious players, this AI bull market that people are waiting for it to explode? Other people are waiting for it to reach ever higher. What would you say about the AI part of things and also how much those names are driving market sentiment these days? Yeah, I mean, clearly the top seven stocks in the S &P are powering the S &P. And there's a great Howard Marks podcast out this week or last week where he basically opined that, OK, the MAG7 are not particularly ridiculously valued, but the things that he worries about are the other 493 stocks.

2:29I think if you look at the AI boom, if you look at the unit growth, in other words, You know, GPU shipment, server shipments, data center builds, kilowatts of power provisioned. That's going to keep going up. You know, I think we're early in the demand cycle. And we're sort of at the stage where lots of companies, large companies have started to adopt AI. They've dashed into it, as they should. Struggling a little bit, I would say, with use cases. Struggling a little bit with the true economics. So which tasks can you use an NLM for and which not? Which headcount can you get rid of and which not?

3:09And any white-collar employee at a large company is figuring out how do they take advantage of AI and not get swamped by it. So I think in the real world, it's early. It's very early. The technology is, when you first start using it, it's really impressive. And you don't have to be too far into it to realize its limitations. So I'll give you a very simple example. We obviously produce a ton of earnings analyses. And what we would love to do is have an LLM, the minute earnings printed for any of the companies we cover, we cover 50 or 60 stocks, basically go and compile a custom spreadsheet based on how we like to prepare the numbers.

3:50and then say, you know, hey, the spreadsheet's ready, now you can write your thing. Not enamored with the idea of AI actually writing the stock analysis. It tends to be too generic, but it should be able to fill the spreadsheet. And so far, I've got to tell you, it just doesn't. So we tried many tools, all the big LLMs and some niche tools as well, and it's still not there. And so for what you would think would be a relatively simple task, which is go and look at a press release, pull some SEC reports from the web and compile a spreadsheet which looks the same every quarter, it's still not there.

4:23So it's early. And again, in the real world, we're probably going to enter the trough of disillusionment stage soon where people get frustrated about what it can and can't do. And very often in technology cycles, that's followed not too far behind by a drop in valuation multiples because excitement about technology falls. And so I'm old, as is obvious. so I've been here for the dot-com boom, the mobile boom, the social media boom, all of these things and they're all basically the same which is everyone correctly gets all excited about new technology, bids the stocks up, reality lags for a while, the stocks fall back, reality quietly catches up while the stocks are on the floor and then all of a sudden people wake up to them again and the mainstream wakes up to them again and then the companies and the stocks take off.

5:11So I have no reason to suspect that the AI cycle will be any different. If you look at the stocks right now, you know, is NVIDIA expensive? You know, it's not expensive based on its growth and its margins. It is expensive if you think it's a monopoly that a lot of folks are gunning for. At some point, someone's going to succeed in deflating its market share, deflating its margins. So the trick with all these things, I think, is don't assume anything is forever. Learn to use stock charts. There's no more reliable measure of price than price itself. And just don't be swayed by a narrative. Just look at price behavior.

5:49If price is going up, well, then price is going up. If it's going down, it's going down. And just ignore the hype one way or the other.

5:56Alex King:We had Kirk Spano on from Margin of Safety Investing, and he was talking about how the LLM model will be commodified at some point, and that will really change the investment thesis, or inherently, it's a different investment thesis than people may think it is. What would you say to that notion as it develops? Yeah, I think that's a smart insight. If you think about the LLM, I'm going to slightly torture comparison this, and pure techies will pull me apart for this, but if you think of the LLM as a kind of operating system, and obviously it isn't an operating system, I understand that, but a kind of platform upon which applications can be built, then he's probably right, which is there will emerge two or three dominant by market share LLMs, and people will write domain-specific applications on top of those LLMs.

6:51So you'll find apps for Gemini, apps for ChatGPT, and so on and so forth. So I I mean, that's probably correct because, you know, the time to put huge dollars into building, training, nuancing a brand new LLM, that's probably in the rear view, at least in the current incarnation. I have a one thing that interests me is if you think about the way that LLMs are built right now, the software design is what leads to this colossal compute and power requirement. So right now, to be very simplistic about it, what does an LLM do? It comes up with a pattern, and it says, based on everything it knows about the world, what's the next value likely to be in that pattern?

7:39And it makes educated guesses. And it's a bit like in cryptography, there's a phrase called a brute force attack. And it's a bit like a brute force attack. you have to deploy an enormous amount of compute to make an educated guess as to what the next thing that's going to happen is. And I suspect that the thing that will change the cost model for AI, the thing that will lead to just a reduction in power, a 10x reduction in compute requirement is a different approach to the LLM. And what I mean by that is, if you think about two different people, a sort of formal rational person and a very emotive person, a formal rational person in the conversation will follow the conversation literally and will try to think about what is it that a rational response should be to the thing that was just said to them whereas an emotive person will tune into all of the cues and signs that are not in the words and it seems to me that the and they'll respond more fluidly more naturally and with less effort you know if you're a formal rational person you come home from a day at the office much more tired than an emotive person comes home because it's been hard to work for you and i think llms are a bit like that so i suspect that coming down the path at some point will be an llm that's built on different methods better understanding of the patterns they're looking at you know stock trading algo is a similar you know we develop and sell a whole bunch of algo services and you know they essentially do the same thing they look at prior patterns and they try to guess what the next price movements are, but they're pretty computationally intensive.

9:18At some point, the world's going to run out of power and compute, and the model has to change. So I think in the end, the thing that changes this bottleneck you're seeing, I mean, you just can't get enough power provisioned in the United States. It's not possible. You can't get it quickly enough to the data centers that people want to build. It's not possible. You can't ship enough GPUs quickly enough to the people that want them. so that has to change so either ai falls on its feet no sorry falls on its face and just goes away i don't think that's likely or something changes so i think that llm design could change and that would then you know not be a commodity it would be a different kind of platform and or we could see much more efficient silicon and um and silicon programming software by which I mean, obviously the CUDA platform that NVIDIA uses.

10:07So you can see Broadcom, ARM, and other people gearing up to sort of pick away at NVIDIA's market share. And, you know, if ARM gets to play a part in that, ARM is much lower power by design at its core, has been for 30 odd years. And so that's another way, I think, where the current setup of vendors, the current setup of buildup could get deflated. So the thing I would look for is watch price in all these names. At some point, this roaring bull market will correct. Of course it will. That could be tomorrow, could be in a year, could be five years. I don't know. But of course it will correct. So you have to watch for that.

10:45And secondly, if you look at changing the guard in vendors, NVIDIA has been dominant for a long time now. All empires fail and NVIDIA will be no different. And when it comes down, when it comes crashing down, it'll be at the hands of somebody that delivers 10 times, I think, power performance. Traditionally, in semiconductor, it was a 10 times improvement in cost performance. But this time, I think it'll be power performance because power requirements translates to total cost of ownership for the data center vendor. So I think there's plenty of stuff to watch out for. But I don't think one should convince oneself that because all these things are likely to come down the path, it's all over for NVIDIA and all these data center stocks now.

11:22You know, while the price is going up, again, the price is going up. You have to separate price analysis from things that may happen in the future and just have those on your radar that might hit price in the future.

11:34Alex King:And then do you have any thoughts? I'm sure you have some thoughts, but any opinions you'd care to share on this government, on the US government getting into quantum names? I don't, to be honest. I mean, Quantum, you know, the stocks have clearly been on a tear. And it's a bit GME-like, you know, there's a retail enthusiasm for them. They've been driven up on momentum alone. Yeah, that's fine. You can make a lot of money doing that as long as you know what you're doing. They dropped a lot in the last few days. I made a little bit of money shorting those last couple of days. I got out, thankfully, just before the government announcement because I thought, well, short gains in the bull market, they're usually ephemeral, so don't be greedy.

12:20And, you know, is quantum a technology of the future? Probably. The question is whether these smaller vendors, you know, INQ, Rigetti, all these D-Wave, all these names, whether they'll be the leaders or whether you're Google's, IBM's, Microsoft, whether they'll be leaders. And, you know, I don't know. I think that if you listen to the founders of those companies, that they'll all tell you that the quantum advantage is quite some way away. I know Google made some announcement yesterday about having found some quantum advantage. So I think it's one of these things where in 10, 20 years, we'll look back and go, there you go, those are the killer apps.

12:55But we're not there yet. And so I think that the stock prices, even if they're supported, let's say this federal government investment comes to pass and it actually happens, they'll get some support from that, as has Intel. But that doesn't change the fact that the prices are far, far above fundamentals. So, you know, I don't plan to invest in any of those stocks personally. If they're runaway successes, that's like I can live with missing them. The risk of reversion to any sort of fundamental basis is so huge that for me it's a keep away. And congratulations to anyone that enjoyed the run up.

13:26It's been tremendous. But I think one has to be careful at these levels, would be my take.

13:31Alex King:Speaking of being careful, what would you say about Tesla's earnings this week? It had a bit of a drop post earnings, some questioning about what's coming down the line and how much excitement there is to be had or how much excitement is warranted. What would you say to Tesla bulls and to Tesla bears and to just market observers about Tesla? Well, I own Tesla stock. It's not a particularly large position, but I do own some stock. And I own it because I think they're going to merge it with XAI. And I think that that's a sort of a no loser scenario. you know tesla as a car company obviously has challenges its leadership in ev is slipping federal tax credit situation is changing um any you know you can't come up with tesla's price based on any comparison to the fundamentals of the the car or the energy business but if you think about xai you know xai is a pure play llm business and right now most people can't invest in it.

14:34So if you merge Tesla and XAI, and I'm pretty sure they'll do this, then right away, you give the market a pure play, vertically integrated, this is how they will describe it, of course, a vertically integrated AI, robotics, LLM, and probably Twitter's in the mix there somewhere, company to invest in. Well, most people can't invest in OpenAI or any of these other names. So I would imagine that will get some pretty successful capital markets marketing. I think it's to the benefit of the stock. If you're an XAI shareholder today, you have some liquidity, of course. These high-profile, highly valued private companies are not completely liquid, but you don't have the sort of liquidity of a public stock.

15:16And so I think if and when the two companies are brought together, XAI shareholders get liquidity. They get all sorts of other securities that will be created based on the underlying common stock. And Tesla shareholders ought to get a re-rating based on the excitement of the new model. So that's my logic for owning it. The earnings of the vehicle company, I think, are somewhat incidental to the stock. That's okay, many such cases. So I think Tesla has a bright future, but I think it's based on a combination with XAI. That's my opinion.

15:42Alex King:Were you surprised by anything in the earnings or what was said in the earnings call? Not really. I think it's a pretty mature setup. I think everybody who is a long Tesla shareholder understands that you're not investing in an auto business. You're not investing in an energy business you know you're not investing in a you know optimus robotics business you're investing in musk and musk will succeed for as long as he succeeds and then like everybody at some point he'll stop succeeding um but if you look at the discussion over the comp plan right now so i voted for and i voted my shares in favor of all of the board's recommendations and it puzzles me as to why anyone would vote against really because i mean is it is it corporate governance 101 have your ceo capable of earning a trillion dollars which is unusual but would you teach it in a civic ethics class probably not but you know is it a good idea as a shareholder to have the ceo highly motivated to get the share price moving up i think it is so So I think Tesla bulls, Tesla holders are there for the Musk lottery ticket.

16:54And Tesla bears, who have been bearish on the same point for a very long time, are of the opinion that, well, it's just a car company and a failing one at that. You know, it's losing market share. The vehicles are old. Where's the full self-driving? Blah, blah, blah. And, of course, they're correct on fundamentals. But, you know, fundamentals are only peripherally related to stock prices. And Tesla, more than most, is not driven by fundamentals. You know, it's a more mature version of the quantum stocks we've just talked about. So no, there's no surprises specifically for me. I think the stock has a bright future, but I don't think it's because of, you know, cheaper electric vehicles.

17:27I think it's because of an XAI merger.

17:28Alex King:Anything that you would say about what's happening between gold and the broader markets and this run that gold and silver have been having and the run that the markets have been having? anything that you would provide contextually or or highlights that you feel like my investors might not know enough about yeah um i think gold's gotten ahead of itself in my opinion so for disclosure i made some money on the way up in gold we have a an etf rotation algorithm that we use that was long gdx and gld for quite some time so that did really well and then i made some money on the way down in the last week.

18:08Again, I got out, I mean, two days ago using the GLL, two times short ETF, and then the short AIU spot price ETF. Gold, I think, is a fear purchase. And the thing that made me take a short position, it only held for a few days, but it did quite well, was the pictures of people in Australia queuing up to buy physical gold. I mean, this is just a complete overreaction. If you look at what's happening around the world right now, there is a reconstruction of the post-World War II world order. And the pieces will land where the pieces will land. Nobody knows. The actors don't know. The general population doesn't know.

18:48Journalists don't know. Political analysts don't know. No one knows. But it's changing. And so I think the gold purchase is a fear-driven reaction to that. People will say, oh, inflation. But there isn't any evidence that inflation is rising. More likely, inflation's cooling. The Fed's held rates at a pretty chunky level compared to recent history, not compared to long-run history, but compared to recent history, fairly high level for some time now. There's some evidence of some weakness in the real economy. So I don't think that inflation is a big risk. I'm not saying it can't tick up, but the idea that inflation is going to moon, there's no data for that yet.

19:28You have, I think a general worry driving it. And again, I think it's fear of change. So people will point to tariff policy or the China policy or Russia policy or any number of things that are unfamiliar to them. And there is a dash to something that feels comfortable, which is physical gulp. And I think it's gotten ahead of itself. So of course it can go up, it could double tomorrow, who knows. But I think that, I think it's premature. So yeah, we'll see. But if you just look at a chart and a spot gold, I mean, charts like that can't keep going up at that rate forever. They just don't. Probably there'll be a reset and then perhaps another move up.

20:07I don't know. But I think this recent dramatic run to the upside, that's a bit of speculative fervor. Another way to look at it, and again, in the spirit of staying ice cold, for anyone who has access to stock charge, you can do this within Seeking Alpha. Don't just look at the price, but look at the volume and try and look at the volumes transacted in gold futures or the gold ETFs up here at the highs. It's tiny. So what you don't have up at these levels is big institutional buying. You have retail and small momentum players buying. That's not evidence of a big rush to gold from the institutional investor base.

20:46If anything, if you look at the volume profiles, that happened some time ago, and I would expect them to be the large investors to be starting to take some profits around this time or at least take some hedges. This sort of goal continuing its race up, I don't buy it personally. We'll see, but that's my take.

21:02Alex King:It's your take. You're not alone in that take, I would say. A lot of comfortable positions on that side of the aisle. Earnings-wise, we talked about Tesla reporting this week. We have Intel coming up today. It'll be after our conversation, obviously. anything that you would say intel has been in the news a lot also speaking of government uh help or intervention we've got some government help with intel as well i know that you brought a semiconductor chart with you if you'd care to share that with our audience that would be um i think insightful for them and helpful how you're thinking about that space or that space within the tech space?

21:43So I'm going to show you the S-O-X ETF. So this is not Intel specifically, but it's worth, you know, with any single name stock, you always have to situate it within the overall market and also within its overall sector. So this is a slightly messy chart. So let's talk about semiconductors and then let's come back to Intel. So this is the S-O-X semiconductor ETF. I like this ETF. It's pretty well balanced between the top vendors. SMH is another popular one, but it's heavily weighted to Nvidia, which means it jumps around a bit more than SFXX. This is the April lows down here, post-liberation day lows.

22:20So it hit a 148 low. We use a couple of different methods for charts. We use Elliott waves and Fibonacci levels, as do many, for guesswork, basically, as to where over a period of time might the stock or the ETF go to. And we also use moving averages, as a matter of fact, to look at them as tripwires on the way down and the way up. Liberation, post-liberation day lows, I should say. So it hits 148 down here on April the 7th, which was the bottom for pretty much everything in the market. We then get the famous presidential tweet, great time to buy. Hopefully people took that as signal because one thing I would say about this administration, you know, like or dislike the administration, doesn't matter, you get plenty of signal from them.

23:03So this was a screaming buy point here. we got a big run up to 186 within a few days a normal sort of pullback and then we've just been in this tremendous run up for semiconductors from around the 22nd of april uh for let me see six months now yeah six months almost the day we've moved from 162 to right now 290 you know remarkable and um the question i think is at what point does semiconductor become a source of fun so Let me explain what I mean by that. If you're a large account investor, there are many ways to make money. But one way to make money is rotation. So you'll buy a sector at its lows and you'll wait for, or perhaps seed, some good news stories, have the sector run up.

23:53And you'll take gains at some major highs and then go put your money somewhere else. So at some point, when the sector has run up as much as semiconductor has, so again, I remind you, 1.48 April the 7th, 2.91 right now. When it's run up that much, you're going to see people at some point take capital out of that sector and rotate it into sectors that are not as run up. And so it's possible, possible that we see semiconductor as a source of funds and enterprise software, for instance, as a use of funds. So basically capital pulled out of chips and put into software because most of the software stocks, not all of them, but most of them are quite big now.

24:34Not unreasonably on fear of AI replacement and so on and so forth. So the big question with Semiconductor right now is, I think, has this topped? Arguments why it's topped? Well, it's at a big, big, big Fibonacci extension. So for those familiar with the system, this Wave 3 here is around about a 3.6 million extension. That's a big move up, okay? It also, on a much simpler level, it just can't really break S-O-X. It can't really break 2.9.2. for any period of time. It's been trying since the 6th of October and we're now on the 23rd of October. So for more than two weeks, it's been trying to push up the level.