How to Profit From the Coming AI Boom ft. Beth Kindig

0:02Hey, everyone. If you like this podcast, go behind the paywall to get privilege access to the smartest minds in finance. Visit realvision.com slash rvpod and use the promo code podcast10 to get 10 % off our essential membership for the first year. Join the Real Vision community and learn how to become a better investor. And now to today's episode of Rao Pal Real Vision.

0:36so we've had all the experts on over two weeks to give you some sort of insight into what the future looks like here is it a boom or is it a bust and these are really important questions. And we've covered it from many angles, from commodities to macro to economists, giving everybody's view. Now, if you remember in the beginning, I kicked this whole thing off with my thesis. And one of the core parts of my thesis is that I'm bullish. I do expect volatility. I don't think things go up in a straight line, but I'm generally more bullish than most. And I'm very bullish on technology and extremely bullish on crypto.

1:17So I really wanted to get Beth Kindig on. She was the girl who nailed it for NVIDIA on Real Vision. She's done extraordinarily well in predicting and forecasting tech. And I really want to get her ideas of how to play this and figure out, does she think it's a boom or a bust? Join me, Raoul Powell, as I go on a journey of discovery through the macro, crypto and exponential age landscapes. In The Journeyman, I talk to the smartest people in the world so we can all become smarter together.

1:57Beth, fantastic to get you back on Real Vision. Thank you, Ralph. Great to be here. Yes, the first time you and I have chatted, but you've been on before you made one of the great calls of Real Vision in the recent years with NVIDIA. So I'm really looking forward to having this chat. Before we get going, let's go to your background. How did you get here where you are today? What's your story? So 10 years ago, actually about 13 years ago now, I landed in Silicon Valley and I was positioned called a tech evangelist or developer advocate, which means that you package product and you describe the products to a technical audience and they then hopefully adopt your platform.

2:37Or another use of this role is mergers, acquisitions, and very large eight to nine figure partnerships. So let's say a company is going to choose a mobile application security company. How do they go through dozens of mobile application security vendors and choose which one to move forward with again for an eight figure, potentially nine figure vendor partnership. So essentially, I was doing that for six or seven years, and it was brought to my attention that where that competitive analysis would be very valuable would be the public market. So to give you an idea, within three years, I was on the largest stages in Silicon Valley.

3:21I was presenting at Black Hat, Games Beat, Tech Week, Chicago, multiple conferences basically throughout throughout Silicon Valley and other areas such as New York, Advertising Week, New York, those kinds of places. And I was getting press because of the ability to basically say, out of all of these cybersecurity vendors, I'm just giving you one example of a project I worked on. This is the one that has the most cutting edge technology that will give you a competitive advantage. Well, this happens to be very useful for public market investors who need to first and foremost navigate the competitive landscape there are so many products on the market it's very hard to sift through uh how those products you know what is the what is the competitive advantage of those products but then also which ones are most likely likely to succeed uh so in 2018 i began to write for the public markets and it started to go very well right right away my One of my first articles that I ever wrote in 2018 was that NVIDIA had two impenetrable moats.

4:28One was the GPU-powered cloud and the other was CUDA and that they would become an AI powerhouse. And I repeated that publicly for years on tier one media, podcasts, things that are recorded and archived. And then another one was that Microsoft Azure, which at the time was not getting much attention, would compete alongside AWS due to hybrid cloud computing and their inroads to enterprise. These are just some examples of some of the very first work I did for the public markets. So once all of that started to materialize, you know, it was encouraged that I launch a research site, which is fast forward where I'm at today.

5:08I've had a research site for four to five years. We've been audited through a third-party accounting firm, large one in San Francisco, and our results far exceed basically most other ETFs or tech portfolios, particularly around the fact that I do know tech and that my skill set is how do you sift through all of these products on the market and recommend only one and make sure that that one that you've recommended performs in the future. Perfect. So when you're thinking about technology, what's your overarching thesis of how do you approach it and how people should approach it? I do agree with the thematic investing approach.

5:50It's very venture capitalist inspired. And because I was in the private market, that is an excellent way to begin. I think where the problem is with some of the more thematic investing is that there's not enough granularity, again, into number one, the competitive space or the competitive analysis. Number two, the fundamentals, especially right now, need to be very, very clean and very straightforward. You know, those are a couple of things that we look at. And then third would be risk management. So within something like AI, which again, we've talked about for five years, we had a 45 % allocation to AI going into this year.

6:31People are talking up Stanley Drunkenmiller for 29%, which is great. But just to give you an idea as to how patient we have been in building this. But at the same time, every AI stock that I've owned going into this year had an excellent bottom line. And so I don't give tech companies any leniency around the fact that they have to be a really solid business as well. And then, of course, making sure that they are on track with their product roadmap, that they're not over-promising, under-delivering. There's all kinds of flags to watch out for. But then, of course, tech is by far the world's most valuable industry, so it's well worth the effort.

7:12So you tend to approach it bottoms up company by company as opposed to a broader thesis on why tech matters and how it outperforms or anything like that. So you're just you're generally focused bottoms up on the on stock picking. Yes. Although I will say that overall where we we will pick a theme. So a great example, I guess, if we could all travel back in time, would be mobile. And I'll tie this into how I look at things today. if you could go back in time around 2010 and pick a tech portfolio, there's no reason to own anything but mobile. Mobile gave us the native application of Facebook, which is when Facebook exploded.

7:50It gave us a search engine and Google Maps in our pocket. Google exploded at that time. Of course, Apple is mobile. So there's no reason to go looking around and spread yourself too thin here because mobile was the place to be. Now within mobile, how do you find those big winners? So for us, it's AI. And that sounds very buzzword today. But I do want to just reiterate, this has been a thesis of ours for many years on record, archived. But within that, this is, in my humble opinion here, the number one investment opportunity of our lifetimes, the generation before us, their lifetime, the generation after us.

8:31This is an enormous opportunity. And there's nowhere else to invest right now, in my opinion. We do add some diversification, but it's like I said, we had 45 % AI exposure going into this year. When we look at pullbacks, our plan and strategy is to continue to build AI. I'm the other way around. Crypto is my bigger bet than that, but AI and a bunch of these technologies that I refer to as the exponential age, this nexus where AI is almost a foundational layer now on top. you know this kind of it's and everything gets built on top of it like you i think it's a cambrian moment i think that the generous of ai that's going on right now is as big a breakthrough as we've almost ever had maybe along with i know it sounds ridiculous but along with the splitting of the atom this this kind of creating infinite knowledge is something that we can't even yet get our heads around and it ties into everything people just think of it as chat gpt or a few things, but it's not.

9:31I mean, it's literally going to power our worlds. How do you think of AI in the broader perspective? So one thing that I boil it down to, I guess, is that the impact it will have on GDP as what you're describing is unlike any technology in modern times. And it's because it will drive down costs, increase productivity. So the contribution there is infinitely higher than something like mobile. And the reason I anchor a lot back to mobile is because we all know that the FANGs, if you could have invested 10, 15 years ago in the FANGs, you would have. So what I'm describing is four to five X larger than the FANGs.

10:11So of course, what we're going to do, what will be the result is some massive winners, just like I would call the FANGs massive winners within mobile and AI will blow mobile, you know, away in terms of it's not only contribution to GDP, but you have a$15 trillion market. Mobile was about a$4 to$5 trillion market. Immediately right out the gate today, we have 3x, but I think that that's very low because it's not incorporating all of the software innovation that we can't even imagine yet. But given the numbers we have today, because people like to think of AI as a hype, what I would encourage people to realize is that in 2010, mobile was not a hype.

10:57And we're probably more like 2008 or 2009 right now in terms of where we are with the vintage of AI in terms of where it's going to go. So just keep all that in mind that if you believe that mobile would have been the right way to position, then AI certainly will be because it's so much more massive in terms of its contribution and size. Also, what's really interesting about AI is all of the others, mobile, internet, there was no infrastructure in place. So it took time to scale. But what we're seeing is AI just fits over existing rails. So the growth of AI is staggering. I mean, we saw the zero to a hundred million chat GPT.

11:40I think it was six weeks. We've never seen the technological adoption of this speed before. Yes, agreed. And a lot of that is coming from NVIDIA right now. I think there will be competitors, but we have NVIDIA sitting right there. And then of course, you know, some of these other projects, OpenAI, DeepMind, obviously they're very instrumental, but for the public markets, you know, NVIDIA has really provided the H100, which cannot be overstated in terms of its importance for how quickly AI is growing right now. Now, yeah, let's talk about some of these stocks. So NVIDIA, obviously, it's moved a lot fast.

12:27How can you think about this stock in the future? How do you discount some of this? How do you figure out who else could walk into this market eventually? Because obviously, they've got such a stranglehold on everything right now and there's such a backlog of chip orders but how do you how do how do people even get to grips with nvidia right now or should they just not and wait for some sort of correction or wait for something to lean into we are a very big fan of waiting for corrections uh one of the best things we did was not only recommend nvidia to your to your subscribers in january but we actually bought in october at the low and we issue real time trade alerts so all of this is in real time and it's on thousands of people's phones so all very proven.

13:11And we bought at the October low because we're very patient. We are watching this to determine as well, like you said, do you just add, you just close your eyes and add? Or should we be a little bit more patient, wait for that pullback? I think it also helps to put into context just why NVIDIA is so important, which is I would almost say their real AI moment was actually a couple of years ago with the A100 because they combined inference and training onto one chip. But now the H100, the reason why, to me, that was their AI moment. That was when a lot of their R &D and a lot of their success, basically, and their product releases were now cornering a market.

13:50And then the H100 has now opened up such an ecosystem of software development that it's a very special moment in time. People do call it the iPhone moment. I agree with that because of the development that can now occur. And a lot of it is not only that it, you know, greatly increased performance, but it has a transformer engine and that gets rid of having to label datasets. So you can basically find patterns mathematically. And that piece is why so many large language models are now coming to market. The T in GPT is for transformer. And so it's just this opening up of so much innovation. And, you know, you can actually train a model with a transformer engine in basically a day, which greatly reduces the time compared to what was available before.

14:50So it was a massive drop within the AI community and ecosystem in terms of where you can go now. And that's why I think there was so much excitement. But like you said, how do you pick the next winners? And I think that's where we're entirely focused right now is being very patient and building out this AI portfolio. What do you think of this NVIDIA Bearcase that's been floating around recently that their orders aren't real and all of that? I don't know if you've seen that at all. Yes, I have. And what I would say is one of the proxies we used about two years ago or three years ago was watching big tech capex which was growing very fast and was leading to basically down the funnel to nvidia and then we've if you sit on these calls with big tech companies google microsoft amazon meta they're all talking about allocating more and more of their capex towards ai acceleration so for me like you know this these are verifiable budget.

15:53This is CapEx. It's very easy to go look at these financials of big tech. All these management teams are basically saying, we're just trying to buy as many GPUs as we can possibly buy right now from NVIDIA. When I heard it and it came across my desk, my first thought was, well, we can track where their orders are coming from. They're coming from big tech CapEx, and it's been a strong proxy to use for NVIDIA's success for the last couple of years. So I don't buy it in terms of there being, you know, what's driving these sales compared to AWS, Azure, Meta's ambitions. And I'm leaving out Google Cloud.

16:33Hey, everyone. We're going to take a quick pause and hear a word from our partners. We'll be right back.

16:42I remember speaking to Ahmed Mostak from Stability AI, and, you know, he was just talking about, I mean, they're melting GPUs. there were people driving them so hard and having to order so many um he he got me into the idea of nvidia a while ago just by saying listen like it's a bottleneck everybody has to go through that pipe right now there's no other way around it it feels like do you think that changes over time or is that the one play for for that particular space and then we look at layering out the bet elsewhere in the space yeah i mean these are some of the most important questions i think to ask and it's something that we think about daily but i was asked on a webinar what's the next nvidia and i basically said the next nvidia is nvidia nvidia it's like apple though which is that there will be an android and so we think the android will be amd and uh we have a large position in amd because of we think it'll be um very strong contender i would call it i like to call it the dark horse which is an underestimated competitor that comes out of nowhere i like to give stocks nicknames.

17:50But I think it's a dark horse because it was with Intel. And if you look through what happened with Intel, it's mind blowing. And the media has never appropriately reported on this, in my opinion. And it's important to understand how AMD took market share from Intel because they have this experience of taking on the 800 pound gorilla. And now they're going to take that exact same strategy and they're going to bring it to Nvidia. So for me, I prefer not to choose. We have a very large position in Nvidia and a large position in AMD, and I hope they both win. But just going back to some of how AMD competed with Intel was all on design, all on architecture.

18:31They basically combined a seven nanometer chiplet with a 14 nanometer that allows the chip to use whatever is needed most at that time. It's called the multi-chip module. And they also beat Intel on memory bandwidth, just very power efficient. And then they put them in a headlock, if you will, if you're thinking of like a WWF match, by lowering the price. And it became a virtuous cycle because they came in with a more powerful chip, a more unique design that was able to be power efficient because it used the seven nanometer when it was needed, flipped back to the 14 nanometer when it wasn't needed.

19:06And that multi-chip module design was not only superior on performance and power efficiency, but it came in at half the price. And what does that mean? It means you can just load up and buy more CPUs. And so it was really the Zen 2 architecture. And they went from, when we first covered AMD, 4 % data center share to now going into the 30%. Just a phenomenal comeback from Lisa Su. So the reason that I think this is getting very exciting because the MI-300 will come out towards the second half of this year is because her and her team are coming at it again with an interesting architecture, which is a shared coherent memory.

19:53The big headline around the MI-300 is that they were able to run the Falcon. It's like a 40 billion parameter large language model. They were able to run that entirely in memory. So they didn't need to go back and forth with the external memory. That was really interesting because it's bringing the heat on NVIDIA in terms of just pure hardware. Where AMD has a lot of work to do, so just to be really clear, NVIDIA's H100 is a marvel of innovation. This thing is incredible. If you hear Jensen Wang on these earnings calls, he'll describe how they make it. It's a huge piece of equipment and there's tons of software pieces and it takes a robot to build.

20:36AMD can compete on hardware, though. And they are running the world's largest supercomputer right now, and they're launching the next largest supercomputer. So it's the El Capitan, the fastest computer in the world, is run on AMD. But where they struggle is the software development platform. So Rock M is an open source development platform that is hoping to compete with CUDA, and it is very unlikely to have the entry that this is where I, let me just rephrase it. This is where the friction can be because software developers have learned NVIDIA's CUDA. So it's more of the software side that I think AMD has work to do.

21:21So we'll see how that goes. But those are some of the bullet points on why I'm bullish on AMD with a caveat that we have to watch how successful the open source development platform Rock M is. So let's move a little bit lower down the food chain. How does this affect stuff like Taiwan Semi and all of that side of the equation? Yeah, Taiwan Semi is really the only fab in the world, foundry in the world that can produce these advanced nodes. So when you talk about this bottleneck, a lot of the bottleneck also occurs around Taiwan Semi. and there's so much geopolitical implications to Taiwan Semi that it's, this is just the world of semiconductors that some people choose to just go towards software, but it's hurting you to go to software because AI is being innovated at the hardware level.

22:15But we actually have a small position in Taiwan Semi too, because it clearly all roads lead to Taiwan Semi conductor with, the 5 nanometer, 4 nanometer, 3 nanometer. The sooner that Taiwan Semi can get on the United States soil, probably the better. But overall, the bottleneck that you've spoken of is also occurring there. So let's go further up the food chain. Okay, the applications layer of this stuff. Where do the big players, Google, Microsoft, Apple, how do they all fit into this architecture? If Taiwan Semi is the kind of, not Taiwan Semi, if NVIDIA is kind of the epicenter of this revolution right now.

22:52Taiwan Semi is like at the foundational layer because nobody can do anything without them. Then we've got the big guys with their large language models and what they're trying to roll out across all of their various applications. How are you thinking through the big players? Yes. So, you know, I would say that is also one of the best questions to talk about because AI will test tech investors because tech investors have gotten comfortable with almost the opposite because mobile is so consumer driven. AI will be very enterprise driven because the most immediate impact that AI will have is driving down costs, increasing productivity for enterprises.

23:31There will, of course, be some consumer overlay. But what is really ideal for a stock or for a company is if you take a consumer facing company like Google and they can inject their AI technology into the ads machine or their Google search. So they don't have to go out and go try to get, you know, lots of consumers to adopt something new. Consumers will simply continue to use search. It'll just be improved search. Advertisers will continue to use Google. It'll just be improved ROI, easier to find your targeted audience. So those pieces are really important because in the past, these innovation cycles such as mobile were more democratized.

24:13So anyone could really create a mobile app. There were millions of mobile apps. Anyone could create a website. There were hundreds of millions, I suppose, of websites. But where will we actually, where's the product market fit? Where's the market for AI? And that's where I think big tech has a serious, undeniable advantage here, which is that they are not going to create an app and go run around and try to make people adopt it. Obviously, chat GPT was fun and interesting, but Google doesn't really need that. They can simply improve their own ad machine and their own search engine as is not have to change behavior.

24:54Microsoft is a very interesting company across all big tech because of its connections with enterprise. So if you're going to drive down costs, increase productivity, who's sitting there best to do that would be Microsoft's suite of cloud products. So again, rather than going out and saying, brand new, everybody adopt this co-pilot, we're just going to add this to Office 365 and integrate that into our product suite. So there's that friction that I think will be very real for startups or other smaller companies, which is how do you get people to adopt AI is not such a problem for big tech because they have the perfect product suite to just bolt that on and prove the revenues they already have from the customers they already have.

25:46And so those are some of the things that are really important. And then, of course, another piece about Microsoft that a lot of people overlook is that that investment in OpenAI, basically because those APIs are run on Azure, there is a nice funnel or, you know, Azure is downstream from all OpenAI APIs. So the more the OpenAI is successful, the more Azure will be successful from some of that usage. So there's certainly some nice connections there as well. The other side of this equation is obviously AI can create infinitely more data for storage. How do we think through? So we've gone through the compute side.

26:25What about the storage side for all this stuff? Because it's just like monstrous amounts of exponential data all the time now. Yeah, big data and analytics, huge need. What I will say, so the exponential rise of data for our portfolio purposes, just to be super transparent. we don't think it's time just yet to build those positions. We're getting very close. If you listen to some of these big data analytics, earnings calls, they'll be quite clear they don't have AI revenue yet, although they're situated really for AI. And one of those reasons, if I will, is that big tech is the biggest producer of AI right now.

27:09So, you know, like we just talked about, and they have a lot of their own big data and analytics products and platforms. They don't necessarily use a Snowflake or a MongoDB within their own systems, so they have their own already. What you would need is startups and smaller enterprises, SMBs, whatever, all along those sizes of companies to really explode on the AI trend before the public stocks today are really seeing that new trajectory. And that's why NVIDIA had this crazy, you know, 200 % growth in data center is because their chips are right here right now in extremely high demand. So in my opinion, when big data and analytics layer is ready to pop, I don't know if it'll go up 200%, but you'll see a very strong inflection point at that moment.

28:05We don't have that inflection point yet. So I'm trying to be a little more patient. I'm wondering if we'll get one of those big sell-offs that tech loves to have massive sell-offs. So I like to have a couple ideas in my pocket for when those sell-offs come down the line. Big data and analytics is not something I have in the portfolio right now. Certainly, these will be needed, especially the more different types of data for machine learning models. but if you look at some of the snowflakes the confluent the mongo dbs we're not seeing that inflection point yet and i don't think we will until ai broadens away from big tech the one that i've really interested in in all of this kind of nexus thought i think the company that sits in the middle of all of this is tesla um you know even when just think about the uh the data storage issue i think it was elon yesterday somebody was talking about how much when Dojo fires up fully and starts processing all of that video data.

29:06He's like, yeah, the compute side is pretty hard and it's a fight for GPUs and we have to develop our own hardware, etc. He goes, nobody thinks of how expensive and difficult it is to store all this bloody stuff. When you've got 5 million cars on the road recording real-time streaming data that needs to feed an AI, it's unprecedented amounts of data. That's very true. Tesla has a lot of data. And I think the one thing I would say about Tesla is they also have a lot of competitors. So, you know, whether it's the auto OEMs or whether it's the autonomous driving systems or whether it's NVIDIA, which to me is a very hard competitor right now would be NVIDIA with Dojo or Automotive.