ANEW - 4 Things To Consider When Looking For Investment Opportunities In AI

2023-11-17 02:15:00 ET

Summary

• Potential winners in AI beyond NVIDIA.
• Potential ways to play the rise of AI.
• Potential opportunities amid the rise of AI.

Artificial Intelligence has been a major trend this year in the markets, but which types of firms are best positioned to capitalize on it? MoneyTalk’s Greg Bonnell discusses with Dominic Rizzo, Portfolio Manager with T. Rowe Price.

Transcript

Dominic Rizzo - This really is an incredible innovation. And like you said, it's actually not that new of a technology. Everyone's heard about AI for a long time. If you look back at the original papers on AI, it goes back decades, actually. So why has AI captured the global zeitgeist today, right? That's what I always get asked internally. And there's really two key new technologies that are driving this explosion.

So the first one is GPUs, Graphics Processing Units, from the likes of NVIDIA ( NVDA ), primarily, but also new companies are coming up with GPUs, as well, companies like AMD ( AMD ). And the second -- and I'm sure you've heard this term before-- large language models, specifically the transformer architecture of those large language models. That's the T in ChatGPT. And so if you put those two technologies together, you have an explosion of productivity.

So maybe an example to kind of help everyone understand why these two technologies are so powerful. So if you gave a traditional CPU, a Central Processor Unit, from the likes of Intel ( INTC ), the task of reading A Tale of Two Cities by Charles Dickens, and you said, Mr. CPU, please read the book and tell me how many times Charles Dickens says the word "the" in the book, it would start on page one, and it would read.

Greg Bonnell - Word three, it was "the."

Dominic Rizzo - It was "the--"

Greg Bonnell - There's your first "the."

Dominic Rizzo - Exactly. "It was the best of times, it was the worst of times." I always forget the second line, but "it was the," "it was the." Right? The GPU, on the other hand, would rip the book into 100 different pages and pull out all the thes simultaneously. So right there, you can see that incredible productivity enhancer. Right?

If you put that with a large language model, particularly the transformer architecture, that will give the GPU context. So it will say, okay, if Charles Dickens says the word "the" on the first page, it's most likely preceded by the words "it was." So we basically then took those two technologies and we ripped up the entire Internet. And that's how we captured the global zeitgeist with something like a ChatGPT this year.

Greg Bonnell - Right. So what are the building blocks here for success in AI? We know the graphic processing units definitely gave us the leap forward. The public has to get their hands on it with these language models. What about companies who are trying to be successful in AI? What do they need?

Dominic Rizzo - Yeah. There's really four things you need to be successful in AI. So the first, we kind of alluded to it before, the compute resources. AI is incredibly silicon-intensive. Just to give you a stat on that, AMD has said that they expect the AI chip market to grow from $30 billion in 2023 to $150 billion by 2027. That's basically a 50% CAGR.

The second, you need talent. There's actually not that many people in the world who know how to do this well. Third -- and these next two are probably the most important, in terms of who's going to win in the marketplace -- you need data, and you need distribution. And so why do you need those two things? If you're going to rip up that book, you need a lot of different books to go study, right? So that's the data. And then the distribution. You need to be able to get it out to your customers very, very quickly.

As we know, the world is moving faster than ever. And so if you have a captured customer base, you're able to take those insights from that compute resources, from your talent, from the data, push it into your distribution, and monetize that product relatively quickly.

Greg Bonnell - So you mentioned NVIDIA off the top. Obviously, they have the GPUs that were seen as the leaders in the space. If anyone looks at a one-year chart of their stock price action, clearly they've been a beneficiary. What about other areas for investors to consider? I mean, longer term, who's going to be a player in this space?

Dominic Rizzo - Yeah. So there's different periods, if you go back throughout history and you look at different computing resource buildouts, what happens? The first is you see that infrastructure-level buildout, right? And so that's the likes of the GPUs from NVIDIA.

We think AMD has the potential to be a really nice second source provider in that marketplace. They actually just said on their conference call just the other week that they expect their new MI300 GPU to reach over $2 billion in revenue next year. That's the fastest product to over $1 billion of revenue in company history.

So this is a really big market, and there's room for other players. But then you have to look at the rest of the digital semiconductor ecosystem. Where are those GPUs made? They're made at TSMC ( TSM ). How are they made? They're made on machines from ASML, these EUV machines. Each one costs 250 million euros. They're the most amazing things that you'll ever see. I mean, the size of three double-decker buses wide. It's really incredible.

How are those chips designed? They're going to be designed on software from companies like Synopsys ( SNPS ) and Cadence ( CDNS ). So that whole digital semiconductor ecosystem is really going to benefit as we see that AI chip market grow from $30 billion to $150 billion. And just to put that number in context, the whole semiconductor market in the whole world is only $500 billion today.

The other place you're going to see win, and it's kind of a trope to say this, but those Magnificent Seven companies. Those Magnificent Seven companies. Think about who has the compute resources, who has the talent, who has--

Greg Bonnell - And who has our data, right?

Dominic Rizzo - Who has the data, and who has the distribution. It's those Magnificent Seven companies. So when I think about navigating this environment responsibly, which is what I'm trying to do on the strategy, is trying to find these areas of winners in order to see who's going to win. And there's four things we look for to identify those winners and the strategy. The first is that they sell linchpin technologies. So these are technologies that are mission-critical to the success of their customers or make their users' lives dramatically better.

The second, they need to be innovating in a secular growth market. The third, this is probably the most important tactically, they should have improving fundamentals. It usually comes in the form of revenue acceleration, operating margins that are expanding, or free cash flow conversion that's improving. And finally, it's very important in tech, and sometimes people forget, you need reasonable valuations. So if you have those four things, you can identify the winners in those different pockets of the ecosystem.

Greg Bonnell - I was going to ask you what things you're watching for in terms of roadblocks for the development of AI. Is it basically a company that doesn't have those four things?

Dominic Rizzo - Yeah, so there's a couple of different roadblocks that could happen. So regulation, of course, being one of them. I think AI needs smart regulation. When you have an incredible technology like this, it's actually responsible to make sure that there's smart regulation in place. One thing that you have to make sure is that regulation doesn't stifle innovation, though, too, and that it just doesn't ingrain the incumbents. So like I said before, the incumbents have a natural advantage here, right? That's actually pretty rare in technology innovations.

Sometimes you have things called sustaining innovations, and sometimes you have disruptive innovations. AI is very much a sustaining innovation. So look at the difference between, say, the PC to mobile transition. That was a disruptive innovation. You had different winners for the mobile ecosystem than you did for the PC ecosystem. Who won in PCs? Windows, Microsoft ( MSFT ) and Intel. Who won in mobile? Apple ( AAPL ), TSMC, Arm ( ARM ). Right? Disruptive innovation, new business model, new way of selling.

Sustaining innovation is when those players are the same winners this time as they were last time. And like I said, those big companies, they're the best positioned for this. So you have to make sure that the regulation doesn't actually ingrain them even further than their natural advantages. So you have to watch regulation. But I think this is an incredible technology, and it's really going to be an incredible productivity enhancer for the global economy.

That's what's so exciting about it. It's not that these large language models are just used for chat. It's that they're used for image generation, or cybersecurity, or code writing. There are so many different areas where you can apply this. I mean, even this job. One day, you may be talking with an AI me.

Greg Bonnell - So this is where you get a threat. At least you're the AI in this situation.

Dominic Rizzo - Yeah, not you.

Greg Bonnell - So you're going to be the AI. I was like, what? What am I going to do.

Dominic Rizzo - Well, look, I actually think-- so I have a great associate, and we were working on a project together. And he kind of came up to me after, and he said, you know, Dom, I'm a little worried. What does this mean for my job? And I said, Austin, you have an incredible opportunity here.

If we think about all the time that you waste updating the models internally, or writing the notes internally, that's all in the same format. We're going to unleash your time. And now you can go work on incredibly new value-add tasks than, say, just inputting numbers into a spreadsheet, or reading that extra email that, you know, wasn't actually that useful.

Original Post

For further details see: