A New Trend in AI is Emerging: Efficiency

0:05Tyler Crowe:Efficiency in AI could be just what the doctor ordered. This is Motley Fool Money.

0:21Tyler Crowe:Welcome to Motley Fool Money. I'm Tyler Crowe and today I'm joined by longtime Fool contributors Matt Frankel and Jon Quast. We got a really interesting show today. We're going to talk about some stepwise gains in the world of AI, but not in the way we've been talking about it for quite some time. And we're going to hit a mailbag question because we are already getting them even after announcing that we're going to start taking them at our email address. We'll get into that in a second. But first, probably the biggest news day story outside of, you know, global war conflicts and things like that was there was some pretty big legislative decisions that happened for big tech.

0:56Tyler Crowe:and it wasn't exactly the way they wanted it to go. In a couple of Pibble cases, one of them, Meta, Platforms, and Alphabet were found liable for a woman's mental health struggles related to social media addiction, and they had to pay punitive damages. Also, Meta lost a similar case in New Mexico related to misleading statements about safety on their sites. Now, admittedly, this one's kind of a hard one to discuss on an investing podcast. I think there's a lot of emotions wrapped up in our use of social media and how that affects things. But there is going to be some investing takeaways here. But I wanted to get your guys' thoughts here.

1:30Tyler Crowe:Are either of you shareholders in Meta? And as you kind of like viewed these verdicts and things like that, has it really changed your view of the company?

1:39Matt Frankel:So to your first question, Tyler, no, I'm not a Meta shareholder personally. That's not a indictment on the company. I've gone on record that Meta is one of my more favorite companies in the Magnificent Seven. But to your second question, does this change my view of the company? The answer is not yet, but it could potentially. Specifically with the issue of mental health, I think that all the statistics point to an indisputable conclusion. We do have a mental health epidemic, unlike anything we've ever seen. And I believe two things can be true here. First, social media apps are engineered to maximize user engagement.

2:15Matt Frankel:I don't think that that's a controversial take. It is how they generate revenue. Of course, they want their users on there as much as possible so they can generate ad revenue. The second thing, I mean, I wouldn't blame social media apps entirely for mental health issues, but it does seem to be a contributing factor when it comes to this topic. I think if you're new to this discussion, the book I'd recommend is The Anxious Generation by Jonathan Haidt. It does a great job of laying out the rise of anxiety in Western culture, also potential remedies. some suggestions by the author. I know this might sound off topic for an investing podcast, but I do believe it is pertinent.

2:53Matt Frankel:The jury verdict here is far from the final conclusion when it comes to Meta. Meta is gonna appeal this case, it's gonna go higher. But if social media is ultimately deemed harmful, like other harmful things, we could see some substantial reforms from legislators that could have an impact on these cash cows. So I'm not saying that it's the final word, but it is something material to keep an eye on.

3:14Jon Quast:I'm not a Meta shareholder. But we do own shares of Alphabet, you know, which is YouTube's parent company in my household. My wife does in her account. It doesn't really change my opinion on either company. I've already had my kind of opinion formed. I'm a parent of young children. I think all three of us here are parents of young children. So I'm sure we have our own opinions of social media when it comes to, you know, how it impacts the mental health of our kids. I feel like people are becoming a lot more cognizant of the dangers of social media right now than they were a few years ago, which is what the basis of this case was.

3:45Jon Quast:It was people who use social media and were addicted to it a few years ago. But my kids, I can tell you, won't be on social media until they're at least in their late teens. And one of the reasons is that it's so difficult to say what's safe and what's not. So this doesn't change my opinion. I already kind of had this opinion about these companies.

4:02Tyler Crowe:This is a hard topic to discuss. And I want to try to bring like the investment thesis around to this, because one of the comparisons that I've seen with social media outlets and these cases that we saw was this, they're calling it, like, for example, Bloomberg called it the tobacco moment, basically where legislative action starts to chip away at the business because we start to realize the physical, mental harm that these can cause. And it leads to a lot of legislative stuff. But here's the weird part about it is even after decades of litigation and fines and anti-smoking campaigns and declining tobacco use in the U.S., tobacco stocks have still been excellent investments over the years.

4:44Tyler Crowe:And I want to contrast that a little bit to other companies where punitive legislation have kind of crippled the company. And I think of examples like BP with the Deepwater Horizon spill. Companies that have been making perfluoroalkylate substances like PFAS, or we also call them forever chemicals, you know, the liabilities on those, you have like Dow, DuPont 3M, they've tried enormous like corporate changes to try to mitigate the liabilities here. And of those, none of like over very long periods, none of these companies have really like fared nearly as well when, for example, compared to tobacco as like actual investments.

5:19Tyler Crowe:Litigation against, you know, company meta is obviously bad and payouts are not like ideal. But if social media is treated similar to tobacco, like in this tobacco moment, Couldn't social media still have similar returns as investments in tobacco stocks? I mean, like you said, these are still cash cows.

5:38Jon Quast:There's a lot to unpack there. I'm not sure if social media is about to have its tobacco moment. But I will say that the most interesting thing about this, it's not necessarily the fines that were just imposed. I mean,$3 million split between Meta and Alphabet is nothing for these companies, but it's really the precedents they could set. So the verdict related to safety on social media, that could certainly have some legislative impacts. I mean, for example, there are countries that already ban social media for people under 16 years old. So something like that could happen here potentially. That wouldn't exactly be crippling to these companies.

6:13Jon Quast:It would have a material impact, but it wouldn't be crippling. The mental health lawsuit, which is the one that was split between Meta and YouTube, is a little more interesting to me. So Facebook has over 3 billion users worldwide. So would this pave the way for everyone whose life has been harmed by a social media addiction to potentially try to get millions of dollars from these companies? Virtually anyone, I know I could, can make the argument that their lives would have been better or more productive, at least, if it weren't for social media. You know, how much more could I have written over the years if I wasn't, you know, on Facebook?

6:47Jon Quast:Keep in mind that this was an initial decision, as John correctly pointed out. Meta's going to appeal, as they should. The final ruling could be a little more consequential. I personally don't think we'll see a wave of successful litigation against these social media giants for mental health lawsuits, but we'll have to wait and see here.

7:04Matt Frankel:Yeah, I mean, it's an interesting hypothesis to compare tobacco to social media. I would agree with Matt here. I think it's far too early to draw that parallel. This case is going to go up the chain and it remains unseen how things would be regulated if they become regulated at all. So stand by.

7:23Tyler Crowe:Yeah, this is certainly something that's been kind of on the back burner for a long time. We've been thinking about it. But now with these litigation actions, it really starts to bring this into the forefront and something we'll be following in the coming weeks, months, years. We'll see. After the break, we're going to talk about AI and one of the more interesting gains that we've seen in the past couple of weeks.

7:43Matt Frankel:Have you been sleeping on your mattress a little too long like I have? My back's getting sore more than it used to. and I feel like Homer Simpson with my body shape and printing on the mattress. But like you, I'm busy with a job and kids, and who wants to go to the mattress store with the family only to deal with a pushy salesperson? That's why I was excited to learn about Lisa and their premium mattresses that you can shop for from the comfort of your own home. We wanted stability and comfort, so we went with the Legend Hybrid. It has over 1 ,000 individually wrapped springs for extra support, but here's the great thing.

8:12Matt Frankel:It virtually eliminates motion transfer just in case one of us has a restless night. No matter how you sleep or your budget, Lisa has a mattress for you. They're made from premium materials right here in the U.S. with a focus on using sustainable materials like recycled steel. Check out which mattress is right for you at Lisa.com for 20 % off, plus get an extra$50 off with the promo code FOOL, exclusive for our listeners. That's L-E-E-S-A.com, promo code FOOL for 20 % off, plus another$50 off. Support our show and let them know that we sent you after checkout. Lisa.com promo code fool.

8:48Tyler Crowe:This isn't a single news story that we want to talk about here with about AI, but it's kind of a kernel from a couple of recent press releases that I want to think about as an essential topic when we think about AI. And it's making it happen more efficiently. I'm going to let you guys kind of cover the topics a little bit more with the news stories. But one of them came out from Google talking about some of their ways of running AI models. And the other one came out from chip maker ARM, where they were announcing new AI-specific chip design. So, John, why don't you cover ARM, Matt, and you can cover the Google announcement after that.

9:22Matt Frankel:Okay, so, yeah, ARM. It's always been known for these lower power CPU designs compared to Intel's x86 architecture. But in the past, it's licensed its technology to companies such as Apple. The news that we are getting now is that ARM is actually going to make its own CPUs using a Fabless model, so just like NVIDIA or AMD, and Meta is going to be the first customer here for ARM of its custom silicon. I think it's a big deal because it is a pivot for ARM's business model. I would temper the news with a comment that Meta is a customer, but it's not an exclusive agreement. Meta is going to continue to buy products from multiple companies.

10:02Matt Frankel:It's probably more about diversification of the supply chain than anything, but it is something to watch.

10:07Jon Quast:Yeah, so on the Google side of things, they said that their recent research showed that a new memory compression method that they call TurboQuant could potentially reduce the memory requirements of large language models, you know, the AI models, by a factor of six. And understandably, shareholders of memory-focused chip makers like Micron and SanDisk were panicking about this. You know, memory efficiency, it's a big focus of AI development recently. And the demands are so high that these memory companies literally can't make chips fast enough. And if Google's right, then that might not be the case anymore.

10:41Tyler Crowe:Yeah. And this is what I found kind of interesting when I saw both of these stories come at the same time, because we have been talking about AI deployments and things like that. And obviously the knee-jerk reaction in the market was for memory companies. Oh, this is bad. and if arm were to take market share with lower you know power demand chips i could see some of the downstream ai infrastructure companies that have to do all this build out they would say oh yeah they're going to take hits as well and i want to get to how you guys are thinking about this as well but this was kind of like as i started reading the tea leaves between these like two announcements this was kind of my hot take is i think the efficiency gains that we're talking about here with using less memory, using less power, these aren't just like good for the AI infrastructure build out.

11:28Tyler Crowe:They're absolutely necessary to come anywhere close to meeting the goals and targets that we've been talking about.

11:34Matt Frankel:Yeah, Tyler, I feel like you've been playing Doc Brown on this podcast and I've been Marty McFly. Any single time we get an announcement for these data centers, there's massive power requirements and you've been yelling, great Scott, and I've been yelling, what the heck is a gigawatt? You've actually helped me see that they're basically saying that we need all this power and we can't generate that much power. We can't generate at the scale that the hyperscalers are talking about. And there's data to back this up. 30 to 50 % of the data centers in 2026 will be delayed because of power shortfalls, according to a report from Sightline Climate.

12:07Matt Frankel:Morgan Stanley projects a 44 gigawatt shortfall through 2028. There are already some data centers in California that have been built but not turned on because the grid needs some fortification. And this leads to an undeniable conclusion that the current path is unsustainable. We need something to change. We either need a power generation breakthrough, a compute breakthrough, or an AI model breakthrough. And I think that we're a long ways away from having a true compute or a power generation breakthrough. The AI model breakthrough is the easiest path forward. And that's what this thing with Google is potentially talking about.

12:42Matt Frankel:We've maybe come up with a more efficient model. That said, we do need to tap the brakes just a little bit when it comes to the memory requirements. So some memory stocks are selling off because of this. The Google announcement only optimizes a small part of the memory needs, specifically the key value cash. It doesn't reduce the memory needs across the board. It reduces a fraction of a fraction, not a fraction of a whole. The memory imbalance will likely continue in my opinion.

13:07Jon Quast:Yeah, and I would add to that, that take this with a grain of salt and zoom out. Memory stocks like Micron are still up over 300 % over the past year, even after the recent pullback. And if LLMs evolve like most analysts believe they will, one-sixth of the current memory usage will still be a lot of memory chips that they need, and it will still keep these companies very busy for the foreseeable future. Every new technology gets more efficient over time, consuming less power, the hardware becomes smaller, et cetera. Think of like, you know, flat screen TVs. This is a natural evolution of AI technology and it's a good thing.

13:44Tyler Crowe:Yeah, one is certainly going to help. And again, I'm going to emphasize this as my takeaway as we watch things like supply chains get strained and we talk about these massive numbers. The only way to make this possible, like John was saying, is we need some breakthroughs in terms of efficiency, in terms of power generation. Something is going to happen along the way because the numbers that we're talking about are just so hard to wrap our heads around in terms of the physical supply chain. It's going to be difficult to happen. So I'm very encouraged to see stuff like this because it is going to make it more viable in the shorter term and hopefully clear some of those backlogs that you were just talking about.

14:20Tyler Crowe:And coming after the break, we're going to hit the mailbag. These days, I'm all about quality over quantity, especially in my closet. If it's not well-made and versatile, it's just not worth it. That's honestly why I love Quince. The fabrics feel elevated, the cuts are thoughtful, and the pricing actually makes sense. Quince makes high-quality wardrobe staples using premium fabrics like 100 % European linen, silk, and organic cotton poplin. They work directly with safe, ethical factories and cut out the middleman, so you aren't paying for brand markups or fancy stores, just quality clothing. Everything they make is built to hold up season after season and is consistently rated 4.5 to 5 stars by thousands of real people like me who wear their clothes every day.

15:01Tyler Crowe:The Quince Mongolian Cashmere Crew Neck Sweater may be the most comfortable one that I own. It's light, soft, and was a lot more affordable than you'd think quality cashmere would be. Stop waiting to build the wardrobe you actually want. Right now, go to quince.com slash motley for free shipping and 365 day returns. That's a full year to wear it and love it and you will. Now available in Canada too. Don't keep settling for clothes that don't last. Go to quince.com slash motley for free shipping and 365 day returns. quince.com slash motley. Before we get to our first mailbag question, we want to make you part of the conversation as well.

15:40Tyler Crowe:If you have a stock or investing in question for John, Matt, me, or anyone else on the show, you can now email us at podcasts at fool.com. We'd love to have mailbag segments whenever possible. So send in your questions. Just remember to keep them foolish. That email again is podcasts at fool.com, podcasts at fool.com. And we're going to finish out today. And this is a mailbag question that comes from Jay Fung. I apologize if I mispronounced your name. But the question is, is setting up automatic investments always the best move versus wanting to buy the dip, which is very much in the parlance these days when we're talking about investing for individuals.

16:17Tyler Crowe:His email goes on, I know the answer is probably a mix of both, have automatic investments going and then save a portion to buy the dip. I haven't been able to get myself to pull the trigger yet on automatic investments since the markets have been so high. I'd love your insights. And am I unwise to hold off setting up automatic investments? Now, before we answer this question, we do have to stipulate because this is a podcast, we can't give personalized advice and none of this should be taken as personalized advice. So Matt, John, as you give your answers, just remember, we're going to think of a hypothetical situation and what you would do personally in this situation.

16:52Jon Quast:Yeah. So thanks for that, Tyler. And thanks for the question. I do a combination of the two, just like the email said, but I'll build it out a little bit. So I like to build stock positions and ETFs, if that's what you're asking about, automatically over time, as it really takes the emotion out of the equation and it mathematically forces you to buy more shares when stocks are cheaper. So think of it this way. Let's say that I want to invest$5 ,000 in a certain stock. I might commit to investing$1 ,000 on the first day of the next five months instead of all at once. So the way that this works is if I buy$1 ,000 worth on April 1st, and then the stock falls 20 % before May, I'm getting my next round of shares at a nice discount.

17:30Jon Quast:I'm buying the dip. If it goes up, well, my initial shares will be sitting on a nice gain, and I'll be happy in that situation too. So at the same time, I like to maintain a little bit of cash on the sidelines to be opportunistic. So in my example, if a stock that I'm gradually buying falls by 20 % with no change in the fundamentals or my investing thesis, I've been known to take some of my extra cash and make my next purchase a little bit larger to take advantage. So to directly answer the question, I will never try to talk somebody out of setting up automatic investments, regardless of whether I think the market is cheap, expensive, whatever.

18:04Jon Quast:Simply accumulating cash to, quote, buy on a dip, it more often results in missing out on a big gain. So I clearly remember that many people thought the market was ridiculously expensive in 2015 after roughly tripling from the financial crisis lows just six years before and decided to pull some cash out, stay on the sidelines and, quote, wait for the next dip. But then the S &P gained another 60 % before it hit any significant correction at all. So keep things like that in mind.

18:30Matt Frankel:Yeah, I want to build on this, Matt. You talked about taking emotions out of it. You know, people, we have feelings. We're emotional beings. We're not going to change that entirely. I think that's okay. But personally, when it comes to my financial decisions, I don't want to be making that based on my feelings because my feelings aren't reliable. They do change. I'd rather be making my financial decisions based on the facts because the facts don't change. And so I want to bring a factual study into this conversation. So Fidelity analyzed some returns, a hypothetical 5 ,000 annual investment from 1980 to 2023.

19:10Matt Frankel:So basically$200 ,000 invested over 42, 43 years. If you invested all of that money on January 1st of each of those years, you wound up with 5.1 million. If you invested 417 a month, which is 5 ,000 a year, but broken out monthly at the start of each month, you had slightly less, 4.8 million. I think that this supports a belief that the earlier you get the money in the market working for you, the better.

19:38Jon Quast:Yeah. In that study, I'll spare you the mathematics, but it essentially works out to your investing for an additional half of a year overall in the 5.1 million case. So that is still a form of averaging into positions like I'm talking about. So it's just wider intervals. You're not trying to buy the dip. I've read that study and the biggest difference was getting your money in sooner. Like I said, for the first year, you'd have$5 ,000 in the market right away versus, you know, an average of$2 ,500 in the market at any given point that year. So that makes a big difference over time. And second, I will say it's not always practical for listeners to invest in large lump sums once a year.