In this podcast, Motley Fool analysts Dylan Lewis and Tim Beyers discuss:
- Margins, pricing power, and other metrics to watch from big tech companies.
- More details about Twitter's acquisition.
- Why Chinese tech stocks are dropping as Xi Jingping begins his third term.
The battle of research papers is on! Motley Fool engineering manager Tim White joins Tim Beyers to discuss the simple reason why tech giants are very interested in generating images from text prompts.
To catch full episodes of all The Motley Fool's free podcasts, check out our podcast center. To get started investing, check out our quick-start guide to investing in stocks. A full transcript follows the video.
10 stocks we like better than Twitter
When our award-winning analyst team has a stock tip, it can pay to listen. After all, the newsletter they have run for over a decade, Motley Fool Stock Advisor, has tripled the market.*
They just revealed what they believe are the ten best stocks for investors to buy right now... and Twitter wasn't one of them! That's right -- they think these 10 stocks are even better buys.
*Stock Advisor returns as of September 30, 2022
This video was recorded on October 24, 2022.
Dylan Lewis: We've got what to expect when the biggest companies in the economy report earnings this week. Motley Fool Money, starts now. I'm Dylan Lewis joined by Motley Fool analysts, Tim Beyers. Tim, how's the going?
Tim Beyers: It's going well. Fully caffeinated, ready to go.
Dylan Lewis: Likewise, and I'm excited to jump into it. We have really the meat of earning season. We got the banks earlier, but I think we're getting into the companies that really move the economy at least as it relates to the stock market. Big tech reporting earnings this week, Apple, Microsoft, Meta, Alphabet, Amazon, all reporting in the next few days. These are some of the biggest names in the market and they are exposed to pretty much everything that's happening in the world all the time. Tim, what are you paying attention to when these companies report earnings?
Tim Beyers: Everything. But if we have to isolate it into two big metrics, there's a couple of things. I want to see where they land in terms of margins and across the board, like their operating margins, their net income margins, are they having heavier losses or not? Are gross margins expanding or are they declining? It's not going to give us perfect insight into what the inflationary effects are, but it'll give us something. We can believe that these companies have durable pricing power, and now we're in an uncertain economic environment. Let's find out. Do they have durable pricing power and if some are better than others, why are they better than others? I personally think this is going to be a really good report for Microsoft. That would be my bet. I think it's going to be a report for Meta.
I'm really curious to see where Alphabet lands because that could be either a great report, or it could be like I would have expected that. I'm really curious to see how these play out. That's one thing. The other indicator is how do they allocate capital here? Because all of them generate quite a lot of money, quite a lot of cash flow. Do we see dividend increases? Do we see buyback plans? Do we think that you have executives and Board members who say, "hey, you know what? Our stock has been absolutely battered we are increasing our buyback plan." All of them theoretically have the right and the cash to be able to do that, so that would be very interesting. We're going looking at those two things here, Dylan. If we see some incidences of degrading pricing power, I think that'll be worrying for some people. I'm not expecting that, but I think it's something to watch.
Dylan Lewis: The strong dollar narrative has been huge in the macro picture over the last couple of months. These are companies that do a good amount of business in the United States but they are also massive mega-tech companies that are exposed to the whims of a lot of international economies.
Tim Beyers: Sure.
Dylan Lewis: Are you looking at anything in particular with respect to that international revenue coming back to the US or the impact of the strong dollar?
Tim Beyers: Not necessarily. Not independently from everything else. But, it would be really interesting to see irrespective of the currency effects, how much growth they're experiencing overseas. That's always something. Because as these companies grow, they can't just grow off of US business they have to grow by compounding their opportunity internationally. The currency effects will be real. Some of the headwinds will be significant. When we saw Netflix report for example, there were some significant headwinds that Netflix faces and that is a global company. I wouldn't be surprised if you see some of these companies report like five percent currency effects, for example, on some of their revenue growth. But I also want to see like in Latin America, in Asia, in Europe, and Middle East, how much are they growing. Are they seeing a significant slowdown? I don't expect so, but that's more interesting to me than the quality or the size of the currency effects.
Dylan Lewis: Is this more a quarter to be focusing on the key business metrics rather than just anchoring to the dollars that we're seeing coming in?
Tim Beyers: I think so. I also think we want to know what it says about the business. Like in the case of Meta, what I want to know is, how are they dealing with the advertising headwinds we know they're dealing with. The worry with Meta is we just saw some pretty terrible results from Snap. How much does that translate to Meta and how much does that translate to Alphabet. Because they're all in different ways, advertising businesses but not all advertising businesses are equal. That's what we want to see. Is there a fundamental difference between let's say social advertising, Alphabet search advertising, and Meta brand advertising. What does that look like? Are they all experiencing the same degree of economic headwinds? I think logically the answer to that is probably not. But let's find out.
Dylan Lewis: Another company with advertising exposure, Twitter. We left them out of this discussion of mega tech. They are also set to report this week as well. They did not have an earnings call last quarter. I imagine that will be the case this time around. Tim, I think at this point, the folks who own Twitter are the folks who own Twitter, the folks that do not and are watching everything with the Musk saga are watching on the sidelines. Those camps are pretty entrenched at this point and I don't want to dive too much into the Musk saga other than to say, we have found some more details about what we're seeing with this deal, $44 billion acquisition, and how it may affect the business going forward. News reports came out this week that because of the $13 billion in debt that will likely be used to finance this deal if it goes through, Twitter could be looking at a billion dollars in annual interest payments on that debt. That is a huge spike from where they currently sit and I think something that would fundamentally change the way that this business looks.
Tim Beyers: It absolutely would. The natural question is, what does Twitter do in order to finance or account for that big of an increase in debt service. It's speculation at this point, would it go to a billion dollars in interest payments or are they going to do things differently? How this deal gets paid for and what happens to Twitter's balance sheet is a source of both interest and concern for investors. There's so much uncertainty here. I think this is the shrug emoji stock on the market right now, Dylan. I say that a little bit flippantly, but I also mean it. There's so many things that we don't know. In order for Twitter to be materially different because we're looking at a funhouse mirror right now. What we're looking at is a company that is distorted.
We're seeing a distorted vision of a business that will fundamentally change. In what ways it changes, we don't yet know under the regime to come in and change it. Are we going to operationalize for much higher growth? Are we going to operationalize for much better margins? Are we going to sharply reduce costs? Are we going to do some combination of the three? There's some material changes that need to be made in order to unlock value at Twitter. Those who own Twitter right now, let's be clear here, this is a stock that's on the Rule Breaker scorecard so I have a thesis of this as a lead advisor on Rule Breakers that there is value to be unlocked.
But I cannot tell you right now, Dylan, how that value is going to get unlocked. We haven't been given that picture yet. Those of us who are investors in Twitter and I personally don't own shares. But again, we've got it on the Rule Breaker scorecard, is we think there is a big difference between the assets that Twitter has and the value of those assets. There's probably a mismatch there. But how Twitter, as a company unlocks the value of those assets, that's an open question. We've got to get a real plan that we think is reasonable to say, "Yeah, OK. I can see how this new regime unlocks those assets."
Dylan Lewis: I think that question has played Twitter for a lot of the time that it's been a publicly traded company.
Tim Beyers: For a long time, yeah.
Dylan Lewis: I like the way that you talked about the financials piece of it and the strategy piece of it. To a large extent, one will dictate the other. If we see a massive debt load hit this company, and their current interest payments over the last 12 months were about $60 million. Going up to a billion dollars is a massive swing from that. This was, if you're looking at the GAAP numbers, already a company that was $100 million in the red over the last 12 months. That's not even looking at the cash in the outdoor. But so to say, it's not a company that has a ton of money and can easily service this debt with its current business strategy. Something is going to have to change.
Tim Beyers: No, something is going to have to give and so the news reports and admittedly, this is speculation here, Dylan. But there'll be vast cost cuts including mass layoffs. The thesis is that's going to happen. There's already been layoffs at Twitter so how much more can they cut and how lean can they run. That's one part of the potential thesis here. How lean does Twitter get as they try to unlock assets. But then the other thing you do, when you load up a balance sheet, when you dramatically increase say your debt service, or you dramatically increase the amount of equity, you sell and put on your balance sheet. You have to make up for that in some way and so you've got to grow.
You've got to grow, or you have to increase margins, or you have to lower your cost so your margins get better. Things are dramatically going to change. It does appear, Dylan, that you're going to see some combination of dramatic lay-offs, dramatic cost reductions, and then some new things designed to either juice growth or improve the top line margin, the gross margin. Whether that's via a subscription product or some more targeted ads that they can sell at higher rates. There's going to have to be both in order to do this differently. The business as it currently exists has been we got great assets. How we operationalize those is, we're going to get there someday, and now the clock is ticking to figure this out in a very sustainable way.
Dylan Lewis: Yeah, I think we're approaching someday, right?
Tim Beyers: Yes.
Dylan Lewis: I think Musk does not want to own.
Tim Beyers: Someday is now.
Dylan Lewis: That's right. Before we wrapped Tim, Chinese stocks, Alibaba, Jd.com, Tencent, Baidu are all down and down and some cases double-digits, mainly tied to Xi Jinping unprecedented third term and news of that coming out and some recent shuffling in the Chinese parties leadership. This actually came up on Saturday's show where Bill Mann and Ricky, were talking about international markets. We are now seeing the news affect the companies, and so I wanted to trace back to this a little bit because I think generally what we like to see is some path forward and some element of certainty when we're looking at companies and we're looking at the environment that they operate in. It seems to me like that is maybe cloudier than ever in the Chinese markets.
Tim Beyers: That's right. Bill is exactly, he's been right about this. I want to give Bill credit. Uncertainty is part of the reason you can get a return in the stock market. You're investing with a degree of uncertainty and so you get paid for taking that risk. Uncertainty is not a bad thing, it's a good thing. The problem is that when you have no clear way to understand how much uncertainty you're dealing with, then boy you better get paid a really big premium in order to take that risk. I don't know that we're getting that. I don't know that we're getting enough. Like are you getting compensated for taking risks investing in Chinese companies? For me, the answer is no.
For somebody else, it might be yes, but know what you're getting into here. I think this is uncharted territory, and because it's uncharted territory, the risk premium really has to be big. The way that works in investing and doing things like valuations, the discount rate or the required rate of return has to be really high. Like if this pays off, am I going to get 15, 17, 20 percent annualized returns if this pays. Well then OK, I might be willing to take that risk. But if not, then I don't know. I don't want that bad. To me, I can't properly assess how much of a premium I'm actually getting. I don't want to invest in a black box. I'm not up for it, Dylan, this is a black box. The treat inside may really be a trick that I really don't want to have anything to do with.
Dylan Lewis: You almost pre-empted my question there, Tim. I'm sure there are some listeners that are looking at, these companies were already very beaten up just on the growth prospects. The zero-COVID policy in China and lockdowns, the increased government intervention. We've been seeing an increased regulation. We've been seeing them fall even further. I am sure there are some people out there that are looking at some of these businesses and saying these valuations are starting to look more and more attractive. Is there anything that puts a business like this on your radar or is it just you're putting it in the too-hard bucket?
Tim Beyers: Well, I put it in the too-hard bucket, but things that would change it for me is some clarity on what regulatory changes are actually going to be put in place. Because for me, the black box is, we've known what we know about how Chinese regulators deal with companies. We used to have some knowns about that. Now with the third term for Xi Jinping and some promises of real crackdowns got more government control over businesses. That's now a big black box. It's an unknown.
If some of that fog gets removed, if it lifts and I know what we're dealing with in terms of a regulatory scheme for some of these Chinese companies, OK maybe, then I can look at one of them or more of them on the basis of an informed speculation because I know a little bit more about how they're going to be regulated. But I don't know that right now, and so until I do, I don't have any time. There's no chill in China right now, and so without that, sorry, I'm out.
Dylan Lewis: As we like to say, sometimes there are no called strikes in investing, right Tim?
Tim Beyers: Yes, exactly right. For me, I'm leaving the bad on the shoulder, Dylan.
Dylan Lewis: Well, Tim. Thank you so much for stepping in the batter's box with me today, always great chatting with you.
Tim Beyers: I appreciate it. Thanks, Dylan.
Dylan Lewis: One more Tim Beyers. Stick around, we've got Tim Beyers and Tim White, his co-host on this week in Tech from Motley Fool Live, our members-only live stream. They break down the artificial intelligence races between Meta and Alphabet and the simple reason they're very interested in creating images from text prompts.
Tim Beyers: There's a lot more happening in AI lately. I think some of what's happening is easily misunderstood. I think it's easy to underestimate just exactly what's happening with AI. Let's talk a little bit more about the types of AI and in particular generative AI, we've had that discussion, but let's redefine that.
Tim White: Generative AI is AI that's used to create something out of nothing or create something out of something else if you will. In the case of the AIs that are making the news lately, it's AIs that can create images from a text prompt. Both Facebook and Google, as well as a lot of other players have created these generative AIs that are creating images out of text prompts and they're using a thing called diffusion to do that. The way that it works is essentially, imagine take an image and unblur that image is where they started with.
We took a blurry image, we tried to make an unblurry version of it. We teach the AI to do that by giving it an image and then blurring it intentionally and then saying trying to get back and giving it a cookie when it does it correctly. You do that a bazillion times and eventually, it learns how to unblur images. Then you do the same thing where you say like, hey here's a prompt of a teddy bear on the moon. Eventually, you basically unblur an image that starts with noise, just like text noise, and unblurs it into that. It's very complicated, and there's all kinds of research papers that have been written on it. I think the reason it's hit the news so hard is it feels real.
A lot of AIs behind the scenes and invisible, and this very much feels like something that you can touch and you can play with and is truly amazing. Especially when it comes to images, there has been generative AI around tech for quite a while. We also have heard a lot lately about +Co-pilot, which is a generative AI that can help developers write code by analyzing all the code that has ever been written and checked into the GitHub code repository. That generative AI really allows AI to take everything it's ever known. This big data thing we've been talking about for a long time, and turn that from one type of thing and into another.
Tim Beyers: What's really interesting about it? I mean, the science-fictiony piece of this that I think is getting people excited is the next logical leap you make in your mind once you start hearing this is wow, OK, I can write a line of something, I can write a sentence and that sentence a computer will take and build something completely from scratch out of that. The point that you're making Tim is these AI models have hit so many data sets so persistently for such a long period of time in computer time that this it's hard, but they've had a lot of practice. It's not like we're just standing up something from nothing and we've hit a cliff, and now computers are brilliant enough to do this. It's been a mountain of training that's gotten us to this point.
Tim White: I think that's why many people underestimate AI and what it's capable of is the sheer amount of data that we have been able to use to train AIs. That data is like all of the text of the Internet that Google has.
Dylan Lewis: Right.
Tim White: All of the tweets that have ever been written, all of the code that has ever been checked into GitHub. All of the images that have ever been uploaded to the Internet. That includes the catalogs of every museum. All of this stuff is available out there to feed into an AI as the basis for it to create something from, which is the same way people create. We take everything we've ever been exposed to and use that to create our own new things. That said, there's a lot of ethical concerns about whether it's right to take the style of an artist and create new works without them, in their style, without paying them. I think there's some ethical concerns we still have to shake down.
But what really blows my mind about this is if you look back to where we have thought AI was going to be in the past was you ask it to do something, it does it. But what we know now is because AI is always watching, always listening, and the sheer amount of data that it can constantly be processing is way more than we ever thought, it is going to be able to push those predictions to you before you even asked them. In the old scenario of Captain Picard asking for his tea, Earl Grey, hot from the computer, the computer should know already that he wants the tea based on his.
Tim Beyers: The computer already does know, Tim.
Tim White: I think if we think about it, all of this data collected should be telling him you need some tea now to calm them down, or you should be doing this or you should be doing that, and I think we underestimate the amount of data that AI has access to and how much it can predict and create from that data.
Tim Beyers: Yeah. This is both the interesting and terrifying thing about this movement is what you've just laid out. I'll use the scary version of what you just laid out here, Tim, and here's the sense. It is the AI who is always watching. You could either be really scared by that or you might get super served by that, and honestly, it could be both. Let's talk about the so-called battle of the research papers we've been talking about here, Tim, which is Alphabet and Meta, each putting out their own research papers about what to do with the most popular form of this, which is text to image processing and what those two want to do, what they think is possible. Let's talk about why this is so interesting. I think we were talking about this yesterday. The most important reason they're interested in this is ads.
Tim White: If they can create an image that grabs your attention because it's been carefully constructed from everything it knows about you, it knows you like blue, it knows you like this car, it knows who lives in this area, it knows the weather by you is this. It can assemble those to create an image that hyper-appeals to you to click on.
Tim Beyers: This is really interesting. And so the more data that these AIs are studying and the more data about your life that you have input into the Google machine or into the Facebook feed, let's just take a very simple version of this. You search for new shoes on Google and the Google machine, to your point Tim, knows that you are a runner, knows that you like black, knows that you are a big fan of basketball, might send you a pair of black Nike running shoes. We don't know, but it may match you up and it may have a really spectacular image that is custom-made that Nike has commissioned at a higher ad rate for Google in order to present that ad specifically to you.
Tim White: Right.
Tim Beyers: This is the dream scenario.
Tim White: Deep learning AI can already do the reverse, which is it can look at an image and describe that image and put it in words. You may search for blue shoes, and it's not just looking for blue shoes on the text of the Internet. It can actually search for images that contain blue shoes that happen to match other images that are similar to images you've clicked on based on what's in those images.
Tim Beyers: The genius of this is that text-to-image processing creates a potentially much sharper ad delivered to you that hits you and hits you in the fields, we could say. Other applications here, one of the things that all of this brings up is a principal called natural language processing. I think of natural language processing as indexing here, Tim. But let's talk about what NLP actually is and how this plays into the AI models we're seeing.
Tim White: Natural language processing is a type of AI that lets you speak to or type to an AI as the way you would naturally speak, that's why it says natural language, as opposed to the way most people use Google where they just type a couple of words and hope for the best. This is you can actually type a whole sentence or speak a whole sentence to your robotic system of choice and it understands what you're saying. But vice versa, it also can generate natural language from its corpus of all the words it's ever been exposed to, and that language can be marketing copy. There's tons of websites out there that will let your AI generate marketing copy and tweets and the text that goes along with your Instagram post.
Now we can generate the image for you Instagram post and the text to go with it all from AI. Apple, I think really stepped into the arena this week. They have got a patent on the ability to take an image on your iPhone and change the pose of the person in that image. This is the kind of thing that again, these generative AIs can basically guess at what the other side of the picture might be. If your picture is one side of person's face, it can guess what the other side of that person's face might be in generate it from that. It does that based on analyzing all the image, but it can do that with language too. I had one where I'm a big DVD player and I had an AI generate a whole bunch of dungeon rooms for me.
Tim Beyers: Nice.
Tim White: It generate this entire adventure for me and it literally created the whole thing with full descriptions and full text and everything. It is amazing what these AIs can do and I think the future is going to be a lot different than we're expecting because of the amount of data these things have access to.
Tim Beyers: Yeah, I agree. I think natural language processing, just to explain what I was saying with an index here, when you have a common phrase, like an English phrase or attributes of phraseology, so you put in something or you speak something or you write something in natural language to a machine, they have associations of images and definitions and other types of attributes, maybe like a graph format. That makes these things very smart. Why that's so interesting from an AI perspective and what I love about it, Tim is that's the contextual processing that the human brain does naturally and computers do not. NLP brings a level of context that I think is fascinating, arguably a little bit terrifying, but clearly fascinating, that makes all of these additional features possible.
Dylan Lewis: That's all for Motley Fool Money today. As always, people on the program may have interest in the stocks they talk about and the Motley Fool may have formal recommendations for or against, so don't buy or sell anything based solely on what you hear. I'm Dylan Lewis. Thanks for listening. See you tomorrow.