Links shared in Slido during the live broadcast:
This Week in Tech links.
Full, unedited transcript:
[00:00:00]
Tim Beyers: Fools. Welcome to This Week in Tech. He's Tim White. I'm Tim Beyers. Jim Gillies is still in the background because that's what he always does. Tim, why don't you get your intro out of the way, and we'll get into it here. We've got a little bit more year in review.
[00:00:20]
Tim White: I'm Tim White. I'm tech leadership here at Motley Fool. Been doing enterprise tech for 30 years, and excited for a snowy weekend, maybe.
[00:00:28]
Tim Beyers: Maybe. We'll see. It's been cold. It was cold yesterday. I'm Tim Beyers lead advisor for Rule Breakers, also on the Trends team. Go ahead and get your questions in. Let us know what you're thinking about as far as tech goes. But we have another part of the year. Last week we reviewed the year in AI. Now we are reviewing the year of, can we call it everything else?
[00:00:59]
Tim White: I think so. I think it's everything but AI.
[00:01:01]
Tim Beyers: It's the everything but AI year in review. We need to start with the Year of Efficiency, Tim. I find this stat absolutely staggering; 524 tech companies laid off. 152,074 employees in 2024. Your reaction to that is.
[00:01:21]
Tim White: Some companies did it better than others. I'll say that was probably my initial reaction. But I think that a lot of companies were pursuing that year of efficiency and really trying to get their costs down to a different place post-pandemic, where a lot of companies experienced a pretty big bubble, especially if they were involved in remote work at all.
[00:01:41]
Tim Beyers: We also need to get the shared links in there. I'm not sure if we've got that yet. But we want to make sure the shared links are in there. I'll give my thoughts on this here. I think the Year of Efficiency is going to become the Years of Efficiency. I don't think we're anywhere close to finished here. The reason I say that is because there is a certain drive to squeeze unit economic costs as much as possible. Because we had so much blunt force investment at the end of 2024 in everything from data centers to GPU hardware. I think we are looking at ways to squeeze out costs wherever we can to fund development initiatives in other places. I think most of that is going to be AI, but I also think some of it is going to be like custom software development. I have no doubt about that, either. I think a lot more efficiency is coming. The question for me, Tim, to maybe give a little breath around this is, will there, for example, be cuts in R&D? Classically, I would say no, but is it possible that you are going to see much sharper R&D investment? Maybe not cuts, but your R&D growth rate goes down substantially to what it's been historically.
[00:03:21]
Tim White: I think we talked about that happening last year, and I think that will continue to happen this year, where companies are very more picky about what R&D projects they pursue and only pursue the ones they think are big. As far as where that efficiency comes from, I think the short answer is it always has to be people. That's the most expensive part of running any tech business. That said, I do think that there's the potential you might see some people pursuing Cloud efficiency. We had some high-profile folks last year, getting out of the Cloud in order to normalize their costs, and try to get things under control. Of course, a lot of people buying huge amounts of AI stuff, as we talked about last week in their own data centers, which is not in the Cloud, so that's probably another place you'll see that.
[00:04:07]
Tim Beyers: Resource optimization is another way we could say that. Moving on. The CEO Merry Go Round in 2024 was ridiculous. The ones that you mentioned on here, Digital Ocean and AWS are obviously the biggest ones, but more than 1,800 CEOs resigned or removed in 2024. I think the other thing, too, Tim, is maybe a bit more focus. This is something I'm going to look for in 2025, particularly amongst the tech companies we're covering. What are the incentive changes? That's going to be interesting because there's been a lot of CEOs that moved out of their positions. Do yourself a favor Fools, if you are so inclined to do things like investigate proxy statements, which is where what CEOs get paid. If you're looking at those, look and see, do a search in the proxy statement and look for two things under incentives. Look for the word incentive and look for artificial intelligence or AI and see what shows up. I bet there's going to be more things that show up than I would have guessed last year, Tim.
[00:05:24]
Tim White: For sure. I agree. Because I thought that last year would be the big year for that, and I suspect that is yet to come. We had a prediction last week that more people will be using LLMs and generative AI this year, whether they know it or not, and I think that's still the case.
[00:05:40]
Tim Beyers: I think that's still true here, as well. Look for more turnover here. I would expect, I'm not sure, the one that we didn't get on here because it was towards the end of the year. But the one that still makes me sad to this day is Pat Gelsinger at Intel.
[00:06:01]
Tim White: He made too many mistakes and just couldn't really recover from them, because unfortunately, I think the world had changed so significantly since his mindset about how to run the company had been formed. But, brilliant engineer, brilliant tenure, and his original run at Intel. I think it's just more sad that they brought him back in a place where he wasn't set up to succeed.
[00:06:25]
Tim Beyers: I agree with that. If you're so inclined, go ahead and put in the chat, the CEO that you think is most likely to be on the Merry Go Round in 2025. You can make your own reckless prediction there. We'd be curious about that while we move on to antitrust. Antitrust is going to be super interesting. As of a year ago, antitrust was a little bit on the back foot. We were talking about maybe things settling out. But by the summer, boy, it had really ramped up with teams under the microscope. Then the big one, the really big story here, which I think we're still seeing the fallout from, Tim. Honestly, I'm not even sure that we've even begun to see what the full fallout is the Google Antitrust losses here. What are your thoughts on that?
[00:07:29]
Tim White: I think that they had a couple of different lawsuits that had a couple of different outcomes. I think that obviously, these companies that have had judgments go against them are hoping that a regime change will go in their favor. Whether that's true or not, we'll see. But the key here is that those lawsuits have not yet really had any penalties come to fruition. Until that happens, I don't think we're going to really see any impact from them that's material.
[00:08:04]
Tim Beyers: It's really hard to say because we talked about this last week and we had a conversation. A couple of our colleagues were in the office while we were doing our planning this week, and we had this conversation about how there was so much worry about Google, not just with the antitrust microscope, but the disruptions from OpenAI and so forth. We're search is dying. But at the same time one of our friends, Cheryl, literally asked you say that. What phone are you using? It's a pixel. What do you use to access your AI? I use Gemini. It is still embedded in essentially everything. I think the unwinding is maybe going to be less unwinding than was originally anticipated and will probably take longer than anticipated.
[00:09:07]
Tim White: That said, I expect that it will continue to be wielded as a weapon of influence with tech companies by the government and that that may have some second-order effects as well as perhaps we saw this week with Meta's announcements.
[00:09:24]
Tim Beyers: You know what? That is fair. For those who don't know, Meta announced with at least the incoming administration is taking some credit for this that Meta is removing what had been its professional moderation practices and instead going to a style that is closer to what twitter, X uses, which is you can put on notes onto stories that appear on Facebook. Whether or not you agree with that or not is immaterial at this point. It is an example of the government exerting some amount of influence. I think you're right. The threat of the antitrust cudgel could create some knock-on effects we can't yet see. The end of tech non-compete. I was thinking that this would have broader effects. We were talking about it in terms of specifically tech non-compete, if you've ever been a contractor and I have, non-compete agreements can be a real bear. You can really shackle you. Obviously, I was in the creative industry. I was a creative contractor working in writing and analysis and things like that. I was not a tech contractor. But I think in tech contracting, non-compete agreements are, I think they're even stricter here. This felt like a very big deal here. Tim, do you expect to see more portability in 2025?
[00:11:00]
Tim White: I do. All of the spin-off companies that we talked about last week that spun out of Google and became OpenAI and many other companies like Anthropic those all would not be really as possible without the end of non-competes. The whole idea is that oftentimes those who would lock you down for five or ten or whatever years. Hilariously, I was just watching a documentary last night that Leo Fender, the inventor of Stratocaster and other guitars, when he sold his company to CBS back in the '60s, he had a 10-year non-compete to do anything guitar-related. That obviously, had some pretty big effects, but it would be interesting to see if that wasn't legal back at the time, what would have happened?
[00:11:45]
Tim Beyers: That is very interesting. Let's keep moving. Everyone but Waymo and Tesla gives up on full self-driving and robotaxis. I'm really curious the thinking on this, Mike. I guess where I'm on this, Tim is, even though Waymo and Tesla have not given up on this, I still think we are way overestimating how soon this is coming because we always, always overestimate how quickly. Tell me I'm wrong here. What am I missing? Maybe I'll frame it this way. I get the reason why some in the industry would give up here because the two companies with what is arguably most important for robotaxis, full self-driving would be relevant data and usage of relevant data. Tesla and Alphabet are undisputably, they have the biggest data advantages. I get why others would be like, we could never catch up to that. But then when am I going to see my robotaxi?
[00:12:59]
Tim White: Apple, which also runs their own maps product and thus has quite a bit of data, as well, gave up on their effort back in March of last year. GM gave up on cruise. Uber gave up on self-driving taxis. Lyft at one point had one, which I think they've long ago given up on. I think it's just super interesting because everybody is like, well, we have to have this. You have to have this to compete. I think what we've found is that the cost of doing that and the regulatory scrutiny and all of the bad press you get from your Waymo taxi, accidentally bumping someone in the butt or whatever, compared to the hundreds of taxi drivers that literally maim people on the regular. [laughs] It's a tough business to get into, and I think that a lot of companies just said, we've got too many distractions right now. We got to focus.
[00:13:56]
Tim Beyers: This would be one area where I think data can have a non-tech but very valid impact in this industry is if your data is good enough like what you want to be able to do is prove, like if you're in the robotaxi business and you are going to be to the point you're just making there, Tim, what insurer is going to handle you unless you can prove that actually our robotaxis are safer. The data is one of the things that allows you to write a more affordable policy. You would hope because what you could say is, our incidence of accidents is this much lower. The incidence of serious industry is this much lower. The data is super important just for getting regulatory approval and actually writing policies that won't kill the company.
[00:14:51]
Tim White: Interestingly, you made me wonder who is currently insuring Waymo. The answer is there's a company called Swiss RE, which is a reinsurance company. [OVERLAPPING]
[00:14:57]
Tim Beyers: That's a giant company.
[00:14:57]
Tim White: That does it. They also have supplemental insurance for, if for whatever reason, the taxi can't actually take you where it needs to go, for example. I thought they might have gone for self-insurance, but apparently not.
[00:15:16]
Tim Beyers: It is possible because it's a reinsurance company that their first line of defense is self-insurance, and they may have a defined reserve. Then if that reserve gets breached, then they go to the reinsurer and say, hey, we need money here. Our reserves have been breached.
[00:15:40]
Tim Beyers: That'd be interesting.
[00:15:42]
Tim White: One of the big stories was that Waymo is only currently operating in San Francisco in Phoenix, places that notoriously don't have a lot of ice [laughs] and snow. I think it will be really interesting to see if those vehicles can ever get out of relatively good weather condition areas.
[00:16:02]
Tim Beyers: That's a good point because we are in the middle of snowy Colorado here, and we do get particularly in areas where we live, the main roads always plowed. The not-so-main roads, not really.
[00:16:21]
Tim White: My street never gets plowed unless there's, four feet of snow, so.
[00:16:25]
Tim Beyers: It is always a bit of an ice rank on the regular streets. I understand that we've come to rely on the Colorado sun melting snow and it's usually pretty reliable in that area. But anyway, let's keep moving here. Postgres, absolutely eating the world and SQLite, too. I think I'd like you to maybe expand on this a little bit, because what we've seen with Postgres if I've understood our conversations properly here. Postgres is a SQL database. It's highly popular, open source, and it is rapidly, rapidly becoming a platform, shockingly so in 2024, how quick that happened. Maybe you could talk about that.
[00:17:19]
Tim White: PostgreSQL is an open-source relational database, so classic relational database like Oracle, Informix, etc. SQLite is another open-source relational database. They work differently. Postgres runs a server that you can connect to with a network client, so you can have a central Postgres server and connect to it from multiple clients. SQLite runs as a file system only and runs in a process. You're firing up some code running on a machine. It can access the SQLite, and once that code is done, SQLite is no longer working. Both of them, I think, have taken over the world because of the rapid increase in needs for databases in two different realms. One is what they call local first development, which was a big trend last year as well, which is that you write your code, assuming that you're working with a database that's right on your device, often a phone or a tablet or some smart device. Then there's some other process that is in charge of synchronizing the data from your local database to some central place and there was tons of projects last year that were working on things like that, using SQLite and using Postgres. SQLite obviously very popular because it's very small and can run on a lot of devices. It's installed for free as part of any Python install, for example. The reason I think that these databases became so popular is in reaction to the big Cloud providers having their giant mega a very expensive databases, many of which are actually just using Postgres or SQLite under the hood or something that is cloned from another project like MongoDB. The costs of running those are super expensive once you get up to Amazon levels of using a database like that. A lot of open-source startups and so forth are using Postgres because it is extensible with plugins. You can write a plugin for Postgres to let it do things like store AI vector databases or use it to store geospatial data. Those plugins are relatively easy to create and install on Postgres compared to writing your own whole new database. Because of the open source/free nature of Postgres the extreme extensibility by writing plugins and a lot of plugins that are very popular and are available in Cloud hosts if you are using Cloud hosting. There's just been a lot of innovation in that space, and I think that is really critical to the next generation of apps that are going to be even more data heavy than the current generation.
[00:19:45]
Tim Beyers: I want to make an argument here that I wonder whether or not you agree with because one of the things that we saw last year is that Postgres was taking some. It was a real threat. It was a real alternative to MongoDB, for example. That wasn't part of Tom Gardner's sell argument, but it's something that we watched fairly closely. Tom made the argument is that it was technically was falling behind a little bit. I'm not so sure. I think you can disagree on that, but I also think that you could have a real argument about that. I think there is a legitimate argument to be had about that. One of the things that's true, though, is that Postgres ascending as a platform, Tim, I think that validated David Achari's strategy at MongoDB to say, we're going to go win workloads. Because if we try to go head to head like this on a free platform and say, our platform is better than this free platform, I think you're going to lose that argument. If instead, you go workload by workload and prove unassailable value at that workload and build up, I don't think it's still not an efficient way to sell, but I think it's an effective way to sell, and the rise of Postgres seems to validate that MongoDB is under threat and the way that they're dealing with it makes some sense.
[00:21:11]
Tim White: Part of the Year of Efficiency story is suddenly everybody cared about open source again, and there was all kinds of other problems that we'll get to in a minute here about open source drama. But we went from, no, I want the best and I'll pay for the best. I don't care what it is. Whether it's MongoDB is the best or Oracle is the best or various things running on AWS or the best, I'll pay for the best to. Hey, is there a way we can do this for free? [laughs] I talked to many DBAs over the course of this last year, database administrators, and many of them were saying that clients were aggressively getting off of Oracle in onto Postgres and SQLite. Aggressively getting off of other big databases because there were plugins available on Postgres to do it for 80% as good and for zero cost. That is a huge differentiator when you are not able to get loans at 3% and so forth. That Year of Efficiency, I think, has driven a lot of people back to open-source, and that's part of the reason Postgres is eating the world. But even Amazon, if you look into shared links, there's a very interesting Amazon article about how much they are spending on Postgres. They have many engineers who are contributing back to the Postgres open-source project to make it more robust, more suitable for running on AWS, more capable of handling a variety of workloads. Even Amazon, who has their own database products, is acknowledging people want to run Postgres and we want to be the best place to do that.
[00:22:43]
Tim Beyers: I think there it is likely, in my opinion, just everything we've talked about with Postgres, Tim, that you're just going to see more and more use cases for it, you're going to see more and more usage. As long as these plugins are easy to build, I expect one of the dangers for everybody else is that if Postgres is unleashing developer creativity, watch out. Just watch out. It will get more legs and become an increasing threat to everyone. This is an interesting one to watch.
[00:23:19]
Tim White: There are many companies that are building their own product on top of Postgres that I'm watching closely as for future IPOs. Neon DB is one of them. I think there will be others. Cloudflare now running has the ability to run SQL light inside of a Cloudflare worker, which is super interesting as well. I think there's a lot of bits and pieces of the database world that are likely to change because of everyone suddenly saying, like, wait a minute, we could be doing this for free. Why are we paying the big bucks to Oracle, etc? Which is why, as we talked about last week, I think Oracle is pivoting to data centers.
[00:23:55]
Tim Beyers: Yes, I think that's right. Well, let's stay on the open-source theme here because we want to talk about ARM versus RISC-V. Tim's going to explain this better than I will, but I'll just kick it off here and saying. RISC-V is an open standard. RISC it's a chip architecture, reduced instruction set computing or compute. So the RISC-V architecture is an open-source architecture. Somebody could take it, you could take that RISC-V architecture and build around it. ARM is a RISC architecture that is not open source. It is proprietary. When RISC-V started asserting some a bit of market presence, there were real questions about, what does this mean for ARM? It is continuing to be an ongoing discussion that we'll get to a little bit later. But let's talk about all the things that happened as a result of this emergence, Tim.
[00:24:59]
Tim White: To be clear, the RISC-V standard has been around for a while. It really came to prominence because of export controls on the ARM designs. Essentially, there were various governments that put things in place so that the ARM reduced instruction set designs, that instruction set architecture could not be exported to China, for example. As a result, suddenly, there was a lot of interest like, well, if we can't use ARM-based designs, can we use RISC-V designs? We have an actual news story from Justice last week in 2025, where a Chinese chip developer has finally gotten to the point where they believe they have a working RISC-V CPU that could be brought to market, and potentially the first laptops that use it could come out in the first quarter here, they are much less efficient and productive than the ARM-based ones, and we'll see why that is, at some point. But I think it will be interesting to see if anyone outside of China who must use RISC-V is going to do it that actually becomes a real threat to ARM.
[00:26:09]
Tim Beyers: I guess this is another one where incentives would matter here. We just talked about Postgres here. If you can build, is it enough to say, look, if I build my chips on RISC-V, I avoid the ARM licensing fees. I don't know how big the ARM licensing fees are. I don't think they're that big. Well, I guess I don't know here, Tim.
[00:26:38]
Tim White: Absolute numbers, I don't know. All that made me question that statement was that Qualcomm and ARM have been fighting all last year. That's another story I could have included in this section is Qualcomm versus ARM. There was lawsuit after lawsuit between the two of them to figure this out. Qualcomm notoriously made the ARM-based CPUs for many Android phones for a long, long time.
[00:27:04]
Tim Beyers: That's their Snapdragon.
[00:27:05]
Tim White: The Snapdragon platform. Google then, much like Apple, decided to make their own ARM-based architecture for their phones, the Pixel phones and some others, much like Apple did with their own. A class chips that they have on their phones. So there was just a lot of fear, I think, in Qualcomm's part, that they're going to lose a lot of business here as people start making direct contracts with ARM and so there was a lot of contention around that licensing. I think that will continue to be the case as between laws and export controls hitting this stuff, companies realizing that they can design their own chips with these architectures and get a lot of integrated benefits. We have a statement that someone mentioned in the comments, what we already have a story, which is Jensen Wong talking about how much faster he's going than other people because of vertical integration. I think all that is going to be part of the story of chip design in the next year.
[00:28:05]
Tim Beyers: That does seem right to me. What is probably a dumb question about this is one of the allures of ARM. Not ARM necessarily versus RISC-V, but ARM overall is that by virtue of an ARM package, you are getting not just the chip design, but you're also getting support, tooling. You have the basis for the software stack. There's some known tools you can use to develop software to the ARM stack, and that is all maintained by ARM. You essentially get the white glove service. You get the pre-packaging. As a developer, you don't have to go out and fumble around and make mistakes in your writing of software to the underlying platform. Or does RISC-V being an open-source, a widely understood open-source project have its own well understood packaging, so it's less of an issue.
[00:29:09]
Tim White: A couple of other bits and pieces here on the ARM story. Microsoft released their Prism translation layer to match the Rosetta 2, one that Apple came up with so that they released their laptops this year that were based on ARM architecture and actually running Windows and pretty decently and with shockingly good battery life, which is one of the hallmarks of a reduced instruction set computer. That has really pushed forward the idea that, okay, well, if you can run Windows on, then that really makes it a lot more viable for people, I think, than if you can only run Linux on it or only run Mac OS on it. We also had the Cerebras IPO, which we talked about last year. Cerebras is a company that's making AI-specific chips that are in this zone of reduced instruction set, but also designed very specifically for AI workloads and Google tensor processing units. They came out with their new Trillium 6th gen tensor processing units. Again, chips specifically meant to run AI, but all in this idea of chip architecture suddenly getting a shot in the arm for the first time since probably the late '90s.
[00:30:24]
Tim Beyers: It's lots of innovation happening there. Let's keep moving here. In terms of online privacy, Google saying they were killing third-party cookies.
[00:30:42]
Tim Beyers: It's interesting in that every time I go to sites now, I do get the hey, will you accept cookies? It's not like they're dead. They're not dead. They are we're supposed to die and then they got zombified. They did not die.
[00:31:00]
Tim White: I did a lot of scrambling in December 2023 and January of 2024 getting ready to deal with a world where third party cookies were debt because Google said it's happening in January. I was like, I'm ready. They'd been threatening that it was going to happen for years, and they had set a deadline. Then somehow it didn't happen. They eventually announced that enough advertisers had gotten back to them and said, no, thanks. We can't do this. We can't handle, a world without third party cookies. That they ended up saying, well, I guess we maybe aren't going to do that. We will try and do something else instead, which is they're going to have some more complicated nuanced way of doing this, which I'm sure everyone will love in terms of this. Of course, the cookie banners do nothing in our pointless political theater, which has made the Internet a worse place, but the idea that Google was going to kill these cookies was a huge story last year, and the fact that they didn't was also a huge story and if you work in web development, Google repeatedly made moves last year that were catastrophic to a lot of businesses. This was just one of them and it's really interesting to see, in a year where they were really pounded with antitrust stuff, it turns out they're pretty big monopoly, and if they do things, it matters a lot to a lot of people.
[00:32:21]
Tim Beyers: The likelihood of more scrutiny here is, I would say, is high. Let's keep moving on open source drama. There were hacks introductions of I'll let you explain this a little bit more, but I want to quickly hit on the, Matt Mullenweg and WordPress and WP Engine. This is what happens in Tech, where you have prior allies going at each other and it gets personal and it ratchets up very quickly and in a very ugly fashion, but maybe your take on the various open source dramas here, Tim.
[00:33:04]
Tim White: Sure. A few years ago we had the open source drama, but we're Elasticsearch. Elastic is a company we've talked about a reasonably amount of time here and they had a product called Elasticsearch, which essentially Amazon just copied and took and ran and charged money for, and Elastic got not a single dollar for doing that. They changed their license. Well, last year, they changed it back to say, no, we are open source again because there was a bunch of people who stopped using them because they weren't open source. Other companies have created all these source available licenses that included things like Terraform from HashiCorp were source available instead of open source in order to try to prevent big companies from taking their hard work and essentially making a whole bunch of money off of it without them benefiting in any way, shape, or form. That was a big story. Many companies, many projects, back and forth on that. The other thing was open source hacking. A lot of these open source projects are run by one or two people that are just trying to make the world a better place with their software and one of them called XZ Utils, which is a Unix security tool, which is, again, way down in the guts of people using Linux for servers. Essential, but invisible, essentially. That developer was having some personal issues and just couldn't keep up with all of the requests for patches and so some Chinese developers essentially put in a bunch of patches saying, hey, you need to fix this. This is a huge security hole. Like, you got to fix this. They accepted this PR, pulled it in, and it was, 10s of thousands of lines of very low level code, very hard to fully understand. This got accepted into the project and went out there, and some random person noticed that their code was running slightly slower than it used to and dug in and realized that there was all this, like, massive malware that had got introduced in that. It was patched relatively quickly, thank goodness, but every big open source project that I talked to, they're dealing with this constantly, whether it's someone has, put in a bug into some package they depend on. Someone is submitting PRs to them that contain malware and they have to, like, fight them off. Someone is claiming that they have found a security hole in their software and is saying, like, pay me $1 million, and I'll tell you what it is. It's a wild time out there for a lot of these open source projects that are increasingly, as we talked about with post Christmas Equalite, critical to the operation of a lot of businesses. Google, as I mentioned, made a lot of changes last year. One of the things they did was their helpful content update, which essentially, when you searched on Google, instead of showing you websites, we just show you a bunch of Reddit posts about something. That basically killed the ability for any website to make any money because their websites didn't show up in search results. As a result, a lot of people that were paying for WordPress, via WP Engine or other ways, were no longer paying for WordPress because their sites went away. I personally believe that pushed Matt Mullenweg the founder of WordPress over the edge. He was already on the edge. I think it pushed him over the edge into holy cow, this is a big problem and so he started looking around and realized that WP Engine, which is a company that WordPress had invested in, WordPress is a content management system that runs about, 60% of the websites on the Internet that you go to. They were getting no money from them, and WP Engine was making all of the money, and that the deal that they had cut wasn't giving enough of the money back to WordPress and so he started an incredibly weird and public feud with WP Engine and shut off, access of WP Engine the lawsuits back and forth. He basically let a bunch of his employees leave because they disagreed with this whole thing. Personally, I think that was partly because he didn't see that the company was going to be as big as it had been because Google had essentially removed the need for websites to exist. Then AI, of course, also another big blow to the idea of a website existing and needing to be there when you can just ask AI. I think that that started to push a lot of buttons with the CEO of WordPress and I don't know where that's going to end up. It's an ugly, situation, but a lot of drama with these open source projects last year, I suspect that will continue into this year as we see them become both more critical. Business is more interested in using them and less interested in paying.
[00:37:40]
Tim Beyers: That's the tension. More interested in using them, less interested in paying and in the case of Mullenweg, he called WP Engine this essential, partner that they had invested in. Called it a cancer on automatic and WordPress. That went poorly. Can you just quickly, just for those who don't know what it is, I know what you're talking about when you say a PR, but when you're pushing something in to just define a PR super quickly.
[00:38:10]
Tim White: Sure. The idea is that you have a piece of software usually existing on GitHub, Microsoft's code management platform, and you'd like to make some changes to it, so you create a pull request that says, hey, I have these changes, and I would like you to pull them into the project, please. That is how a lot of businesses manage their code, even internally. I might submit a pull request to one of our projects saying, like, hey, I have these changes that I need. Here's what they are, and they can be reviewed and you can look at them. One of the big challenges is once those pull requests get to be over 40 or 50 files and especially of low level things like C, it can be very hard to review them and fully understand what is going on there and if you are a single person operation maintaining an open source project, it can be extremely challenging to do that, especially when everyone is screaming that there's security holes are real.
[00:39:01]
Tim Beyers: The reason I wanted you to define that is because that should give context Fools into why the automated generation of AI code and injected into systems is potentially such an issue because an AI could maybe do, I'm going to guess here, Tim, but, like, you could have 100 pull requests a week that are AI generated into a system. If you have no way to incredibly review all of that code to make sure it is accurate, does not introduce exploits and so forth, that's a serious problem. This is where we are. That's why I wanted you to quickly define that. Very quickly.
[00:39:46]
Tim White: Well, just to be clear, there are AI tools that are now being used to review PRs to try to do this. It's the classic AI versus AI thing or whatever, for a single person.
[00:39:54]
Tim Beyers: Watching the watchman. That is exactly what's happening here.
[00:39:57]
Tim White: But for a single person, open source project, even AI tools aren't going to defend them against hundreds of malicious PRs.
[00:40:06]
Tim Beyers: Very true. There was a bunch of M&A this year, and we'll go through this quickly because I want to park on cyberwar, a little bit before we get to our predictions. There's so much in here, and you can always ask us about us in future shows, but we've got 19 minutes left. I want to make sure we hit some of our questions, but the one in here, actually I'm going to park on two. AMD buying ZT systems. I think that was super important because it helps AMD get into the business of actually doing some amount of customization and integration, for their large scale chips, particularly the ones they want to get into data centers. It's the answer to NVIDIA trying to build more at a systems level than just selling cards with GPUs on them. NVIDIA doesn't want to do that anymore. They want to sell a whole stack whole systems. This is one of those things that helps AMD get to that place. I thought that was important and will have increasing importance in 2025. The one that didn't happen, where Google was going to buy HubSpot and then didn't and I'm glad that that didn't happen, even though I could see the logic for it at the time. I don't expect that that is coming back and then this is my big disappointment of 2024, which is IBM buying HashiCorp, because I thought that HashiCorp was just getting cheap enough tremendous assets. Terraform is a tremendous tool still very widely used. I think IBM got a deal here taking out HashiCorp when it did. Anyone you want to comment on in there, Tim?
[00:41:52]
Tim White: No. Other than, I still am laughing about DirecTV buys Dish for $1. Still. Anyway, I can move on to Cyberwar.
[00:42:03]
Tim Beyers: Let's move on to the Cyberwar here because this is the recurring theme during 2024, if there was a recurring theme, the number of hacks, the number of breaches and, of course, this is highlighted by the CrowdStrike outage, which wasn't an attack, but just reveals to some degree, the frailty of systems because this all got started by an error being introduced. An automated push that got introduced into the system that just took everything down because the code was just the instructions that loaded into the system was just errant instructions, and it just destroyed you eight million plus Windows computers around the globe. It's just a terrifying example of the frailty of systems, but your thoughts on Cyberwar in 2024, Tim?
[00:42:59]
Tim White: I think there's a lot of interesting bits to talk about, but I think the biggest story for me was that in the end, despite this being a CrowdStrike caused outage, Microsoft is who took it on the chin.
[00:43:10]
Tim Beyers: The really did.
[00:43:11]
Tim White: Hard and they ended up being very contrite and really actually saying, like, maybe we should actually change the way that our, security architecture works for Windows and a lot of things like that. They also just got hit with a bunch of their own hacks and breeches and everything else and so I think Microsoft was the Cyberwar target of the year. I think it was just shocking to see how many people were like, wait, this shouldn't have been possible to happen CrowdStrike or no.
[00:43:42]
Tim Beyers: Absolutely agree. To their credit, though, Microsoft is thinking about and they did. They had the big summit with CrowdStrike. Say Adella was on stage at Falcon, which is CrowdStrike's big customer conference and they introduced some new ways with CrowdStrike, but also independent of CrowdStrike, new ways to make it easier to manage your security updates and, I think, give a bit more guardrails to teams in order to manage Microsoft systems. If that's a good thing that comes out of this, I think that is, good to see. Hopefully, it doesn't repeat itself in 2025. I'm going to move past because we already talked about Pat Gelsinger moving on. The Intel implosion was heartbreaking. You should look out look at the Ag drones that was amazing, but let's move on to quantum because there was a question about it in what Jensen Wong said about quantum. I'm going to paraphrase. I'm not sure if I have it right, Tim, but what Jensen Wong said is, like, we're probably way farther away from quantum than we might think, and that really placed a Paul, I'm going to double check myself to make sure I'm right, but I thought that's roughly what he said, because quantum, at least as it's being pitched, I don't know if this is the reality is something that would be threatening to something like, a GPU based supercomputers and so forth. I think he's talking his book a little bit there, but I think what we saw at Denver Startup Week still blows my hair back. I think we're way closer to at least quantum adjacent tech, which I'll let you talk about here. We're way closer to that than, any of us actually realizes.
[00:45:52]
Tim White: What he said is that truly useful quantum computers are 20 years out, and I don't necessarily disagree that, you're not going to have a quantum iPhone in the next five years. It's not happening. But I do think that quantum adjacent technologies, which is super tiny atomic clocks, which is photonic ICs where you're actually pumping light around inside of ICs, all these other technologies that are both quantum computer enabling and side cast offs of the main quantum computing industry, are really going to be interesting. Google is still full blast developing quantum chips. They announced their big Willow chip toward the end of the year. I agree with all that. While I agree with Jensen Wong that no one's going to have a quantum computer on their desktop for at least 20 years, that's probably right. I do think that you will see maybe not full on quantum computers be interesting in the next five years, but all of the technology that it takes to make them will have knock on effects for a lot of other companies in the next five years. That was a really interesting thing to learn this year.
[00:47:00]
Tim Beyers: I think that, too. I think that the way the technology markets develop, the purpose built stuff that is on the periphery, but solves very high value use cases, you build from the outside and come in Fools. What Tim is talking about with those atomic clocks and these quantum adjacent technologies, would be very purpose built for very specific things like improving GPS, for example that's one. But you could easily or at least I could easily see if it's high value enough in a supercomputer environment where you have a big budget and a lot of work to do to solve something very specific would you see a quantum platform be applied to that? I think the answer to that is yes, but I also fully agree that quantum PCs on the desktop, no, that ain't happening anytime soon. That does not mean there won't be genuine purpose built, interesting platform technologies that use quantum in a very high value way. That seems to be a lot closer than I think we are considering at the moment. We saw some of this at Denver Startup Week and in what Google's working on. Quickly on these last three, I actually want to start at the bottom, because these Chinese Risk 5 processors, how likely is it do you think that China has got this right? The Chinese engineers have got this right, and you are going to see a run of common use risk five based systems that are running amok in the Chinese market.
[00:48:53]
Tim White: I think it's likely that they'll have to be there because the demand is there. They want these kinds of chips, and so someone will fill that demand. Will they be competitive with arm? I don't think so anytime soon.
[00:49:05]
Tim Beyers: I think I agree with that. What Jensen Wong said, also, along the corollary of what we just talked about with quantum, he said his company is growing faster than Moore's law. I actually think he's probably right about this.
[00:49:21]
Tim White: Moore's Law was specifically about the number of transistors doubling every x amount of time. I don't think that's happening with what Jensen is talking about, but what he's talking about is the net performance gain because of their vertical integration where they can tune the software, tune the firmware, tune the way the compilers are written All that stuff. The net performance gain is faster than the performance gain of just even doubling the number of transistors. I think that's probably true for now with what they're doing.
[00:49:51]
Tim Beyers: Then, last one, and we're going to get to predictions and questions. Mark Benioff says Salesforce is hiring and there was a question about this. No software engineers in 2025 due to AI. There was a question about the knock on effects here that I saw. What will the knock on effects of that be? This is from our Pal Dwight Kurt Schrute. What are the second degree implications for investors? Well, I think what it does is, I'll let Tim talk about the engineering perspective here, from a investor perspective here, Dwight, I think this raises the bar. For Salesforce. Whether they mean to or not, they're introducing expectations that if I were Mark Benioff, I would not want. You are telling me that you could run at a scale and efficiency that I should expect. What would I expect, Dwight? I would expect margin efficiencies. Why wouldn't I? You can scale without software engineering. Interesting. What is the margin expansion going to look like? That's my first question, Tim. But from your perspective, what's the engineering implications here?
[00:51:04]
Tim White: We talked about how many engineers were laid off last year at the top of the show. I think that Salesforce probably would have reduced or not hired engineers anyway, and this is a convenient way of saying, and it's because we have cool tech. That's my opinion.
[00:51:20]
Tim Beyers: This remains to be seen here. But they genuinely believe that they can turn a lot of those automated workflows that have been in Salesforce forever. They can make agent force a massive product that they don't need a lot of additional software engineering to do it. They can re-purpose a lot of their existing tooling to get this done, scale up their sales. Theoretically, 2025 should be a year of at least incremental margin expansion for Salesforce, if this story and the narrative, the way Salesforce is telling it is correct. That is something to watch. Let's move on to predictions here because we don't have a lot of time left here. Do you want to go first or you want to go second?
[00:52:04]
Tim White: I'll go quick. I think that 2025 is going to be the year of the AI Stack Wars, where people are vying for developer experience to get AI workloads to run on their AI accelerated hardware. We had lots of questions about AMD. I think this is where AMD is going to have to go if they're going to continue to be competitive here, is they are going to have to make it so compelling for developers to run their AI workloads on AMD hardware, and they're going to have to do that with a whole package of developer experience, like a stack of things you use to do this. I think you're going to see Sara Brass do this. You're going to see Nvidia continue to do this here. You're going to see Amazon do it. You're going to see Microsoft do it. They're all going to want you to run your AI workloads on their hardware, and there's going to be a huge war to get people to do that. I think a lot of it's going to come down to making it as easy as possible to do so.
[00:52:54]
Tim Beyers: I think that's right. The stack wars always shake out where developer experience, like, if you hook the developers, it does matter what's in the stack, but it also doesn't matter. If the developers get what they want at the point of entry, if the developers experience is solid, you will have creative developers figuring out how to maximize that tool. I completely agree with that. Mine will be I think DeepSeek has changed the game entirely. For those who don't know, DeepSeek is a highly efficient Chinese large language model. It is built in software. It is largely built through resource constraints. I think that is absolutely fascinating. It has completely altered the economic argument for OpenAI. They can no longer spend like a drunken sailor as they have and that you can't, what was it you said, Tim? It's something like six requests, that you use it and open AI is now, losing money on everything.
[00:54:12]
Tim White: I pay $20 a month for ChatGPT. As soon as I make at least six requests in a month, they have lost money.
[00:54:17]
Tim Beyers: They've lost money.
[00:54:18]
Tim White: I make more like 600.
[00:54:22]
Tim Beyers: It's crazy. They are bleeding money. My prediction is they have to give up on this idea that we must raise a trillion dollars in order to scale this up at the current level of like, with the economic equations that they are using today, where after six requests, they start burning cash, so they need trillion. They just can't do that anymore. But if they don't realize that, they're going to be WeWork, and they are going to flame out. If they do realize that and they start competing and trying to not trying to outdo DeepSeek, but saying, I see what you've done here and I'm going to figure out my own efficiency equations here, then they can be Uber. They're either going to be Uber or WeWork but not in between by 2027. One or the other.
[00:55:21]
Tim White: Either flame on or flame out; one of the two. This is their year to either keep in the game or get taken out. I think it will be very interesting to see which one it is. Flipping over to questions, thank you all for submitting so many great questions. Lovely great stuff in here. We're not going to be able to get to them all. To know, we have a lot of questions about AMD. Those are the top rated ones. I kind of gave my take that they're going to need to make it so easy for you to run your AI workload on AMD that it would be a non starter and to continue to compete on price, which is how they have been successful up to this point, Tim, what do you think? It's going to be, the future holds for AMD?
[00:56:00]
Tim Beyers: Fully agree with that? I think this is the year of the AMD ecosystem. This is what I'm looking. Fools, I don't know how low AMD will go, but what I can tell you is, I think the market will recognize if the AMD ecosystem expands materially such that it starts to look viable as a comp, better than the Nvidia ecosystem or the Nvidia stack, but just an interesting and viable alternative. The two words that matter are viable alternative. If the ecosystem gets built out in 2025, look for AMD to rally would be my argument.
[00:56:41]
Tim White: There's always going to be some companies that are going to want the best of the best and buy Nvidia, and there's companies who are like, We can't afford that, and we just need some AI, so give us that AMD, goodness.
[00:56:51]
Tim Beyers: Alex is asking, What do you think of TransMedics being shorted and it's big drop today? Is it still a buy? It is for me, Alex. I can't tell you if it should be a buy for you, but I'm not too surprised that there's a lot of skepticism around TransMedics. For those who don't know it, TransMedics doesn't just have the Oregon care system, which essentially is a large cabinet whereby donated organs are kept in a metabolically healthy state inside there. It's not cold storage. You increase the viability of the number of donated organs. You take that party because one of the problems with cold storage is it's you have to get the timing exactly right, and a lot of organs end up being unusable. The transportation was always such a problem. What TransMedics did like two years ago, they also opted to build out a transportation network. They built what we've called on this show a whole product, not just the cabinet, but also the airplanes and the distribution, and to do the whole product. That costs a lot of money. It's going to be very lumpy. Because they are trying to completely reform entire industries. Could it be a good short term short? Of course it could. It's going to be highly volatile, but I really believe in it in the long term because this is what technology is for. This is absolutely a migraine level problem and trying to solve it is super hard. So there will be bumps.
[00:58:33]
Tim White: Dave asked, did the market overreact to Jensen Wong saying the comments about quantum computing? obviously, they overreacted. Someone else asked, What's your favorite quantum stock? I don't have a public company in quantum that I love. Right now.
[00:58:44]
Tim White: I'll say alphabet. I will alpha.
[00:58:47]
Tim White: Alphabet's probably your best one, which, of course, has many other reasons to like it other than quantum. I think that they will continue to innovate there. That's probably the answer, but a lot of the smaller companies that are just getting started, especially a quantum adjacent is what I'm interested in, and those aren't public yet, but I'm definitely keeping an eye on them.
[00:59:09]
Tim Beyers: I think those are yes, absolutely. When those come public, that's going to be absolutely fascinating because it'll have, like, that vertical software feel, and those who know me know that I love vertical software companies. I think that's super interesting. We'll end on Dave J. What are your thoughts on where Uber is heading in the next five years, opportunities versus distractions? I'm glad the Robo Taxi distraction is gone. I am also hopeful, Dave, I think it is a massive efficiency story. It is about scale. More cities. Appropriately adding more drivers and what you want to do, like the real tech advantage there, and what Uber's been really good at this is building out their logistic software. You reduce dead time or what in the airlines they call deadheading. Deadheading is in order to get to if you are, like, a flight crew member, and you have just completed your trips and, like, you're in Atlanta. But you are assigned to be on a flight to Hawaii. You have to get back from Atlanta to Denver to take your Denver to Hawaii flight. The Atlanta to Denver flight is deadheading. That's dead time. That employee is not making money for the airline? This is what Uber wants. I want drivers to be always on a job. If they are riding an idle, that's time that Uber doesn't make money and the driver doesn't make money. There's more efficiency they can squeeze out there, Dave. This is why you're seeing the margins go up. They found creative ways to eliminate more of that dead time. You're going to see a lot more.
[01:00:57]
Tim White: Finish with this nugget, which is that I want Uber to become the Uber of Uber things. When Uber first came out, every other start up is, Well, we're the Uber of X. Well, I think it's time for Uber to be the Uber of X There's so many things that just having a really good contractor workforce that you're managing well can succeed in, and they proved it with food delivery. I think that they just need to keep looking at those startup ideas from five years ago and make them their own.
[01:01:25]
Tim Beyers: Agreed. Two minutes after we covered a lot. We will be back to normal programming next week with current week shows. But that is our back to back. What a packed year it was, Tim. Absolutely. Absolutely. We'll get back to normal programming next week. Thanks for tuning in Fools. We appreciate you being here. Thanks to Natasha for running Motley Fool Live again today and stay tuned for Fool 24 coming a little bit later. Tim, stay warm ish cause nobody's going to stay warm. It's going to be cold.
[01:02:00]
Tim White: Great.
[01:02:02]
Tim Beyers: Fool on, everyone.