TikTok Crisis, AI Coding Boom MCA8

Welcome to the Morpheus Cyber Podcast. Today, Tiktok Crisis, AI Coding Boom. From social media sell-offs to trillion-dollar coding disruptions, we're tracking the deals that could reshape digital power forever. Grab your phone and vote in our live polls. AHA Slides.com slash MCAA8. Tiktok's sales structure finally leaked, turns out avoiding a ban involves more corporate gymnastics than a Chinese acrobat troupe. AI just turned 30 million coders into a three trillion-dollar question mark. Are we programming ourselves out of programming jobs? Fusion startups are burning through 100 million-dollar rounds faster than they're actually burning hydrogen. But hey, someone's got to fund the sun. Waymo hit nearly half a million weekly rides. Then promptly reminded everyone why humans still clutch the wheel in their nightmares. Vote along with us at ahaslides.com slash MCAA8. Let's dive in. The convergence is here. Five technologies are colliding. AI, quantum, cyber security, crypto, and robotics. Each week we track the signals that matter most before they become headlines. Why do we have this thing called techfuturesindex.com? Well, it's like a stock index that tracks market movements. We track the top 10 in each one of those areas. And also we track private companies in techfuturesindex.com. So check that out. We'll talk about that once in a while. It's not our main focus. It's for anyone who's a technologist, a manager, an executive, or somebody who's just curious about what the future is bringing. And right now the future is changing mighty fast. So we are essentially sub billionaires. And we affectionately call ourselves that because we are not billionaires. But billionaires have teams tracking all of this for them. You don't. We've built this for the rest of us, founders, professionals, investors, and technologists. Enterprise great intelligence without enterprise price tags. I am building the future cyber security. And I'm focused on using Nvidia DPUs capturing every packet going in and out of major corporations, 100% of them sending that over to Nvidia's morpheus model running on blackwell chips. So that we are looking at every packet going in and out of an organization in real time so that we can discover threats, sub-second, and stop them. Because DPUs data processing units not only provide packet acquisition, they also have access control lists. So when you find something in the morpheus model and kill it. So that's why we name the podcast Morpheus cyber, but it's much more than that. So three perspectives, security architect, business strategist, technology operator, just real debate, honest disagreement. We tell you what we're actually thinking, not what's safe to say. So thanks for joining us today. And we have with us Jim Roundsville and Gastine, and they will introduce themselves a little bit here and there as the podcast goes. So thanks so much for joining us on morpheus cyber.com. You'll also notice that there are links that you can get in the show notes that take you to a real time interactive system where we have slide material and polls that you can interact with in real time as you listen to this podcast or as you watch it. So it's got a lot of interactive content. So thanks for joining us today. So hey Jim, what's up first? Every week we're going to take a few different topics and I'll start with one in just a moment, but we go through so many topics. Each one of us, right? We'll look at these different topics, what's in the news today, what we think is important. So we'll start with the first one here today. We've been kind of watching for our TikTok deal to show up and so we're very interested to see what the details are going to be. And we knew that Oracle would be a big player in this. And so when you peel back the headlines of the TikTok deal, you find that this isn't just kind of a clean national security solution. It's kind of a sophisticated illusion of control. Although there's control of Oracle to have all of the control of the storage of data, they don't have any control over the algorithms as to what is served up in the TikTok application. So this can have a lot of problems that go along with it. So they call this project Texas because Oracle's moved to Texas. For most of you who don't know that California has another big corporation leave. They continue to do that. And I heard with their billionaire tax coming that they're going to see a bunch of other corporations leave. That's a whole other subject. But what is this deal about? It's a $1.5 billion architectural work around. So that Oracle is going to store all the data here in the United States. But because they don't have control over the algorithm in any way or form, China is still able to serve up what they feel is important to the American consumer. And that to us is a security risk. And so we've got part of the deal done. There's still more to be done. I would hope I haven't seen that there's still in discussions about the algorithm part of the application. If they don't get hold of that, I can't see how this is a good deal for the US. I don't have TikTok installed on my phone or any of my computers at this point just because of the security risk. I'm a security guy. I'm just not going to do that. But all of my kids, all of my grandkids have TikTok. So I'm interested in this. I want to see the United States. Yeah, for sure. It's a great $50,000. It's crazy. 150 million spent. I mean, 150 minutes a day on TikTok. That's the influence at scale, my friend. That is totally influence at scale. So talking about this solution that they're looking at here, one congressman who's actually said a year ago, who was a senator at the time, this is security theater, not security policy. Who do you think said that? I'll give you a hint. He was a senator and now he's the secretary of state, Mark Arumio. So there is not a lot of agreement. And this guy is obviously in one of the cat bird seats now. So he holds a lot of sway. China's not going to give control that the algorithms, what TikTok's all about, you know, without the algorithm, you know, it's just another social media site. They track 150 different elements for each view to determine what they're going to show you next. That's where the magic is. And they've come out and said that's not negotiable. Interesting. Well, are we going to, are we don't really do very many predictions, but obviously, you know, we're wondering that the time frame is coming up where they're going to have to or get off the pot, right? So we're going to either have to have a deal or we're not going to have a deal or no deal. Something's going to happen here pretty quick or they're going to kick the can down the road. I kind of think they're going to kick the can down the road. Yeah. I don't know the technical feasibility of what they're suggesting. And that's to create this space, safe space where US entities are going to be able to look at the algorithm and try to control it without having the source code or really running things. And that's complicated to me. And is that going to slow things down? Well, let's put it this way. You can't have digital sovereignty when the brain remains in Beijing, right? Yeah. And if it not being fully Chinese, not being fully American, just creates this regulatory risk. How do you manage one and a half of something you don't own? You can't. It's not going to happen. So there's going to be this influence that we have no control over it. Yeah. Pretty much high dance would retain ownership of the algorithm source code under this proposed deal. Not sure that's going to work. Yeah, it's pretty dicey. I'm curious to see how this is going to unfold. And in TikTok generates $18 billion a year in global advertising revenue. That's no small jump change. Right. US wants to cut. Why don't you roll us out of this one because I don't see an end of that one. I mean, we're stuck waiting on Congress, right? Congress knows already that this isn't going to work. You know, the suggestion that was put before Congress is that, hey, you're going to have control of this part in Congress knows. No, not going to happen. That's what Marco Rubio said that. So we'll see more to it. Just in the news, we need to watch it, but it needs to get sorted out. Like I said, all my kids and drunker and kids love that out. And I'd like to use it to I just want to get it figured out. So anyhow, my family's on it. So okay, so moving right along here. I don't know if you guys heard, but there's a new market that's emerging with this AI coding. And there's something like 30 million developers worldwide. I'll make on average $100,000. And so that creates a $3 trillion marketplace for this new emerging technology. Now get hubs co-pilot. And by the way, co-pilot is the perfect name is genius marketing because it allows the developer to still feel like they're in control, but you got someone helping you. I give them kudos for that, but they're dominating the marketplace get hub. And they have almost 2 million subscribers. And this sort of proves that the developers are willing to invest in some sort of a co-pilot type of situation. JP Morgan Chase has given AI coding to all 50,000 developers. First of all, blows me away. They have 50,000 developers, but that's what it is. And better tools make better developers. But there is a downside to this. And the downside is that the AI generated code is approximately twice as vulnerable to security risks as the human written. What do I mean by security risks? You know, they have hard-coded credentials. They have missing authentication. Weak cryptology usage. There's what they're finding quite a bit. So the question really is, what do you guys think? Will AI replace programmers or programmers who use AI replace programmers who don't? Yeah. Well, my thought is, is it, you know, the proofs in the pudding and these companies are acting on ROI. And they're already seeing a 40% reduction in time to ship new features since deploying AI coding tools across engineering teams. So that is a very powerful ROI that companies can't deny. And where do companies go? They go to either make more money or they go to save more money and at the same time have efficiencies. So I believe that we're not going to stop this trend anytime soon. And I think organizations. I personally am a user of cloud code. And I believe that it's just massively powerful a year ago when it first started. It was error prone. It had a lot of problems. But today I can pretty much develop a new feature or what have you. And maybe I have one or two errors in the code. And then I just take those error codes, send them back into AI and it goes and fixes itself. So it's like physician heal thyself and it does. It's pretty dog on cool. It's not a panacea, but for somebody who doesn't know how to code, they can learn to code in a few weeks using English as the language instead of C plus or similar. True story. Yeah, I kind of watched you build an application from scratch and I've been impressed to watch that whole process and it's inspired me to also learn cursor and cloud in the future. But right now I'm happy just to work with cursor. But I read something by cursor CEO, Michael Trull. And he said that vibe coding builds shaky out foundations and eventually things start to crumble. So he's a creator of an app for this very purpose. And basically saying, it's like if you let AI build stuff without understanding the code, if you don't understand the code, then things fall apart as complexity grows. And that's true for simple apps, no problem. You can build a simple app. But as they get more complex, then problems start to appear and it's kind of like building a house with understanding electrical systems and plumbing. Right. So hey, what's so hard to do? Just tell it to do this. Thief and it'll create for you. But under the covers, there's some problems and we don't see those unless you're an advanced developer. And so the demand for advanced developers is going to rise significantly because that need to be able to see under the covers what really is going on is extremely important. In fact, they're starting to build in AI systems just to run security checks on applications, which to your point bill eventually this probably will all get sorted out and it'll get to not 40%, but 80 or 90% completion and dependability. And maybe the 20% are those one offs where this is a very complex problem that they have to deal with. And only in advanced coders might be able to say, how could we approach this in different ways that are best for this particular situation? So, they're good advanced coders. Yeah, good news for you and advanced coders. You guys are going to be around for a long time. Sure, but no, hope to losing you guys. You know, Jim, I agree 100%, but I've watched you and Bill tinker around with these things and I challenge that 40% number. I can't believe it's so low. Why are they low-balling that number? I guess because I see Bill in the middle of when we're talking, I go, gee, that'd be neat if we had a search bar at the top of this. And boom, shock, a lock, he goes in, types it in. I'm used to software development being a quarterly process. You know, they announce you get, you know, you decide you're going to make a change and it might show up within a year if you're lucky and it gets postponed. I'm watching them do it in real time. I think it's maybe 40 times faster, not 40% faster. Now the errors, you know, the security may be dropping, but the speed in which it helps an experienced coder code. I think it's dramatically faster. I think they're low-balling it. So let me talk about it in this way. ADHD is a problem that people have with executive function. They're very smart in the various pillars of knowledge that they have. The trouble with ADHD is that they don't have good executive function. What does that mean? That means if you're doing authentication within your application here and then some sort of a feature of your application here and then some reporting feature of your application here and then some technical information over here. The trouble is is that when these applications get large, the AIs, which the other thing I experience is it's like 51st dates. Every time you go to tell it a prompt, it's like, well, let me go see how that works. You just set that up five minutes ago. What do you mean? Let me go see how that works. You just did it five minutes ago. Well, it's like 51st dates. It has to go back to the code, look at everything. So it does not have context that goes from edge to edge. In other words, authentication is one silo and just like a person with ADHD, brilliant within, matter of fact, probably almost genius within their pillar, within a certain knowledge. But emotionally, maybe they don't have the ability to see the big picture from all across all of the pillars of knowledge. And so consequently, what ends up happening is these big apps as they grow and grow and you have so many different pillars and silos, the application does not have the context and it's not large enough to understand authentication and at the same time over here something else. And so consequently, when these apps get very large, they lose overall context and I call it executive function, which means that is what programmers, coders and technologists are going to bring to VidCoding and AI coding is, okay, I have to plan the overall architecture of this. Now, that's a topic and that's a task that AI can handle. Just like they handle all the complexity within a silo, once they understand that they have to understand the full breadth of the application and all of the aspects and they have to be able to understand different components of that. That's when AI is really going to take off because it's going to have executive function. It's not just going to be brilliant within one area. It's going to be brilliant in that one area or 10 areas of your app, but it's also going to have good executive function, context information and it's going to have a way of remembering things and indexing that remembrance a little bit better. I think that's when AI is really going to be trusted by the organizations, when we can put it into what I call the executive function. Right now under the hood, it's still statistical pattern matching. So it's seeing a group of code that's already been done by someone else and it learned that code. It's going to be a good pattern match and says, oh, I know what you're doing. You're building this. All of them may finish it for you. In fact, let me suggest you what you missed in your code. Let me, oh, would you like to add this to your code because it's seen it already. It's just statistical matching. It does not have the ability to do semantic understanding. So the semantics, meaning that the differences in the ways something might mean. And so you have to be so darn right specific to an AI to get it to really sometimes do what you want it to do. And you can almost make sure you want to say, no, dummy, you know what I mean. And you can't do it. It's because it doesn't know anything more than what you can architect. The human architect is safe for now. Yeah, for sure. I guess until that happens right now, they can't do that yet. And we've seen that. I've watched Bill. And stuff that I've done to you where I'm even, and we're not even talking about coding. We're just talking about the simple things, helping you with a configuration on a router or switch or any number of devices. You have to be very specific on what you say. It doesn't infer anything from what you said. Unless you've got right into their knowledge wheelhouse where it already has seen this before, then it can start suggesting things, but you've got to get it there. That's your job to get it to where it, it can actually help you. Yeah, and the other thing is documentation. The AIs, cheap AIs or AIs that don't have large scale context, you know, the smaller models, the Chinese models, etc. They don't have all the documentation. And the other thing is that even in cloud, in cloud code, the latest, I think, Opus 4.5 is like September of 2025. If somebody changed something materially in Python or C++ or any of the different coding languages, you have to explicitly go tell the AI coding engine, hey, there might be some new information out. And so then you have to provide it with the documentation. Now, once it's provided the documentation, it learns it really well, but that eats up a lot of your context window as well. So there's a finite ability. So there's these things called MCP servers, which extend the capabilities of the AI into other areas. Initially, they didn't have the ability to go out and cruise the web and find things new on the web. Now, if there's a new API or something that came out, let's say in December of 2025, and I need to know what that was because I'm coding in that particular area, I can go out, find that link of documentation and give it to the AI and it will consume it. But there's a limit to how much you can do with that. There are MCP tools called Contact7, there's some others where their expertise is helping coders gain access to real-time information to, you know, that's why it's called Contact7. It gives you real-time information to all the latest documentation for all of the programming languages and routines. Well, the good thing is, is the cost for using these tools is quite low. So I think an average person who wants to learn how to code, maybe you don't actually learn the coding structure, but you'll learn after a time of using it because you're using just common English and you're talking to telling it what you want to build, you start to see how coding actually worked. It actually builds the code before you on the, in whatever coding window that you're using, you can actually learn it if you want to sit and understand it. Every time it writes code, you can just correlate to what you asked it to do and learn it. So it's just kind of reverse engineering yourself into becoming a coder. Now, if future programs are always going to be using AI anyhow, don't spend so much time trying to understand that, the nuance of everything, as well as understand how to use the AI tool because that's the future. Six months ago, you wouldn't do this that and the other thing and now it will. So if you're stuck or you have an application that you wrote six months ago, then go out and look at it again with new eyes because AI is getting better, 15% per month. So that's about almost 100% better in six months, right? So if you're having problems, just put the app on the shelf until you know, some new AI code comes out and it'll help you out. Now, for you, let's talk about the cost. So you can buy like you did Jim cursor, 20 bucks a month. But as soon as you get to the point where you're doing a bunch of code and you're really in it and it's really looking good, they say, Hey, you're out of tokens. And then you you say, Oh man, I'm sweating bullets here. Click. I want to buy more. Well, cursor is pretty expensive in the add on area. So what I do is I use cloud code and I like it a lot. I use the max plan. It's $200 a month. But guess what? It's like having a full time coder, you know, in your shirt pocket, it is powerful. And you don't run out all the time because when you buy a little bit by little bit costs a lot more because those little bit by little bit per token costs more per capita. So it's better to once you decide you're going to do it. Now, let's just say around December, I wasn't using cloud code very much. I can drop it from the max plan 200 a month down to 100 a month or if I'm on sabbatical for a month or two, I can just turn it down to $20 a month or whatever and wait until I get back and then turn it back on. So they're pretty generous about turning it on and turning it off and the amount of tokens that you get and the amount of time. So on the $200 a month, the all I've hit the ceiling like maybe once. And the high roller bail. Yeah. But but the thing is, yeah, I'm still a sub billionaire. You know, billionaire's can you're a medium roller. That's it. But anyway, I hope that helps people understand a little bit of the context. You know, you can't really use open AI or cloud or any of them to do or cursor to do true coding for $20 a month. It's going to they're going to get you. And they usually hit you right when you're really solving a major problem. They say, you're going to have to wait because they do it in like five hour increments and they'd say, you can continue on using this level of model until 3 30 a.m. or whatever. And then it puts it into five hour windows. And if you if you overshoot in that five hour window like during the middle of the day, you hit the top, you're either you're either going to have to go, they drop you down to a lower model, which I don't like to program with lower models. I use Opus 4.5. I don't want to go down to a lower model because it might hose up my system, right? And then cursor has its own new composed model. And it's certainly cheaper and that sort of thing. But we don't know where it is. But we kind of think that it's using some of the cheap Chinese models in there to make it cheaper. But anyway, it's it's really great. I advise people to go you with the $20 a month or even less. You can build real code. The first thing I did was I created a Tetris program. Boom. Just like that. So the first thing when I was showing Jim how to use this is and I made him use the cursor, made him do it. And we need to do that with you, Gus, because you need to see this before your own eyes. You just click up into the agent area and say, make me a Tetris program that has this name and and allow us to see the the pieces come ahead of time before they arrive and then have a reset and then have a high game score and that sort of thing. And you can just keep adding to it. And I'm serious. I don't think I'm smart enough, Bill. I'm just your average sub billionaire living in San Diego. I like to hang out with my friends Jim and Bill. I Austinite friends. And I'm cool with that. Hey, if you guys are a student, any of you guys are out there are students. You can also Gemini CLI offers free access to their pro version. So you can get that for free, which is not their starting. You have to be a student. You have to be a student to get that discount. Because it's free for the $20 month version. So good point. It's pretty good deal. And then they don't have like a business version that jumps right to an enterprise version. It's like $249 I don't know what the tokens are for that. But yeah, you're going to have to look at all of that. But certainly cursors is good starting place. Clawed is great. Bill Swarsbuck. I would agree. I probably need to elevate to Clawed at some point. Gemini CLI is fast there. I don't know if they've already done some things that are actually better than Clawed, but certainly they're a competitor. And you probably should take a look at it. So if you're a student, that's probably the place to start. So I am not an expert in this next topic. But fusion startups are popping out like crazy. $7 billion have been popped out there. But Jim has studied this pretty intensely. And it would be really great to have some of Jim's views on this. He's got some nice diagrams we're going to show by Hey, Jim, what do you think about fusion? I love fusion. That's the future of power. And as soon as we get there, the better. And you know, every year people would say, 30 years, we're going to see fusion next year. 30 years, we're going to see fusion next. It's just the same repeat. It's on replay. We'll learn over it. But now we're seeing, yeah, for good reason, it's dangerous. So so now we're starting to see some major improvements and some great steps forward that now we can probably see it as a reality within three to five years. So that's exciting news to me. I mean, I like this kind of stuff. My son also studies a lot about nuclear energy and fusion and cheer. I know he would be a great person to talk on this subject. But I'm going to take a stab at it here. So there are 13 private startups out there that are in the fusion project market. They're trying to be the first to market with fusion power. So they just find a way to look at this to explain what fusion power it is. Yeah, fusion is exactly what it's talking about. Fusing together, two molecules of a fuel. And when from that fuel, you derive an immense amount of power. It's like an atom bomb, right? Like that's the idea of fusing those two together. It's the same idea. But it is the power of the sun. This is the power of a star. It is power very hard to contain because it is so immense. So they've had to figure out how can we just take one atom or one molecule of a particular element. In this case, I'm going to talk about TTIE because they are using a new fuel. And that fuel is a combination of hydrogen and boron. And they take one molecule of each and fusing together. And from that, they can get, oh gosh, megawatts. We can show the Tomac reactor. This is a picture of a reactor that is also known as a stellarator. So it's a way to contain this fusion in a plasma beam. Or in this case, this is kind of a circular donut. In that particular reactor, I'm going to show you another reactor in a little bit. That's a bit different. But this one is the one that most of the fusion companies are trying to replicate and create. Now remember, this design is like 50 years old. And for 30 years old, I think. And it's, it originally was a huge reactor. And now they scaled it down and made it to where they can build something economically so they can try to create this power by fusing these different types of fuel together. The most common fuel that is out there right now is a called DT. It's called Duterium and Tridium. So one molecule of each fused together. Do they use some? I let them crystals in that gym because we all know that they have crystals. Yeah, unfortunately, nope. I think that's all prediction. This thing is this like a room, you know, big room, great big room. Yeah, pretty cool. And I don't know if they're going to scale it up to make it much larger. But this is for testing purposes only right now, right? But there's problems though with using Duterium Tridium. And here's the things. One is they're 80% neutron loss. That's one of the problems. The neutrons just escape. They just pass right through everything and they're long. Is that dangerous? Yeah, because with it's radio activity that comes with it. So it's radioactive, active, active damage that happens to the reactor. So after a period of time, the reactor has to be replaced. So that's a problem. So and the third problem is that they need to use water to cool it. So there's a huge amount of steam that's required. And so they're trying to use steam turbines, which there's a huge amount of energy loss with turbines as opposed to direct generated electricity. So those are the three things. Also Tridium is incredibly rare. So Tridium Tridium is created or bred inside of a reactor using lithium, which is another element, which is difficult to get hold of. And after all of this, we have this whole problem of waste. All this waste product that happened. So TIE is another company who's come up with a different fuel source, which I thought was really exciting because it's just really hydrogen and boron. And they call it P11b. So it doesn't release neutrons. It's byproduct helium, which is totally different kind of system. Yeah, completely different fuel. It'll be the same kind of system, but different fuel. But the difference is in the fuel types is how you get it ignited. Right? How do you get it started? How do you get the fusion to happen? How do you get these two things to come together and fuse? So with ET, the Tridium Tridium, it takes 150 million degrees Celsius to fuse them together. Okay. I think that's I think that's hot. What do you think? Gus, I can't even wrap my head around that. So in in the TAE version, and we'll take a look at this other reactor, they are using a different method to cause the fusion to start. However, this material that that they have chosen takes between one to five billion degrees Celsius to start the fusion. All right. So now the resist isn't going to happen anytime soon. You said five years? There is not a exist there is not any kind of material that can contain that kind of heat. So they've come up with a different idea of how could they contain the actual the actual what's the word for it? The plasma the actual plasma at that kind of heat so that it doesn't touch any of the size of any time of a reactor. And what they've come up with is using these giant particles and you can see in the diagram or the image that they've got these giant particle beams that shoot at the plasma and keep it away from the sides of the reactor. Oh, so it's kind of like having some magnets like those things you can buy the globe. And it's got magnets that basically suspended in air and there's no magnets in this. But yes, same idea in the tritium in the tritium in the center of the reactor instead of on the edges where it deteriorates the reactor. Is that right? So the DT in the DT fuel. Yes, that's right. But not with this not with the hydrogen and boron. This is a different system. And so they don't use magnets. They're using these giant particle beams to keep blasting the actual magnet magnet or whatever you want to call it the plasma the plasma. The thing that goes as hot as the sun and it's it's being rotated and kept away from the side walls by these giant particle beams. 93 million miles away. 93 million the sun. Oh, yeah. Well, this is going to be right inside of a reactor. So yeah, I think I'm due back on the planet Earth here. This is crazy. Yeah, it is crazy. But the DT fuel can generate 17.6 mega electron volts, which is about 1800 kilowatts of energy for every one kilowatt hour of energy they put into it. So it's one to 1800 ratio, right? Really good returns. Now the P 11 B fuel reactors, they only return 8.6 mega electrons, which is only 30 kilowatts to every one kilowatt expend it. So it's a one to 30 relationship. So much more eco-friendly for the environment, but returns on on the electrical power is much lower. You know, the question one of them does like 1800 times ROI and the other one is only like what did you say 30 30 times? Yeah, but that 30 times one is the one that has billions of degrees, right? Not yes. So the concept is they're pushing molecules together and that's the byproduct is energy. Whereas with the atomic bomb, wasn't that separating the atom? That's where the energy came from. Sort of like the opposite strategy. Yeah, so I guess it's neutron bomb, not the atom bomb. Yeah, atom was splitting the bomb. Neutrons are fusing. So it's the same idea. Over my head. Now Jim, when we were talking the other day, you had mentioned that Microsoft signed an agreement with one of these companies. Can you tell us about that? Yeah, Microsoft just signed an agreement with Helion. So they guarantee that they would produce electricity by 2028 or return all the money back to with this method? With it. Yeah, fusion, using fusion. Yeah, whatever fuel they use are going to fuse together and create this energy. But they've already made it a hardened agreement with Microsoft that they will deliver. Well, they spent all the money. How are they going to give it all the money back? I don't know. I have no idea. They're probably a big corporation. I want to know who wrote that slide deck. Yeah, sure. So basically you got clean energy, right? So absolutely. Advantage. And there's no risk of meltdown like China syndrome type of thing. Is that true? If you stop fusion, everything stops. So yeah, if you don't keep it going, it just turns off. Right. It's not like that's wonderful. Yeah, it's not like our nuclear plants because you have this, you have this rod that's been now excited and it's going to stay hot. You can't just immediately turn it off. It is going to take a long time to cool it back down. So that's not the same thing as this. You turn off fusion, you take a way, the fuel from it, it stops. You have to keep giving it fuel. So what is the promise here of fusion? What can it generate a lot of electricity and make a lot of money? Can it change the way that we use energy? Can it negate the need for fossil fuels in the future? What's the story with it? I mean, it sounds like it's basically putting the sun in a box and you got the power. Yeah. Clean carbon free power. And the early customers are not going to be households. They're going to be data, data centers, industrial plants, defense and solutions. That's where the all the efforts going to go first. And then it'll get down our average users out here when it's been tried. So there's more than just two different fusion approaches, right? There's like what four or more different ones. Yeah. It's someone's going to win that. Or someone's going to win the battle. So from a business standpoint, it's a winner take all race, right? Whoever gets their first going to run with it and the profits go along with that. They estimate it's a trillion dollar annual energy market trillion dollar annual energy market. Oh, you bet. So and like you said, it's not going to be households first. It's going to be data centers. It's going to be industrial plants, defense, installations, places where have to be always on carbon free power worth a premium because the price might be a little higher to begin with. Confusion today sits at a kind of a real dangerous intersection because it's real science that requires real capital. And now we need real deadlines. And so we've got 2028. Somebody's put a stick in the stand and said, I'm going to deliver power by 2028. That's going to push every other company out there that's infusion to deliver the same. So the question isn't I'll take the over on that. Yeah. Yeah. It's their first. Right? Whoops the name of that polymarket. Can we bet on this? Oh, yeah. Let's check that out. There you go. I will bet on it. This is just wonderful. I mean, it's a great solution. I just don't know if it's realistic. Yeah. But thank you. Well, it hasn't been today because I remember all of the people in airports over the years back in the 70s where they were pumping fusion and the idea of fusion because conceptually it works. The trouble is making it work practically in reality. Now, speaking of convergence, right? You know, Waymo just hit 450,000 weekly rides. And what does that mean? That means that they are doing one ride every 1.3 seconds in the markets. But they're only in like four markets. So it's not like a burnbuster. But it could be because of there in four markets and they're doing 450,000 weekly rides, then you times that by all the markets in the US and you've got a very huge amount of money. And of course, Waymo is owned by Alphabet, which is Google and are affiliated in the same holding company as Google. So anyway, there's been problems with Waymo. For instance, they're driving down the road and they're pouring brand new concrete. And the Waymo sees it and it's like, oh, that looks good. Boom. Then it stops, right? Well, a human driver would know that that was wet concrete, but the Waymo didn't know. And those are the sort of things that we can predict might happen. But until they actually happen, then we can do a detector for that type of thing. And until then, you're watching a very expensive, very public beta test with real human beings as the guinea pigs, right? Yeah. Good point. Yeah. I had an experience with a Waymo where I came up on a four-way intersection and I was going straight ahead and the Waymo turned left in front of me, cut me off. Just completely cut me off. It should not have turned left. And quickly, thankfully, I was quick enough to stop and let it pass by. But I thought because you're a human body to complain to you about that. It's an upper. It just let him go. Yeah. Well, my wife and I were doing a run in downtown Austin. And during it, they closed the streets off and stuff, right? So there's police officers moving as the runners go down the Congress Avenue. They have a bunch of motorcycle cops and they go forward and stop the traffic. But the traffic signals continue as they have normal. But the police have turned off and stand there and say, don't go. I saw an experience. I can't remember it was a Waymo or a different brand. But nevertheless, the cop was there and he's like putting his hand out and the car just keeps coming toward him. Now it's not running him over. But we're talking about six or eight inches from him. And every time he steps back or moves or a little bit, the car goes forward a little bit more. That's why they've only rolled it out in a few markets. And I think they're smart. Obviously, Google has the power to you know, to wear with all to roll it out in a major scale. But they're being very cautious. And they have their test markets. And I feel bad for you guys in Austin with stories like that. We don't have them in San Diego. But it is interesting that they've taken this slow approach. And I'll be curious to see because you know, you mentioned that they have what, 450,000, it was only about 100,000 just a few months ago. So I mean, it's accelerating. And it's cheap. The average ride is three dollars out there. It's going to end driverless, you know, like they are. It's going to be even cheaper because the Waymo cars are very expensive. And I'm told that the all the LiDAR and all the different components of a Waymo cost upwards of 45,000 up to 45,000 dollars per car. The average ride is three dollars. That's crazy. That's very cheap. Yeah. So however they're doing it, but yeah, you're right. If they can cut that cost down, think about the adoption level at three dollars or right. Yeah. Well, they say that robotaxi from Tesla is like less than half of the Uber ride. I don't know how it compares to Waymo, but less than half. But you see the Tesla is equipped intrinsically already with all those cameras. And so there's no additional cost. You can buy a full Tesla for 45,000 dollars. When you're talking about putting a Waymo out there, first you buy the car for 60,000 dollars. And then you put on all the equipment that we're looking light. Yeah, they're funky looking. They're funky looking. There's no doubt about it. And you know, that's 40,000. So right there, you know, it's not just the cost of the car like it is with Tesla. So the cost of the driver, I'm not sure exceeds the cost of the Waymo with the automobile combined. It's pretty well. All of these technologies, all these technologies we're talking about today have a trust issue with all of them, right? So we have a trust issue with TikTok's algorithm, who's in charge of that? We have a trust issue with AI coding tools, whether we can trust that they're secure. There's a trust problem with the fusion startup. Are they going to blow us up? I mean, are they going to create more problems by using this technology? And then also now with these autonomous vehicles, that's been interesting.