What's Going On At OpenAI?

Now hello and welcome to the AI revolution. Call. I'm Cody Willard. There right is Bryce Smith and I'm John d'.

Oliveira.

John D. Oliveira. So we like to say John Dio. Follow him on Twitter. That's the handle. Bryce and I run a couple of tech funds. We invest in revolutionary technologies I. E. The AI revolution. We've been in Nvidia since 70 cents. Oh. In my personal account I have been a fund launched six years ago. Nvidia was already, what do you think, probably 15, 20 bucks when the fun launched. We started buying it. Anyway, we, we like the revolutions and the AI revolution is the most obvious one that's creating trillion dollar economies right now. So we're here to talk about it. There's a lot of news. It's been like two or three weeks since we last did this call John, is there something outstanding like that you want to hit first? Mind blowing much?

Yeah, I'll say. I heard it. I don't know the level of truth in it or whatever is. Meta was making offers to try to get people from open AI to hire them and so they're trying to, they'll give them a. Let's see if Cody heard what I said.

But yeah, no, no, this is all not true. This is all scam Altman lies. You know, your source is the biggest liar in AI. I didn't hear it from him.

I, but I heard him asked about it.

Yeah. And he lied. I mean that is what it is. There's nobody getting there first of all. There's nobody turning down a hundred dollar off.

Okay, see I, I wasn't gonna, you know, I was waiting for the call.

Yeah, exactly. And number two, there's no way that that Meta's like willy nilly offering random engineers over open AI a hundred million dollar signing bonus.

All alive.

So.

Well then it's, it's a big story.

Stop lying.

The level of lie is the big story. Then it is.

That is, that is some extreme lying going on. You saw by the way that that Google IO iyo, which makes smart headphones is now suing Scam Altman and Johnny I've for stealing their ideas and using the name IO and you know, again, everything Sam Altman touches apparently is somehow dirty.

So like it's not only the name they're stealing, they're still it. Because I hadn't heard what the idea was.

It was smart devices that you talk to, you know. I mean.

Yeah, but is it. Well, that doesn't say much. I mean I, I, how do you.

Can'T Like I'm just saying that simply they are being. It's. Once again Sam Altman's being sued for doing something slightly dishonest. That's my point.

Is it just the name that he took or is it.

No, I think it's also, I think they're saying that Johnny I've and Sam Altman all knew about Google IO and they like, they were pitched on it and they followed it forever. And Google Iyo, Google Iyo, Google IO.

Io1 IO audio first IO audio computer with earbuds. And you're saying that's what Johnny Ives was making was just a earbud thing?

Apparently.

I just want to say I did get a call from Mark at Zuckerberg. Of course I did turn down the 100 million. That's how much I love. I love being here. So.

Yeah, well, of course you're hoping for an even bigger payday over here at 10,000 days.

You know, as, as my father used to say, I never struck out as a New York Yankee.

So.

Yeah.

So look, I mean, at the point of. I guess let's get back to on track here, that Sam Altman lied and said that Meta was trying to poach their open AIs and engineers for $100 million a pop. And I don't think there's any way that's happened or going to happen. And but it is amazing. It does underscore the fact that people, the fact that we like all of us. When I heard it, I'm like, no, but it was for a moment I'm like, yeah, maybe. I mean baseball players make $50 million for throwing a ball. Like the AI revolution, engineers are actually going to change the world here. And you know, you could see them.

Getting that kind of like where you.

Make that kind of money is an equity, not a signing bonus.

Well, a lot of times with equity, you sit around and wait for something to happen with your stock and nothing done.

I'm just saying re. That's how in the tech world, how do you make a hundred million dollars is not a signing bonus. In the New York Yankee world, you make a hundred million dollar signing bonus, but you don't get any equity in the New York Yankees franchise.

But is that just a mistake of how things have been happening versus things there now? Setting aside it's a lie, right? But what would be bad about signing.

Bonuses for, you know, as a shareholder in these companies? I want the engineers to be dependent on. I want their crazy wealth to be dependent upon the stock going crazy higher.

That means.

Hold on, let Me just finish that being said, if, if, if you're Meta and you're spending $70 billion on CapEx this year to build AI infrastructure, $100 million is literally a rounding error at that point. So it's not inconceivable that those kinds of bonuses could be being thrown around. I'm just saying I 100% don't believe it.

If you were talking about like the top notch, like the man. Yeah, yeah.

I think.

What's your cousin, Alex Bregman of AI Revolution.

There you go. I'll let you bring it up.

Yeah, I think the speculation is that it's Ilya Suskiverse who maybe got an offer that night.

He's not there.

That I actually could. I. I know he's gone.

I thought he was still the head guy.

No, the other guy. Alex. I mean Alec. What's it, Carpet?

Meta sniffed around and tried to buy Ilya's new company and it didn't happen, but. Oh, they are trying to hire Ilya.

I mean a company called like asi.

Yeah.

Going after Super Intelligence.

Intelligence directly.

Like no. Little, no, no way, points in between kind of thing because I think is bogus. But some of these people. Yeah, you did great at one place. I just don't know that they have the, the multiple hits in them.

I. This is always how it is. Other than the PayPal mafia, which even a lot of those guys did not really build much wealth. Afterward, there's the, you know, you. The halo effect of being at an incredible company. Let's put it this way. Chamath Palapitalia was Mark Zuckerberg's right hand guy and Chamath has done nothing but scam people for the last six years.

I wasn't gonna put it out there that you don't like him, but you said it before, so you just.

You know what? It sucks. I like, I'd rather be one of these like permeable blind tech investor pundit guys who loves everybody who does anything in technology all the time. But I happen to be this like cynical revolution investor guy. So on the one hand I believe the technological revolutions, but on the other hand I know that there are scams and frauds and bullshit and bad marketing and pointlessness of Wall street machinations that suck money out of the system. And so I get mad at Chamath for being a scumbag and I get mad at Scam Altman for lying and getting fired and somehow weaseling his way back in it to Chat GPT.

The guy I was thinking of is Greg Brockman, he's still one of the founders.

Yeah, Greg Brockman's still there.

He's the one that got, I don't.

Know, like, is he even technical? I know he is, but I mean, Sam Altman.

Yeah, he's a technical. He is technically adept at financial and corporate engineering.

Yes. No, Brock was the only one left from that original group of them, including the woman who was briefly CEO for like two days.

Mira, I saw that she just raised money at like a 10 billion dollar valuation or something. And they don't have a product or. I mean, they.

Yeah, it's. It's the crazy. It's night. It's 1999 all over again. You know, look, I have on one side of a page a brief description of what it is I want to go do.

I just push back. It's not 1999 because these companies are worth 10 billion and like Meta really is spending 70 billion on this stuff. So it's not just VC money that's being spent and IPO proceeds that are being bidding up Cisco routers and sun microsystem servers, but it, I mean, like being able to have on your resume that you worked at, that you were a founder at Chat. GPT pretty much at this moment in time guarantees you a 10 billion dollar valuation. Can I finish on while we're at it? You know that Sam Altman apparently lied and said he was Y Combinator's chairman and he never had the job.

It's a technicality. He was offered the job and something.

Well, it seems like there's a lot of technicalities that are untrue around that guy. And then did you see that Chamith is like. He was like, hey, Twitter, should I do another scam? Spac. And Twitter was like 85, 90%, like, hell no, don't. And he wrote back the next day like, yeah, I think I'll do it. My theory. And then they. Those scumbags at the all in podcast launched some stupid high, like probably 300 a bottle tequila business, ultra high end, whatever that is. Those guys, all that stuff drives me crazy. And chairman, I'll short whatever stock you bring public because you know it's going to be a piece of. Because you are too. Like, it is what it is.

Cody says all of that and then he goes. And now look at my new.

Like, what, what am I gonna. What am I.

What are you offering? I don't know.

Have I ever scammed anybody like Chamith Palpitalia and scam Altman?

No, never involving money. What else is See, he said, what else is there? By the way?

What other kind of scams are there? In all seriousness, we're in the public forum, dude. Like, don't accuse me of scams over here.

You just.

I accused you of scamming, not involving money.

So what does that mean is what I'm asking.

It clearly means absolutely nothing, which is why you shouldn't worry about anything about.

The whole public forum. Public. That's not very nice to throw it out there like that.

No, I know. Just like you just did to Sam Altman.

Well, he is a. He is a scam artist, that's why. Poor guy, he's just trying to can consume me if he's got the guts.

Wow. Yeah.

No, for the record, Cody is not a scammer involving not money or money in any way shape there's anything in between those two.

I don't know. Really appreciate those kind words. Yeah, let's get back to the AI revolution. In all seriousness, there are some amazing developments happening and I think we probably should like get off of the yelling around scammy and chamber.

Oh, like wait a second, who started the scam talk?

I'm saying I should stop. We should stop talking about those guys and talk about the good stuff, because that's what we're here to talk about.

Okay, what's good stuff?

Bryce?

I built an app. I'll just shamelessly self promote.

Let's start there.

One of the cool things, it's not about me, but the Google AI Studio, I think is the most underrated AI app that you can use, and you can use it to make your own AI applications.

And is Firebase involved with that?

Yes. So there's the, there's the Firebase AI Studio which I've tried to play with and I don't find this helpful because I don't have much coding and back end developer experience. I don't have any experience on that. But Google AI Studio is so simple. I can literally just buy code in English and make cool apps and you know, I can paste an article in there, like turn this article into an interactive app that people can click through and learn that way instead of, you know, reading a page. And the Google AI Studio just does it. And then you can link it up with the cloud and publish it. You know, you don't have to make a domain name or anything. It just, it just exists and it's out there. You can share it or you can.

Embed it on in an iframe on our website, which is what we've done on Like Bryce built a sentiment tracker on our, for our subscribers. And so they can at any point go on the sentiment voting page, which is an iframe web app that Bryce built, and pull it up if you would. Bryce, I think that's what you're doing anyway. Yeah, it's pretty amazing. John, he's. And then just the other day, Bright, you know, I'm always asking Bryce, hey, what's our long, short positioning in the hedge fund? Delta adjusted, not delta adjusted, blah, blah, blah. And the other day he was like, it's like, I'm gonna vibe code of a web app that can track our long, short exposure in the fund. So here's the web app on the web on the tradingwithcody.com website, which tracks my subscribers, our subscribers sentiment.

Right. So you can you come in here, you vote, you know, maybe you're bearish because of the war, geopolitics technicals or. Oh, there's war as an option. You submit your vote, right? And it tracks it. And then Gemini, we have a built in Gemini API which interprets what the subscribers sentiment shows here. And then here's, you know, it tracks it on a, on a weekly basis over time. So I mean, it's a cool, it's a cool little thing that I built in an hour, probably.

And if you were going to try to do this two years ago, you would have gone to fiverr.com or upwork.com and hired some freelance programmer and hoped and prayed that you could convey to him what it was you were trying to do and get it built for thousands of dollars over weeks of work.

Hey, Bryce.

Yes?

What? Did you. Did it do what you intended or did it do some stuff and you're like, oh, that looks good. You know, hadn't thought of that, but. Or like how much of it is like you saying, put this in the upper left, that kind of thing.

But it's, it's both, John. I think that's one of the biggest lessons I learned in the vibe coding is, you know, you kind of come up with a concept, but then it's truly like this synergistic thing between you or collaborative thing between you and the AI. And I'm like, oh, you know what? I think I want this subscriber sentiment stuff over here. And I tell it to move it to the left and bring the tracking thing up to the right and then it does that. Maybe I want to change the color scheme to dark blue instead of red and it'll do all those things. It is like a joint Back and forth. It makes the core functionality pretty quick. Then one of the other things I'll do is how can we add more functionality and just have it generate ideas? And I'm like, oh, I like the third idea you had. Let's add that. That's what vibe coding sort of is. It's not just like one prompt and you make an app. It's like you work with the AI back and forth to make it really cool and functional and pretty and how you want it to look.

And perhaps most importantly, out of all of this, the takeaway is that Bryce and I have been talking about with you, John, for months at a time that we think the idea that software developers jobs are suddenly at risk or something is short sighted. Because this is an example of Bryce and I never really thought about all the different software web apps that we could use in our day to day lives to make ourselves more productive, more efficient, smarter, more engaged, content creation, whatever it might be. It is a constant now with Bryce since he started doing this with Gemini, Google Gemini AI stuff. I mean, it's just like every day he's like, oh, here's another thing I built. Here's another, like it's what we had.

Said where it, it's what we said where it's like you're, it's like demand goes up.

Yeah.

You know, you, you have this thing and it's like you haven't put any program out of business because you weren't hiring them before. But you know, you go to do this thing, you're like, oh, I can do that, I can do that. And you expands on even like the demand for.

And you would just picture that, you know this Bryce doing this creates demand for server compute. It creates demand for different layers of the stack of software and hardware throughout Google's cloud. And collectively, as more and more people are vibe coding random web apps that make themselves more productive and more efficient, there will be more demand for coders in the stacks at Google itself. And needing to make this stuff more pliable and functional and better and faster. Bryce, you're about to show something else. I think.

Yeah, this is just another app I made to show go to an investment conference in Vail last week and we presented on Google as our investment idea. And so instead of giving them a PowerPoint like everybody else does, 500 bullet points on it, I made this, you know, interactive app. It's got our investment thesis in here. You can click through it. You know, it has basically our take. You know, this is from my article I wrote about Google. And then you can, you know, again, it's got Gemini built into it and it speaks in Cody's voice and you can ask the Gemini questions or we have some pre built prompts. You can ask the Gemini about this specific part of the thesis. So again, just another app. You know, I spent three or four hours making this one. This is a little bit more extensive.

And then John. Yet here's another example of how this sort of snowballs, right? So we've been talking about for the last couple three months how Bryce and I are playing around with making a digital clone of Kodi that people can talk to. Literally my voice, my. All of our analysis programmed into it. So can try to convey some of what we're. How we see the world and when it answers a question, questions and whatnot. And so Bryce was trying to get everything on tradingwithcody.com into that new Kodi clone and he was having formatting, you know, like trying to get it all to talk. It's behind a paywall, some of it. Nah.

And that.

And so then he was like, hey, Google AI, can I. Can you create a file? And it created a JSON file and Bryce uploaded that to the Kodi clone and now it has literally all of our latest thinking that's even behind a paywall. And it was only done because Google's AI coding vive coding productivity enhancement to Bryce enabled to just sort of. Yeah, let's make that.

So the JSON file was like an export of all the content that you needed from this place the other. And then you got this thing and you go load it in and now it's got all the context it needs.

Because the JSON file was an export from the website, it was too large for to upload into this AI brain. So I built an app that can parse the JSON file, pull out the stuff we need, and I can sort it by date. So I just want the stuff from 2025 because we open source everything that's older than a year anyways, you know, so get the stuff that's behind the paywall, turn it into PDFs or HTML files and then upload those to the AI mind.

What do you mean you wrote the code? Does that little Google's help or was that how that work?

No, yeah, it was all the AI.

Yeah, but it was code. It was that was it used. I mean you together with it, you ended up with code that it used to filter out what it needed.

Right?

Yeah.

That's pretty amazing that Bryce as a non Software developer with no background in it at all is able to vibe code to a point where we're getting productivity and content enhancements and everything else. And then beyond that enables like it's also still the first inning, like the top, you know, first pit, let's call it the first out of the first inning. And so just imagine in six months how you're going to be able to just tell your computer to do stuff with or even in two years just how interactive being able to create content, code, web apps, vibe stuff, whatever you need in your life to make it better and more productive, you just tell your computer and it uses AI to create it.

I had a similar experience last week. I was working on except mine was down in like creating back end code, not the full app with the front end to it and all that kind of thing. So it's a, you know, coming at it from a different angle. But it was still have the conversation with the AI and together work on whatever and is making this, this thing that has to do with ontologies. We don't have to get into it whatever. But it's fairly complex stuff and in three days I with it wrote over a thousand lines of a module. It's a thousand lines of code and fully functional that you know, it connects to AI in a couple places and parses through stuff and whatever.

So John, like Palantir's big moat for example is their ontological approach and years of building ontological framework for AI in various applications and industries. When I hear you say that and I hear Bryce doing what he's doing and just it it are how good are the moats going to be here in five years or does like the importance. We sold Autodesk this last week because it feels like they're not all in engaging investing everything they can and making their a. Their data make the best AI designer stuff in the world. And it feels as Bryce is doing all of this and watching the AI revolution play out in real time, it feels like the AI itself is getting to the point where it doesn't necessarily need Autodesk's unique code and data and reality embracing of how gravity needs to be designed in architecture and stuff. Because the AI itself is doing that. It's get. It's encompassing that. And I just like. Are the moats around the world just going to slowly disappear because everybody they want just talk or are.

Is it like some things, is it that what Autodesk had that was a moat becomes actually a liability for them because they were all in on doing it A certain way and have a lot of very proprietary code to go handle a certain thing. And then as you pointed out, you know, AI now does that thing without what they were doing, you know, and that makes you wonder like, what do you think about Adobe?

Oh, same thing, perfect example. I'm glad you brought it up. Like the exact same analogy. It's like for, you know, 20, 30 years, people would pay way too much money to have a subscription or a disc originally that would allow them to use software to make pictures and to edit pictures and to put text in pictures and to make better presentations. And like Adobe's moat seems gone. You literally can just Talk to Like 10 different websites, LLMs applications on your smartphone and make the better, you know, the most amazing gorilla picture carrying your daughter on his shoulder.

Like I don't know though, from what I've seen with stuff that we're up to, they seem to be maybe trying to be a little smarter than Autodesk about it, where they are kind of looking what's out there and trying to rapidly incorporate and then go, yeah, you could get these 10 things, but it's 10 different things and we integrate and you'll be able to come to us to one place for the video and the image and the whatever. And they seem like they're trying to play that game. I don't know.

I think you're totally right, John. It for the last year and a half, two years, Adobe seems to have clearly at least seen the threat that AI presents and to have gone all in as selling themselves as an AI company that is the best way to do image and video generation. Still. I just, I think the, I think those LLMs, the app, the AI applications are already better and easier to interface with than Adobe and that and there Adobe's edge has already disappeared. And I think Autodesk is slightly, I guess it's not a perfect analogy because you're hitting on some nuances that are different. Autodesk is more slow, has been slower to try to get all in on AI, but they actually have a bigger moat because you know, making our like architecture designs that actually don't collapse is a fair, it's a moat.

But it's a niche too.

It's a niche, but I think it is disappearing anyway.

Yeah.

Your thoughts on all of this?

Yeah, I mean one of the interesting topics that's come up there was an interview with open AI scientist John and he was talking about this very thing and he was saying, you know, really what is your proprietary data? And he was talking about it's really just the culmination of hours of human labor doing things right. So like drawing blueprints was originally done by hand and then we figured out how to do a computer. So it's computer aided design. And if you have an AI with huge amount of compute behind it, you could have the AI make synthetic data with, you know, all the knowledge that humans have and make those drawings for you. And you could basically recreate a data package. Right. That you could then take that data package and train a new AI on that. And now you've replicated what Autodesk has took, you know, 60 years to build. Sort of, I mean we're not there yet clearly. But that was the premise of how he thinks the things are changing with the proprietary data.

There was a workflow that like Autodesk or any other kind of tools along those lines for architects would do that involved these like layers of what had to get figured out for different aspects. And you know, if you got an AI and it just jumps to the answer and you can't like get to those in between points, then that's not an answer. It has to actually do the work and show the work on, on those in betweens. So there's maybe some more to be done.

Sure.

The other thing about, the other thing about Dobie, they, they were going to acquire Figma and then they dropped the deal. And I know from working in, you know, the industry that uses this stuff that Figma was like eating their lunch. And so if they drop acquiring Figma when Figma is still like Figma came out and it was just incredible. It's like how can you do that in a web like page, you know, and have all the intricacy of a million kind of little shapes and layers and things. And it was doing what it needed to do to kick Adobe's butt at least in the web design space. So I don't know what that, how that plays out, you know either.

Yeah, it is fascinating seeing, you know, how quickly the revel the AI revolution is disrupting even what had been, you know, what you thought were big moats software as a service, high growth companies. And some of this stuff will, some of the stuff is overhyped. Like maybe Autodesk ends up actually figuring it out and having major moat and you know, like advantage, competitive advantage versus the LLMs. And so I don't want to make it sound like this game's over. These are still, all of these pucks and skaters are still moving around and we're just trying to Figure out which I don't think Wing Gretzky ever actually said this, but we're just trying to figure out where this.

I think if I were them, I'd be trying to make a foundation model that took my years and years of Autodesk, let's say kind of data in the way that you had with maybe probably Unreal Engine. And you went from that to like a Sora. It's like if you know how this stuff is done and there's the language of Go, you know, create use the language of Autodesk as the output of the LLM to go make. What's missing there? You know, something like that.

You know, Unreal Engine is another example there of I think a moat that might be disappearing before our eyes. It Unreal Engine, to give some context here is basically Unity has its own version. What's it called? Bryce or John 5.

Unity 5. There's 2 though, there's unity and Unreal. Right?

Unity is what Bryce Unity.

The name of their engine is Unity.

There's a Unity engine and the Unreal Engine and they basically forever have been the platform upon which video games are built. So if you're take two interactive and making the latest Y2K bat NBA 2K basketball game or your Grand Theft Auto, Grand Theft Auto. Most of these games have been built using the platform that just sort of again it's almost ontological. It provides some context of how gravity works in a video game setting and physics engine. And so it's. And John, that would be ontological.

It maybe it doesn't do the ontology.

Stuff but it's, it's an ontological framework.

Okay.

Anyway, the point being that it like these video, like these Sora and these now these AI video generation engines that you can just go to on chat, GPT or Gemini is like, it seems to be encompassing that, that physics reality just off of its training. And so the importance of those physics platforms to create video games are disappearing. And now people literally are going to these Grok or Gemini AI or whatever and vibe coding first person shooter video games. And it's the first out of the first inning. Just imagine in two years the video games that some creative 13 year old kid in his mom's basement is going to create that's going to blow everybody's mind.

Have you seen this? Like a couple weeks ago there was a video, you know, on YouTube or something and it was just lots of cuts to different places, like people being on the street, interviews and you know, all different things. And it was like everybody was like, blah, blah, blah.

So what do you think about the.

Fact that you're not real? And like nothing in the nothing was real, but everything looked very real. And you're like, okay. Because now it's like not just before, it was like, okay, this video, now you could put voice to it and move the lips correctly and have it be wherever. And there you go. It's catching up to Hollywood.

I think what hit me so for so hard, strongly from this investing perspective, is that you software has always commanded premium valuations. And now there's this big overhang on software because everyone is questioning whether AI might replace this software, right? And nobody knows for sure. And some software companies will probably be fine, some won't. But because there's now this threat to the entire industry, valuations come down. You don't get 20 or 25 times sales for most software companies. Whereas, you know, before AI, a lot of software companies would get those crazy high valuations. Now like four or five do.

And on that note, it is interesting from an investment perspective that software forever had the highest gross. Higher gross margins than say, semiconductors and certainly higher than just servers and computers and generalized hardware. But the semiconductor versus software analogy is interesting because Nvidia's gross margins have gone from 40% to 50% to 60% to now over 80%. And software gross margins have like, for example, even Microsoft is forever sort of diversified out of its software businesses and gross margins went from 95% to 85% to 75% and on down. And it is interesting that you can actually see the. The moat in gross margins. Like you can see Nvidia's got a bigger moat than Microsoft right now. That wasn't the case.

So it's not, it's not straight up like hardware has better gross margin. The software, it's. That better moat has a better gross margin than not better moat.

Correct. I think that's what. And that's probably the re. It's always. It's almost like a truism. It's always been that way. And that's why it is that. That's why you can have higher gross margins because you have a bigger moat. You can charge more.

Yeah.

We would be remiss if we don't hit on the single biggest AI revolution development over the last ever would be yesterday's robo taxi launch by Tesla in Austin. Thank you, ladies and gentlemen. Nobody died. Nobody got hit. No accidents yet. There will be accidents at some point. There are accidents when people drive. There will be accidents probably because of human Drivers, more likely than the car itself. Than the car driving, full self driving autonomy stuff. But it will happen. But let's just. Can we just step back for a moment and acknowledge that if we have $30,000 cars that are able to actually fully drive people around major cities, we have changed transportation. Not we, the world. Tesla has changed the world's transportation forever.

Well, first, weren't they going to be down to like $20,000 cars? The one that has no wheel and it's straight off the line and so forth. The model.

Fair enough. Yeah.

And the other thing is, wasn't they taught at least I heard. I don't know if maybe it's just a kickoff thing, but like a flat rate, like that's just.

For now. That's just a marketing boy. $4 and 20 cents. 420.

Oh, oh, yeah, yeah, yeah.

And taxes were 69 cents.

Yeah. They gotta be smoking something to use.

That humor for you. Ladies and gentlemen, Elon Musk and Rob Gronkowski. There's an. Those are a couple of guys.

It's just a little scary to the world that, you know, the, the, hey.

He'S just being funny.

The humor.

By the way, he, he published his drug tests the other day and like he's been. He said he wasn't doing drugs and the dust showed he's not doing drugs. So he's not out there. He even said he's like, dude, I smoked pot one time on Joe Rogan.

I don't know what.

He's not doing a prescription at 1.5 years ago to get out of depression. And it worked. Like, that's all. That's what he said.

But what's scarier? That he's not doing drugs or that he behaved as erratically as he did while not doing drugs?

How would. So dude, if you followed Elon Musk at all, you know, the guy's erratic. Like he's all in all the time. And that's gonna draw like there's a lot on the line all the time.

Yeah, yeah.

So at any rate, dude, if the launch worked, like.

No, let me say no, please. The launch worked. No, I think it's amazing. I think it's awesome. A step that, you know, we'll tell our kids about. Like, we'll say you were alive. You didn't pay attention. You know, this, this happened. This important thing I, at work today used it as an example of something which was, you know, we have some new feature that we introduced to our customers and it was like, oh, we gotta, you know, make sure it doesn't like screw up, you know, it's got it, it's important that it, that works and everything. And it was that notion of geofencing where you know, the robotaxi went and took an area of South Austin, however big, you know, but they said it's just this and they go do it and then they can like expand off of that as a way to be like, well, we'll make it work and then you'll, as it gets bigger, it'll still work. You know, it's like risk thing. And we used to, we used a similar, it wasn't geofencing but a similar notion with, with the stuff that we, you know, introduced. So anyway, I though I, I, I'm very impressed and I, I like, I can't wait for it to, you know, to get to the level of Waymo in terms of its saturation or whatever and then to just zoom past that. Yeah, I mean there's no reason once it starts doing that. That's right.

You know, the, the, the Bryce has an older model three that has like hardware three and we're on like, is it five now? They call it AI four. AI five. Bryce, what are we on one and two?

The current cars are on, on hardware four, AI four, and then hardware five will be next.

So four is out.

Four is out. Yeah. Like my cyber truck has four.

My wife has I think three. Yeah. I don't know how, how you can.

Tell Bryce's Model 3. You can actually go in, I think somewhere in the, you look at your.

Software, it should tell you what some, some like a version number of. Yeah. Hardw the hardware and then.

Yeah, and that Elon said that, hey, anybody who purchased full self driving is gonna, you know, Tesla's gonna have to eat it, but they're gonna have to put in the AI 4 at that.

Yeah.

When 5 in the, give it the next 6 months to 18 months, I would think.

18 months. Yeah, but they really need to do that because. Yes, we're, we're one, we had self driving from the beginning, but it's, we really need to be up a level on the hardware.

Yeah, exactly.

Even if they charged for it, I would do it.

Yeah, but I think they're going to do the right thing because you know, you paid for full self driving and you didn't get it. Yeah. So they're not going to charge you for that. But again, to give that some context, it costs Waymo $150,000 to put LiDAR radar and 28,000 cameras on a single Jaguar which costs $120,000 base price. So they're in for like 270 or $300,000 per car. And Tesla's going to be able to like switch out your cameras for the next generation's cameras and it'll cost them, you know, $500 in labor and $1,000 in cameras.

Let me ask something. So watching the video about the Robotaxi driving around there was delivering the guy to St. Edwards in, you know, in Austin and St. Edwards University and Marc is looking at it, she goes if that is just a Tesla using full self driving to do that then why can't my card do that? What's the difference between. Is just the software. We just said the thing about hardware.

But it's a hardware. That's it.

It's on. Five Robotaxi is on because lots of people have four out there.

Yeah, no. Okay. The difference between your wife's car and those is hardware and software. And the difference between my. I don't have a Model Y, I've got the Cyber truck but. And it's a different issue. But if you have a brand new Model Y and you pay for full self driving, you are one iteration behind those taxis.

The robo taxi is using software that's.

But they will correct just that kind of difference and they're going to send that software out to all of the people who paid for it even before you're able to truly, you know, even before you're in the area where it is able to do full self driving unsupervised. But you will get that next iteration within the next few weeks if you have a brand new Model Y with.

The latest and that area is determined by Tesla and in cooperation with a local government that says yes, you're allowed to do this here.

Yeah, yeah, always. I mean you're going to have to you know, get local permit and everything. But yeah, I mean so you can expect especially because there's already been trails blazed there by Waymo Etc. Elon said next is San Francisco but the low hanging fruit is going to be like San Francisco, Austin, Atlanta, where also Waymo's already at. You'll see it and I bet they'll. And then it'll be like, you know, state will. Regulations will also play a part. So Texas, you're already in Texas with that. So then it'll probably be Dallas and then you know, in Las Cruces, New Mexico where Bryce lives is hopefully sometime next year.

Yeah, I signed up to be on the beta tester. Whenever it comes to your area, you can go on the website and sign up. So I mean, yeah, it's just a matter of time, right?

Software and then permission from the locality or whatever it is.

Well, and that's part of why I think mistakenly the word the market really got excited about Elon and Trump being best friends because somehow like that meant that Trump and or his federal bureaucracy at some point was going to give like hey, rubber stamp to whatever Elon wants to do with full self driving and it'll override all the state and local regulate. That's just not how the world works. Anyway, you're gonna like. It's gonna be a process, it'll be city by city. But there are 2 million Tesla cars that have the potential to have their hardware upgraded and downloaded with the latest software and then turn on as full unsupervised, self driving, autonomous.

And I own one of those mods, let's say I own one of those. I own a new Model Y. Does it the new Model Y come anyway with that monitor facing the back seats? Just because it's like the backseat passengers can watch a movie like it's a thing that's in the car. Okay, it's already there. Yeah, it's already there. So then can I I buy one of the Model Y's? How soon until I'm my own taxi company?

So Bryce, I mean this is a matter of debate, right? I you, I would say 18 months again, just to sort of let it permeate, get established. Tesla's going to do it with their own owned vehicles first to establish, you know, a baseline how it's going to work. And even like those, you know, the brand, the reputation, the quality of the product they can control, the cars will be clean, etc. And eventually, you know, they'll have call it, I'd say 18 months to 36 months.

So if you had a, like a say, you had a fairly high percentage of people who decided who were Tesla customers, Tesla drivers who decided to kind of opt in on this and just go ahead and do that. How big would that make the fleet be? Like if I had 50 of Tesla users, how big is that fleet compared to the, the Uber people that are in my town?

I just, I can tell you right now, Bryce, you answer the question. How many numbers, number of Uber drivers.

Like are in my town?

No, I'll do. Bryce, how many Teslas are on the road is what John's asking.

That could versus how many Uber drivers? If I said there's this many Tesla drivers on the road, Is this many Uber drivers on the road. Tesla flips the switch on this all of a sudden.

Yeah. So, John, there are 1 to 2.5 million Uber drivers in 2022. That's in the US alone.

That's fine.

So.

Yeah. And that. So call it 3 million. Let's round it up. 3 million US Uber drivers.

On the road.

2 million Teslas on the road in the US 2 million.

Yeah.

So say 10 of them sign their car up for the fleet. That's 200, 000. That's 1/10 of Uber's driver base.

That's enough to drive down.

Is there's no tip on a Tesla because the costs are $30,000 for the model Y and not, you know, a car and a driver and everything else. The actual line will be a frac, probably 50 to 70% of what the Uber. So call it 30 to 50% less what an Uber is. And the Uber driver can only drive 10 to 12 hours a day and the Tesla can go 24 hours a day except for when it's charging. So call it 23 hours a day.

Like another way to look at the, like the business model on it or something would be like, I don't want to manage people, you know, get my car, whatever. I'm going to kind of hand it over to the Tesla fleet for those 22 hours when I'm not using it. And it will manage for me. Just like a property manager.

Exactly.

Yeah.

Yeah. Elon said that it's a combination of Uber and Airbnb. It's the. The model is a combination of Uber and Airbnb, which is what you just described, John.

Yeah.

Okay, so you're basically, when you buy the car, you're buying an asset that's going to ultimately pay for it. Not only pay for itself, but ultimately make money for you.

That's what Elon has promised for three, four years in a row that it will be the first car that actually appreciates in value when you drive it off the lot. Let me know. Like, I'm still saying it's 18 months to 36 months away before you generate money with your, your. With your Tesla. I think Bryce is a little more optimistic than me. That saying, Bryce, when do you think you'll be able to turn your Tesla into a robo taxi in Las Cruces for the weekend?

I'm gonna say this time next year.

So.

But regardless of when the starting gun is, what if that happens or do you think that is the case from there forward?

Yeah. Then it's the game I mean, the transportation's changed forever. Soon thereafter, Tesla rolls out robo vans, robo buses, robo semis shop, robo coffee shop.

You hop into the van and you're there. Have a cup of coffee while you're getting wherever you're going.

Correct. And, you know, it's a matter of, you know, trillions of dollars of potential revenue with very high gross margins. So.

And this is just a robo vehicle. Then you've got robo robot.

Yeah, that's. We're not even talking about the Optimus robot. Yeah, and look, we're not blind bulls. Like, there's a long road ahead and one Geo fenced area in Austin for two days does not make a 10 trillion dollar revolution. It's going to be.

I don't think. I think he's a patient bull. Like, you know, when you got into it 10 years ago, you're like, at some point and I'm young enough. So, like, at some point is good enough.

It's definitely six years ago that I bought Tesla. But point well taken.

Technically you held up seven fingers when you said that.

But this was this. Maybe did that. But I meant this.

Okay, I'm just a wise ass. But you knew that coming in.

I did know that. You're rough on me. Always have.

Oh, by the way, when I saw the.

I'm like the nicest guy to you and all you do is pick on me.

Is that true, Barb Rice?

You ever, John?

Never ever. Okay, now we know it's not true. So. But the, the other, the thing about the Optimus, like it was just a couple weeks back, there was a new video. And just before the video happened, I was like, you know, there's all these different robots and everything. Remember I was talking about this Chinese one that does this or that and kung fu and so forth. And I was like, the Optimus just seems so like slow and fragile and whatever. And then like the next day, a video comes out of the. Of the optimist dancing, like doing Running man and doing whatever. And I was like, okay, never mind.

Literally the most amazing thing of that dancing video is seeing the Optimus robot's ankles. Like, that's part of those dance moves. Like the knees, the hips, the ankles, everything is like coordinated and looks like a good athlete. And John, where when it comes to the Chinese robot versus the optimus robot, I mean, I think the quality of the hardware is on. A Tesla is going to be, you know, like the dish. If you've ever been in a crappy BYD Ev in China, you know what I'm talking about. There's even the worst, lowest quality, Tesla is 10 times higher quality than a BYD EV. And the robots will probably be the same thing. There will be some really cool Disney R2D2 robots built on Nvidia's platform, built in China and imported and probably with some sort of tariff because Trump or his predecessor in the Republic, last time he went off.

Here we go. So like yes, so Tesla has a couple. They have, they can like mill almost any. They can decide what they want to make and just make it. They have the kind of vertical integration of that, of putting together, going give me a like a roll of steel and out the other end will come whatever, a car, a rocket or whatever. So they can make all that stuff. And then they have the, the expertise of, is it called designed to manufacture? Like where as you're designing you have in mind how those pieces are going to come together, you know, so they do, they're just experts. And then scale, they have scale. So they're experts at all of that.

And John, perhaps most importantly, they have the technological prowess and the US base to make developers an ecosystem. And the Optimus platform becomes an operating system that has infinite number of apps and services that you can download. And so you could buy the fun dancing robot from China or you can buy the Optimus robot download the dancing app. And when it's not dancing it's taking out your garbage. And when it's not taking out your garbage, it's babysitting your children. And when.

All right, maybe, yeah. Actually the whole multi purpose thing is a really important question because you know, when we watch the dancing, did they load it up and at that moment it was just a dance bot where you know, there's all these things or can it like actually flow from one to the other? And do you know? I don't know the answer to this. Do you know are they using the Nvidia platform that does. They're not. So I'm talking about the platform that does the, the whole 3D world of the thing working in its animated place and once it gets it, it comes out and does it in the regular world.

Bryce, we should clarify. Look, you never want to say Elon's not playing with stuff like I'll guarantee they are using elements of Nvidia's robotics platform. But Tesla has been building from the ground up to this moment, like before the Nvidia had released their robotics platform to developers. Bryce, can you add color to any of this?

I would say we I don't know that we know for sure that they're not using some of Nvidia software like the Omniverse and that sort of thing to help with the modeling Sim, they call it sim to real and so they could be doing some of that. But the hardware in the robot is definitely the Tesla, you know, FSD chip modified for the robot.

Yeah, right.

Probably the Dojo supercomputer has been helping train the Optimus.

Well for sure that. But the Dojo doesn't answer the question of the. What the Nvidia is doing with that other piece.

Yeah John, I think it's, it's, it goes back to sort of that stacks of layers of code and hardware and I don't know that you could peel out and say there's no Nvidia in this hard Tesla solution at all. But I think Tesla's got major differentiation in a lot of the development and it's interfacing with the physical world because of all the work they've done with the full self driving car interface.

Yeah, but there's a piece there that, where it's kind of like the LLM of the motion world, you know, where you get from like talking about something to going and actually doing it where it is separate stuff.

Stuff.

Unless you are using one of these, what's it called, like a world model, you know. And what's his name who runs the AI at Meta, The French guy, John Lecun. You know, he's got the technology called V Jeepa, V Dash Jepa and it's different, it's different than the LLM stuff. And this, maybe this is a tangent. I don't, I think though there's a dependency in that kind of stuff. Where do you really get what, what's going on in the world where you are and could you tell like if you're in a scene that you know that thing just fell off the table a second from now it's going to be broken all over the place or whatever because it like can predict like what goes on in the world. It's world model and that's. So it's going to be, it's going to be important piece for all of them. And I don't know Tesla is actually going to make that from scratch themselves or they're going to go, that's what I'm saying.

They've been working on it from scratch, but I would bet along the way they are at least looking at incorporating layers of technology from partners like Nvidia.

One thing John, is that when they, they had some. This. The Tesla guys speaking, but they said they had a breakthrough and trained the, their AI model based off of just watching a YouTube video. And so I don't know that they necessarily used like the world model because you can use the world model to like have the AI run different iterations of the training, right. And see, and then start to fine tune it. But they did say that they, you know, showed it a YouTube video.

Yeah, I forgot they said that. That's right. And that's really good. It doesn't answer the whole question, but it definitely is. Is a huge step forward to build because if you went from, oh, you gotta have all this person in a suit doing this or that and having it imitate whatever their thing, and now they're like, no, we get a first person video and it can just see it and then it can. Yeah, it was a huge step forward.

So, yeah, I agreed.

All right, last words, fellas.

Oh, damn, I wasn't ready. Oh, my last word will be Cody is not a scammer in any way.

So now it sounds like the old he doth protest too much.

I just want to make sure it's covered. I didn't.

It's fine. It's. It's all fine. But you could, you know, like their time and place. But at any rate, look, I think give me a last word, something that people can take away.

All right, go ahead, Bryce.

I. I mean, we're, we've been pounding the Google table for a few months now that the stock sold off in the last couple days because there's more news about them being replaced by AI and antitrust and all this stuff. And I'm just gonna pound the table on Google once more and say, I think it's an amazing AI investment for the next 10 years.

Good. I like it.

I'll say it's. It's not, not, not because I was the one most picking on them and, and I seen more and more that, yeah, they got some, you know, amazing stuff.

So cool. I think the takeaway, my last words will be that AI is impacting the more the world more than even the most ardent bull realizes already right now.

It's got good one.

John, we need your last AI words.

I.

It's in the realm of this stuff, you know, what Bryce was talking about with what he built and how, you know, it's, it's not intuitive. The fact that, you know, demand goes up, so you think that it's knocking people out of jobs and so forth, what it's actually doing is opening up possibilities. So I think that has been it's start that's starting to prove itself.

Amen. Thank you everybody for being here. As Ameris likes me to say thank you, goodbye, gracias and adios.

See you later, guys.

Thanks.