Episode Transcript
[00:00:00] Speaker A: Well, hey, welcome back. Gregory and Paul show. I'm Gregory.
[00:00:03] Speaker B: And I am Paul.
[00:00:04] Speaker A: And we stream live every Thursday now at 11am Pacific. What time is it for you up in there?
[00:00:11] Speaker B: And 2:00pm, 2:00pm the East Coast.
[00:00:14] Speaker A: East coast time. Okay. Because we started this as an X base and we used to do it live on Fridays and then we decided to move it to Thursdays because of summer Fridays and lots of people taking time off on Friday. So we moved it to Thursdays and for today's show, we spent a lot of time preparing. Last night I was, I was very. I watched a lot of YouTube videos, I read a lot of blog posts. So to recap what we have in store for today, we will start with the OpenAI ads launch to the general public.
[00:00:45] Speaker B: You're excited about this one?
[00:00:46] Speaker A: I'm excited. I posted this. Everyone is not excited. They don't see it from the perspective of a marketer. They're not thrilled about having ads in LLM. I am thrilled about putting ads in the LLM. So we'll talk about that.
[00:00:58] Speaker B: Of course.
[00:00:59] Speaker A: Then we will. We will debate the Hormozi video that you loved. I wasn't necessarily as excited about it.
[00:01:04] Speaker B: Talk about it. So. That's right.
[00:01:05] Speaker A: I watched some videos. I got some perspective on Hormozi, what I think about him. We got Marc Andreessen got an entire take Truth taking Truth seeking AI prompt. I want to cover copyright and the meme.
[00:01:18] Speaker B: That's an interesting one.
[00:01:19] Speaker A: Yeah. Meme world, like meme copyrights, it's a curious gray area that would be interesting to discuss. Since we love memes. I thought it would be interesting to discuss a, actually a lawsuit that a meme copyright holder filed against a popular AI startup for using his content. Then we'll jump into a 16's very long, very well written, very well researched post that makes the argument the AI Jobs apocalypse is a complete fantasy. Fantasy. That was their words.
[00:01:53] Speaker B: Which is a perfect segue into fantasy
[00:01:55] Speaker A: Coinbase to the actual layoffs. Coinbase layoffs. We'll talk about Claude partnering with SpaceX and then we'll end on the trial of the century update Sam versus Elon. I don't even know if we'll have a. We could do a whole show. I actually think we should do a whole show on that. But we'll talk about that later. But. But first, the event is sold out. So.
[00:02:17] Speaker B: Yes. So this is the.
[00:02:18] Speaker A: This is the coveted Nano PAC man that the winner of the Vibe your SaaS startup pitch competition will receive. We've given one of These amazing trophies away. We don't have any prize money. You don't get any venture capital, you don't get any AI credits. You just get the ability to tell everyone that you won this thing, which is why it's so hotly contested. We had 72 pitch decks submitted, up from 22 from last time. We only picked five startups to pitch, which means just being selected to pitch is about 7%. I'm saying it's one of the most competitive startup pitch competitions in the San Francisco Bay area. I know that most of them, they'll actually let you pay to pitch. We don't do that at all. We have lots of really interesting, amazing, talented founders who sent us their decks.
So many of them are incredibly fantastic. I go through all of them. I have a team of judges. We review every single one and then we pick the five that we think should be highlighted to get an opportunity to pitch. If you don't get selected, you can always resubmit, maybe make more progress on your business. Maybe we pick you for a demo. So just because you don't make the cut in the first time doesn't mean that you can't continue to be part of the community. And of course, like, we try to highlight the ones that we think are interesting, entertaining, fun, doing something new, doing something timely. And so actually, for this.
Yeah, yeah, yeah, for this, for this, for this pitch competition, I, I should go through and tell you who the.
[00:03:51] Speaker B: Yeah, yeah. Give us a sneak peek.
[00:03:53] Speaker A: Yeah, yeah. There, there's some really, really interesting, really interesting ones pitching here. I'm gonna bring up the full list. Okay. So this is super cool. I'm very excited about this.
The Drone Championship League. It's a startup that a friend of mine in Europe referred me to. They produce the Drone Championship. Yeah, it's a drone racing league.
[00:04:15] Speaker B: That is sick.
[00:04:16] Speaker A: It's cool.
[00:04:17] Speaker B: Yeah.
[00:04:17] Speaker A: So they, they produce the Anduril AI Grand Prix.
[00:04:20] Speaker B: Okay. Okay, cool.
[00:04:21] Speaker A: Right. Then we got some security companies, White Swan, Identity Security. So we all know identity is a really big issue. And you and I have talked at length about security, cybersecurity, think it's a booming business. I think everyone out there, if you don't know what to do and you're trying to come up with an Idea for your SaaS, find a security angle. Yeah, very excited about security. Then there's a healthcare app called Cara that is a unique and innovative way for people to measure, manage their healthcare at home. Operational procedures and care that they, that they receive to manage in home, in home care and Then there's a company called Proxera, which is an AI governance platform for regulated firms, which I thought was interesting. We all, we've talked at length too, about, like, the challenges that Pentagon have had with anthropic and other aspects of, like, government regulated services requiring AI and they have different needs than the private sector. And then the last one is another security company, trustable, that focuses on sustainability performance. Nice.
[00:05:31] Speaker B: Excited to see the pitch? I'm very excited to see the pitch.
[00:05:34] Speaker A: Super excited. Last time went really well. Everyone gets 90 seconds. They get one slide. They get up there, they gotta be concise, they gotta be passionate, they gotta be excited about their business. We'll ask maybe one or two questions, but the judges have reviewed all of these in advance. And so that was one of my requests doing this was that we spent the time, effort and energy to think through, look at all the pitches so we don't waste people's time, like, asking basic questions about their business and public. If we want to ask a question or two, like, the judges can do that, but we'll run through them and then we will pick the winner live.
[00:06:05] Speaker B: Right on.
[00:06:05] Speaker A: Awesome. Okay, let's move on.
[00:06:08] Speaker B: Yeah. What's a. What's first on the docket?
[00:06:10] Speaker A: All right, so I'll explain this because I'm the one who's very passionate about OpenAI ads, and unfortunately, everyone else is not as passionate about OpenAI ads. Well, not entirely true. There's a founder I work with who does an LLM ad network, and he is probably more excited than I am about OpenAI's ad platform. But you've really got to be an advertising industry person to get excited about putting ads into something. So OpenAI has been testing this for a while. They've opened it up to. They started with, like, large brand partners letting them buy campaigns that they claim went pretty well. There was some notable, like, feedback publicly that people, people were not necessarily so happy with the initial response and performance that they got, but they've made changes, they made updates. I've actually seen the interface. I've seen campaigns running on it. It's quite interesting. It's pretty basic. I'm sure they'll add more features and functionality. Like, this is a area that's been figured out. Like, we've been. I've been working in ad tech for like, 15, 20 years. Right. Like, this has been around, so it's not like, gotta reinvent the wheel to figure how to do this stuff.
[00:07:13] Speaker B: So what makes it, what makes it special? What makes it special? Over go Google Ads. What is the new feature that they're launching with or what's available?
[00:07:21] Speaker A: That's a good question, Paul. I'm glad you asked. Look, advertising is about one thing. Scale. Open AI has scale.
[00:07:28] Speaker B: They have distribution.
I see, I see. So they can reach a lot of people on their platform.
[00:07:34] Speaker A: Look, dude, people probably don't understand this. Google is giant. Like the amount of traffic that they get. Gigantic, right?
[00:07:40] Speaker B: Yeah.
[00:07:40] Speaker A: Like when Pinterest and Snapchat and Twitter and Facebook all launch their ad platforms combined, they were tiny compared to Google. Facebook obviously has grown. It's become significant. They are a rival to the scale that Google has. But all these other social networks that people are familiar with that you would run ads on, they only have like a fraction of the scale of the traffic that Google has. Right. So OpenAI is one of the very few options when it comes to getting scale close to what Google has. Like billions and billions. If you try to go into a platform like Quora or even Reddit and set up a campaign, you can't even burn the entire budget if you have a big advertiser. People probably just don't understand the scale that some of these platforms have relative to other ones. Right. Just because they offer ads doesn't mean they're able to show enough ads to people to satisfy the advertiser. So I think that that's the most exciting part about OpenAI. They're really, really big and they can run a lot of advertising potentially through the platform. They're going to be other features and functionality because it works differently than how other ones work. And like, this stuff is pretty new when it comes to like, targeting and like how they're going to display the ads and they'll work through that stuff. But the promise that there's a new channel out there that has really a lot of immigration inventory available to people, I think it's quite exciting and I'm. And I'm excited about it because all the other channels are saturated. Ads are expensive. Maybe we'll get into this when we talk about Harmozi. Like, ads have gone up in the last five years by a huge, a huge amount. And it's really challenging for like, startups in particular because you just can't buy ads in an ROI positive way if you're small.
[00:09:27] Speaker B: So. So if I'm at indie hacker or early stage startup founder, should I care about this?
That's good. Who do you think, like OpenAI is targeting? Is it $100 million businesses? Billion dollar businesses, Trillion dollar advertisers?
[00:09:42] Speaker A: Yeah.
This is a good question. There's not a lot of information on this so I'm just going to take a guess like that they tend to be still a platform that be full of early adopters. So if you're like trying to chase.
If you're like a e commerce startup that sells a relatively straightforward consumer product, I don't think it's a good option. I would trust it. I could be totally wrong about this, but you should try it. But I can't imagine there's a lot of people on there asking questions about makeup which does really well like on YouTube or does really well on Facebook. That would be my guess. The biggest category on Google is travel. I don't believe the travel stuff would be built out enough on ChatGPT to like really create a fantastic platform for that. Like if you're trying to market a boutique hotel. Right. So I think the obvious early adopter type of products would do well. So that means like B2B SaaS, startups, other kind of high tech products. I don't know if people do a lot of search for like laptops and phones and technology like that that might do well on there. I would try to think of things that fall into like let's call it the nerd sector.
That's I think would do, that's what I would try. That's I think would do relative well. Apps might do well on there. Uh, that's my guess.
[00:10:53] Speaker B: E commerce. Any prediction for E comm? Do you think they're, they're all gonna try it?
[00:10:57] Speaker A: They're all gonna try it and like they might be able to cater some of the content to that which, which would improve the performance and improve the ability of people to like to, to run ads in that. So I mean I, I think it's where it has to go. I, I also believe that like I, I don't think it would be it'll end up any different than Google. Like if travel's the number one vertical on Google I would imagine eventually travel has to be the number one vertical. Vertical in terms of advertising on OpenAI I don't see how that wouldn't play out. It's very obvious use case. Right. I need to find a hotel in such, such place you go to Google and hotels have money to spend to advertise Staika. It's so obvious.
[00:11:33] Speaker B: So yes that ChatGPT search wise is taking is competing with Google. Do you think that with the ads platform they might be competing with Amazon in terms of distributing, getting people to buy things?
[00:11:46] Speaker A: David, we've talked about this.
[00:11:48] Speaker B: Yep.
[00:11:49] Speaker A: Competing with everyone.
[00:11:50] Speaker B: Everyone. Yeah. That's. That's, that's like comical.
[00:11:54] Speaker A: There isn't a vertical or an industry that they haven't decided they want to disrupt. From Instagram to.
[00:11:59] Speaker B: That's right. To the iPhone, coding, co work, E commerce marketplace, search. Why not? Why not?
[00:12:07] Speaker A: Yeah, I think so. Like, I take your pick. Right. But I, I don't. I just, I haven't used it or I haven't been successful with it to use, like ChatGPT do some kind of like travel planning thing. If they could get that to work really well, I could imagine that'd be the biggest I have seen demos where that is the use case that ChatGPT actually was designing for. Yeah, it was like a. An Airbnb ad or something like that. Right. So if your initial question, you're SaaS. Indie. Indie startup. Try it. Absolutely. 100%. Like, you could, you could put a small amount of money there. See. See if it works, you might end up with people who are like, really highly qualified. That's been the result of all the research that I've done when it comes to aeo. So everyone says that the people that come through, that they click on a link through organic AI search, they tend to be much more qualified than people come through Google, which would fit my, like, description and assumption that it's early adopters.
People that, like, are pretty sophisticated when it comes to technology. When it comes to being on the, like, cutting edge of a lot of different things.
[00:13:11] Speaker B: Yeah, I don't agree.
[00:13:13] Speaker A: Are you more excited or less excited now about ads?
I'm putting you. I'm putting you down. I'm putting you down as less excited.
[00:13:20] Speaker B: Less excited.
I think, I think they'll figure their stuff out. The two platforms that I'm watching is OpenAI ads and Reddit ads. Both of them hasn't their shit together. Not yet. I think maybe in like 12 months.
[00:13:35] Speaker A: There's a lot of features that need to get built out. I think people don't understand advertising, which is fine. Like, why would you. It's not the most exciting thing, but there's a lot of tools and technology. You need to make these things run really well. Actually.
[00:13:46] Speaker B: Yeah. Yeah.
[00:13:47] Speaker A: Actually, this is one of my biggest jokes is that every. This is. It's less. Less the case now than it was. But I used to always laugh that, like, because I came from the agency, I was like, look, dude, you guys in Silicon Valley, like, I've never seen an industry where you guys are more excited to monetize through ads and have just a complete. Not only lack of understanding, but like a passionate, like, hate. I was like, the contradiction here is like, right, you got to monetize ads, but, like, not only do you not understand them, but you actively, like, block them and, like, don't really want to learn about them. I always thought that was like a really interesting contradiction that, like, they all want to monetize through it and they all don't know anything about it at all anyway.
[00:14:26] Speaker B: Yeah, but that's, that's. That's where the opportunity is, right? Like, Reddit is exactly same thing. Most Redditors hate ads, but they're. That's the only way they can make money is by converting users into.
[00:14:38] Speaker A: I think that's actually quite hilarious. I think Reddit is even funnier than the Silica. The Silica Valley ones. They're like nerds. They don't want to look at ads. Great. And that they don't really care about. I understand that, but like, the Reddit one's more. Is more ironic. Like, that platform is like, full of
[00:14:51] Speaker B: people who are, like, passionate about not being marketed to.
[00:14:55] Speaker A: I mean, dude, how are they going to ever get out of their mom's basement anyway? Right? Like, they're. They just sit on there and hate on everything. I've stopped posting on Reddit recently. I just have had enough of that place. Every once in a while, find someone who's like, post something funny. You saw my posts where it's hilarious. Yeah, I did the post of the guy. The name of the guy dropping the hand grenade, and I was like, why does Reddit hate marketing? It had like. It had like 5,000 views, no upvotes. I don't know how you do that. I felt I was quite impressed on myself. Right.
[00:15:22] Speaker B: People are silently hating all your posts. They're still looking at it.
[00:15:26] Speaker A: They all read it, though. 5,000 views. Not a single. Not one upvote. Must be a record. Okay, okay, let's do the hormozy thing.
[00:15:36] Speaker B: Okay. This is. Yeah, perfect segue, kind of. So we talked about the ads, platform, and then Hormozy. Did you watch the video?
[00:15:42] Speaker A: I did.
[00:15:43] Speaker B: Okay, do you want to summarize what the video he was saying?
[00:15:46] Speaker A: Okay, so I watched a bunch of hormozy videos because at first I saw this post, he showed me this video, and I don't even know why everyone's tweeting. The video is kind of like. It's kind of half edited. He's given some guy advice about random clip.
[00:15:58] Speaker B: Exactly.
[00:15:58] Speaker A: Yeah, yeah, yeah. So, okay. The guy asked him directly, how do I go from like 150,000 to 15 million? Some astronomical number that he wanted to scale his business. Which like, which is how I like, this is what I don't like about her mosey. If someone asked me that, I'd be like, look, let's be realistic. Like, that's a really big number. Why do we set increments? Like, figure out like what it really. Cause like if you, whatever your product is, you need to operationalize stuff in a way to serve people at 150 million, very different than at 15. But mostly like, I think it's partly why he's popular. His confidence level is just through the roof. He's like, well, he just gives him an answer. He goes, how many ads do you make? And the guy's like, we make five ads or something. And then he's like, you need to do like 250 ads a week. Yeah, you're laughing already. This guy's like a ridiculous answer.
[00:16:52] Speaker B: He's like, he's like a gym bro telling you how to run a business. How many sets did you do? 5. No, you need to do a hundred, dude.
[00:17:01] Speaker A: That's exactly what it's like.
[00:17:02] Speaker B: He's a gym bro. Just like, just do more sets.
[00:17:05] Speaker A: Oh my God.
[00:17:05] Speaker B: Two more rounds and do more sets.
[00:17:07] Speaker A: Well, he also starts with like we do 250 ads a day for his own business. Which like I also was had a couple. Like what does he define as an ad? Is an ad like a permutation that he's testing to this.
[00:17:24] Speaker B: Holy crap. Nice.
[00:17:25] Speaker A: Dude, I spent a lot of time
[00:17:26] Speaker B: you dug into this.
[00:17:28] Speaker A: What is it? Like what is an ad? According to him, 250 a day. So there's actually. Okay, so I pulled these up because there are people in the comments asking questions already.
No, not just asking, giving the correct answer. Like you can mathematic. Like this is where like I'm not a tech data scientist or machine learning expert, but there are optimal numbers of ads that you should create. Like the whole thing is actually like a math problem. There's. It's not just like, it's not just like a random answer. Like 250, bro. Like, no, you can actually calculate based on all the different parameters, right? How much spend do you have? How much inventory do you have? What is your click through rate? I think that's really all you need actually. CTR the volume, the spend, all the frequency. Like there's a few other metrics and you could, you could figure out the optimal amount of Creative. Which would vary depending on the volume of ads that you run. So the more ads you, the more. Yeah, the more ad space that you buy. That's the correct way to define it. The more like the inventory that you're buying, the more ads you'll need because the ads run out. But you can use data to figure out like exactly what that is, like how quickly they burn out. And it varies depending on a number of factors. Right. It's not the same for everybody. So like, there's a, there's totally like, there's like an actual, like mathematically correct answer to this. It's not 200 int a day. So this guy wrote this one. This one was really good because he just like kind of broke it down. Like, not, not on like my level of like, I am Coca Cola spending $10 million a month. Obviously they're going to do it through machine learning, right? This guy's like, look, if you're launching a hundred ads per week with $5,000 a day budget, you're only spending 50 ads. $50 per asset, right?
[00:19:13] Speaker B: Yes.
[00:19:13] Speaker A: He's like, he's like, you need to spend, you need at least 25 conversions per week to figure out if they're any good. Right? So what he gave the guy was like a metric of like how to actually do this. Not just like tossing out a ridiculous number. And it would like, Right. It goes on to say, like, depending on your cpa, like, if your cost to acquire a customer is a, is a hundred dollars. Right. He says you would need $350 per ad concept to get enough signals to determine if it was okay.
[00:19:43] Speaker B: Dude, at this point, I think, yeah, I think just harm, I think Hamorzi is just like a shit disturber. He just says things and get people to argue in his comments. And he's just like, you know that saying, like the, the best way to get the correct answer on the Internet is just by giving the wrong answer. Dude, I, I, that's basically what he's doing.
[00:20:01] Speaker A: I am burdened with being the guy who's like, wait a minute. Like that's a total garbage.
[00:20:06] Speaker B: Like, doesn't make sense.
[00:20:07] Speaker A: I don't even, I don't even. Like, there are times where I wish I was not this person. I'm the guy that sits in the back of the meeting and I'm like, not even paying attention. And someone goes, 250 y's and I would like go, wait a minute.
[00:20:17] Speaker B: There's like, yeah, it's exactly the same thing.
Exactly. So basically you have to discount Everything that he says in his video and then just go through the comments and then that's where the real value is.
[00:20:28] Speaker A: Yeah, they like, like, as I said, like there's like a mathematical way to calculate, to calculate this, right? It's very. So like to summarize it, like what you need is determine the amount of spend required to get enough clicks in order to figure out if that creative is going to beat the existing creative. So it's the last piece that none of this. There was someone else in there who identified which was my original answer, which is that what I have seen is that the performance varies greatly from like ad to ad, product to product, and that usually. Or what I would aim for is outliers. They're usually outliers that eat up like 60 to 70 to 80% of the budget. And the testing is only to try to find additional outliers. So the idea that they're all uniform and perform equally is, is also false or at least in my experience, not true. So the, the third, like, variable in all of this, this is where it starts to get really complicated and where I joke like, I'm not a machine learning, like expert. And if you have multiple products, all of a sudden the complexity of all this becomes crazy, right? And then learnings don't necessarily translate between the different products. So the best example of this is movie advertising. So I worked with a guy one time that was like, he ran campaigns, you know, for like big film company. Like the learnings, you can't share them. You can't be like the Wolverine movie. All of those, like data science and all that stuff we figured out for that doesn't work for like the rom com.
[00:21:52] Speaker B: It's a totally different. Yes, the timing's different, the audience is different, the appetite, what came.
[00:21:59] Speaker A: Everything is different. Like it's, there's no, there's no learning. And this is my favorite part. This was this guy I was chatting with some conference and then he goes to me, he goes, look, dude, I don't even know what to make of like, like attribution and like advertising, like data science anymore. He goes, the team comes back to me and says these are the top performing ads. And he says, I go, those are the worst performing movies.
[00:22:19] Speaker B: So like zero correlation.
[00:22:23] Speaker A: This is the best. Oh my God.
[00:22:25] Speaker B: Worst performing movie.
[00:22:26] Speaker A: Look like the bottom. So I watched a bunch of Home Rosie videos because I didn't want to like, like just come out of the gate and, and like, I don't want to be Kara Swisher and be like a professional Hater. Like, I really don't like that. Like, because. Yeah, yeah, you're exactly saying, like, all seriously hate on Elon all day long. I guess there's an audience for it, but I don't want to be that person. And so I watched some other videos that he made, and, like, he gave some good advice to people who seem to have a good experience for it. It could be fake, I don't know. But, like, there was a guy they had on. Had, like, some e commerce business. He's kind of struggling. He gave some, like, I thought, basic advice about how to improve what he's doing with Facebook ads. And. And the guy. Actually, they went back a year later, and the guy was doing much better. And it sounded like, actually the strategic advice. The strategic advice that Hormozi gave on his business, it was actually moving the needle.
[00:23:14] Speaker B: Yeah. Do you remember what it was?
[00:23:15] Speaker A: Yeah, yeah, yeah. The guy was selling kind of, like, home goods. So like, fences and different types of, like, home.
Like, fences and they. I don't. I don't even know what they're called. Like that with a railing you put on a stairway inside your home.
[00:23:30] Speaker B: Okay. Yeah, sure.
[00:23:31] Speaker A: Handle like a rail. I guess that's what I call it. And there's actually a lot of different kinds. So that kind of business. And he gave them all this, like, tactical stuff about Facebook was all very generic.
I don't think actually moved the deal. The guy didn't say. And then he did tell him, like, you should improve your custom order business, and custom orders is probably going to improve your business overall because you can charge more. And the guy came back like, dude, custom orders. We. He changed our process. We figured a way to do better job with customer orders. Custom orders doubled the order value and increased our profits, and now we have a much better business. So he gave the guy, like, really good advice. So I would never want to hate on anybody that is helping people. And it looks very legit and real. Particularly, like, independent business operators out there who are, like, just trying to, like, make a living, have a job. So that's my perspective on.
[00:24:15] Speaker B: Yeah, pretty good.
[00:24:15] Speaker A: I wouldn't trust him with a $1 billion budget. Yes. I don't know if he's the right
[00:24:21] Speaker B: guy for that, but you're doing 100 ads a day. No, we need to do a thousand.
[00:24:27] Speaker A: The video is great because, like, he's very confident. He was like, you just need to do, like, 250. 250 ads. That's it.
[00:24:33] Speaker B: That's.
Yeah.
[00:24:34] Speaker A: Take the. Move the needle.
[00:24:37] Speaker B: He's done some good stuff. You know, like his ghostwriter did a really good job on his books. That's. That's all I'll say. I like this books.
[00:24:44] Speaker A: Okay. I'm glad I bored everybody to death with like, advertising algorithms and machine learning. Okay.
[00:24:51] Speaker B: All right, all right, let's. Let's jump to AI Andreessen, truth seeking prompt.
Next one.
[00:24:58] Speaker A: Did you run this? Did you try this prompt?
[00:25:00] Speaker B: I did. So, like, there is actually a good snippets or just.
[00:25:03] Speaker A: Oh, wait, wait, explain. Explain it for people who don't. Who didn't follow Andreessen religiously. Like, I like.
[00:25:09] Speaker B: Yeah. Yep. Okay. So Andreessen came out with this tweet, got, I don't know, 3 million views, something ridiculous. Basically, he broke down his kind of universal prompt to fine tune his LLMs to do a better job. So it kind of broke into two parts. The first part is kind of you're experts, right? You're. You're super smart. Please kind of always frame things in a way that, that highlights your expertise. The second half, which is kind of like the fine tuning part, which is really cool, is he kind of adjusted the model to say, never praise my questions.
Don't ever kind of, don't ever kind of capitulate unless I provide new evidence. And then don't anchor on numbers that provide. Be. Be kind of like always push back on the numbers and then always use accuracy as your success metric, not my approval. So that's really cool. Basically what people are saying is that, yes, you have to kind of like fine tune this model to always argue with you. The top comment, which actually I found hilarious is said that basically he's creating a digital version of a girlfriend or a wife, which is basically saying that always disagree with me.
[00:26:22] Speaker A: That's what the prompt does. My God, that's funny.
[00:26:25] Speaker B: Like, always, always disagree with everything that I say. Just do that. And always be the expert and always be right. That's basically what the prompt is.
[00:26:34] Speaker A: Yeah, that's funny.
[00:26:35] Speaker B: So I thought that was hilarious. But there are of course some serious, like, feedbacks, which is like, never hallucinate. He used the words. But that's kind of like ridiculous, right? It's kind of like it's like a magician telling you to not think of a card. Of course you're going to think of the card. You can't tell a LLM to not hallucinate because that's like, it's. It's baked into the. Into the existence of the LLM. So, like, there are some good arguments in the Comments in there. I thought it was really, really interesting that this, this guy, this vc, super smart, kind of like the richest, always the richest guy, smartest guy in the room, kind of building this AI agent to argue with him without realizing the fact that he's kind of just, he's just kind of like training it to be what everybody else is doing around him. Right? Like basically what I'm trying to say is like, like he's going to basically one shot himself by creating this AI that always argues with them without realizing the fact that like he's just jumping against two extremes. One is the default is always going to agree with them and the other end is the AI is going to always disagree with them. You need somewhere in the middle ground, right? I don't think anyone has like really figured out what that looks like.
[00:27:52] Speaker A: Stop being so Canadian about this middle ground. Come on.
[00:27:55] Speaker B: But that's how, dude, that's how like AI psychosis happen, right? You either go one end extreme where you have AI models that always agrees with you or you try to like overcompensate and create a thing that always disagrees with you. Like, like a girlfriend or.
[00:28:09] Speaker A: I have. I can totally see how people fall into AI psychosis.
[00:28:14] Speaker B: Exactly.
[00:28:14] Speaker A: Yeah. There's times like I spend a lot of time, it becomes entertaining or something. I just keep like prompting it and like ask questions because it just goes on forever. It's like a never. It's like Dungeons and Dragons, like this never ending game.
[00:28:24] Speaker B: It just goes on forever. And so I can see you could
[00:28:27] Speaker A: get lost in that. I could totally see that.
[00:28:29] Speaker B: You can absolutely get lost in it. Right. The worst case scenario, of course is when it just always says, you're right, you're right, you're right, you're. You're a genius. That's very obvious. But then like this guy is kind of overcompensated to the other extreme. And then like, huh, it makes you, let's say it, push back on everything. Right? But like it doesn't actually make you any smarter. It's like I could just write a program that just keeps telling me that I'm wrong. Like, how does that make me smarter? It just makes me get deeper into my own bullshit. Right? It's like analysis.
[00:29:03] Speaker A: I tell AI it's wrong a lot now. So I've, I've become very skeptical with AI and I get a lot of answers from it that I don't believe
[00:29:11] Speaker B: or I don't think I don't believe anything.
[00:29:13] Speaker A: That's correct, actually. And I pushed back a Lot, which I don't think a lot of people do. So I don't know. I don't know whether she leaves us, dude, it's. I.
[00:29:21] Speaker B: So what I noticed is that AI, especially Claude, is really good at sounding smart. It comes back with an answer and then you read it on surface. You're like, wow, this is like the smartest thing I've ever read. And then, so what I've been trying to do is just tell it to dumb it down for me. My prompt is actually like, treat me as like a. A 10 year old. And like, I have no understanding of anything. And then if like the simple version can convince me that this is at least directionally correct, I completely discount it because, like, I have to fall back to my own understanding of things. That's like kind of super dangerous, right? Because AI will just keep spewing bullshit. It'll sound super smart. And to a layman person, I know it will. Will basically sound like it's correct. And then the tricky part is sometimes you ask it for multiple choices, which, like a lot of AI companies does. It's like, which version do you like? Do you like A or do you like B?
[00:30:17] Speaker A: Yeah, yeah, yeah.
[00:30:18] Speaker B: You realize that both versions are horseshit. It's like both versions are garbage. So like, you fall into like this weird cycle of like, oh, I have two choices. Maybe, like, if I get to pick one, I'm actually consciously making a decision. But you're not. You're like, the choice is completely.
[00:30:34] Speaker A: No, I don't know. I don't know what the answer is. I don't know.
[00:30:36] Speaker B: I don't know.
[00:30:37] Speaker A: Do we got other topics we gotta move on? I think.
[00:30:39] Speaker B: Okay, let's do.
[00:30:40] Speaker A: I think. Yeah, yeah, yeah. Like, okay, so I just want to touch on this one really quickly about copyright and memes. And so a lot of people think that it got more protection than they do. And it's true that like there is fair use rules that you can use memes over, like organic as free expression. Right? Yeah, I think where people. It's pretty common. It's still a risk. It's still America. People can sue you for anything. So the copyright holder.
Yeah, yeah, yeah, yeah. And like it used to happen at the beginning of the Internet, but I haven't seen it. But I am surprised that sometimes, like Tom Cruise, somebody doesn't decide just to go sue everybody. He could, he could get them all blocked and everything. I think it would make him look bad. But he's got enough money and his. His likeness is everywhere, like on These meme sites, like so just like people own their likeness, they can do whatever they want. So there is a legal risk using these things in general. But what happened was someone used it in an ad. So the this is fine meme, you know the dog sitting in the fire and he's like this is fine.
[00:31:33] Speaker B: Oh yeah, that's right.
[00:31:34] Speaker A: Yeah. It got it to an ad that was used like an out of home ad. And so there are very specific laws about copyright when it comes to advertising and commercial use. Right. And so they sued an AI startup. Yeah. Because it wasn't licensed.
[00:31:47] Speaker B: But it's like it's everywhere though. It's like, it's like Slack icon. I have that in my Slack. I have it right.
[00:31:53] Speaker A: But all of those. Correct. So this, so this is the important, this is like my recommend. My very, very general non legal recommendation is don't use it in a paid context, meaning you paid for the space. I think social media is a gray area. Like do I own that? Is it commercial? And there are instances that I'll get into. But like when you buy an ad, like we buy a TV commercial, we buy a billboard, we put like a giant, you know, thing through the New York City subway. If you don't have those things licensed like that in my opinion is copyright infringement. And there are startups out there that will help you license memes. So like I know someone who start, who started one of these things and they actually do all the licensing for like a Nike so they can get the rules for get the rights to all the stuff when they put on like Instagram, whatever. So there's a whole industry around this and like you should use it if you're going to put them in an ad. But like what you're talking about, like there's a gray area, right. Like Slack icons or whatever, typically it's okay, but like you gotta be careful. It's super commercial. You put on your landing page. It's like 50 bucks for whatever. And it's got like Tom Cruise says it's amazing.
Like, like he could probably don't do that. Yeah but that was interesting that it came up as a lawsuit because I've been big into memes and been pushing memes and use memes all the time. Right. And that's why I was like no one ever actually brings us up anymore. Like what is the legality of all this stuff? And so that's why, that's why I searched the other day. I was like what is, like what is it? And then so if it isn't like an ad. Like there are issues with it and. There are, there are.
[00:33:18] Speaker B: Yeah, that's. That's actually a fair point. I posted a meme today with the couple in the, in the nightclub where the guy's like whispering to the. Or shouting out the women. So like, I'm sure there are actors or they're paid models for the photo, but is there some like, legal concerns if I turn this into a. I don't want to.
[00:33:37] Speaker A: I actually, I actually know the whole history of this and I don't want to bore everyone with it. So like.
[00:33:42] Speaker B: So okay, check with your lawyers, check with your copy lawyer.
[00:33:46] Speaker A: There's actually a really complicated story to the whole, the whole thing because like it's the. The crux of it is that like it's changed what's. What's legal and what's not anymore than it was like when I was younger. I think that one. I think it's totally boring. Let's just move on. Let's talk about the AI more interesting job. Yeah, AI jobs apocalypse.
[00:34:03] Speaker B: Non existent job apocalypse.
[00:34:05] Speaker A: I mean a16z wrote this really in depth piece. I read the whole thing last night
[00:34:11] Speaker B: and they said, break it down for us.
[00:34:13] Speaker A: Yeah, they said basically what I would say, which is that yes, AI will eliminate some rules. And we're seeing it right now already.
[00:34:23] Speaker B: Yeah.
[00:34:24] Speaker A: I'll give it to very good examples. I know someone that ran like a Chinese language translation and like design firm in Vancouver and they're out of business. Super clear example, language translation.
I imagine they did a lot of things for like local businesses and stuff. And so like anybody could just go in there and just translate Chinese to English pretty easily. And it's not. They have no business. Right. So there are some really good examples where like it is just going to end and that's okay. Like that's what happens with these technological revolutions. But all these new roles are emerging and they go back and talk about all of these transitions and how like eventually if there's disruption, the economy evolves. Right. And I think like the premise that I think people miss and the fallacy that this document is like trying to, let's say debunk is the fixed pie concept of work. So yeah, the like, let's say the anthropic Dario worldview is that there's only this much work that needs to get done. And so if AI like does half of it, it means that all those jobs go away. But the reality is, or the like theory is that that's not true, that the economy evolves because people's desires are not fixed. And I think this part is really important and there are some economists who talk about this that my opinion, human desire is infinite. And that is why there will always be jobs. And if we want to have a robust economy and we want to have people that get employed and have jobs, as long as we have it set up so that.
[00:35:56] Speaker B: Yeah, correct.
[00:35:57] Speaker A: It's relatively easy to like hire. Like, I'm against all these like super onerous regulations like they have in Europe because it makes the labor market, in my opinion, less dynamic and harder for people to change jobs, learn things, new things. That's why I'm generally against a lot of like licensing and a lot of things that people think are good, but they really restrict the labor pool and create dynamics where like you spend a lot of effort, energy and money getting this license and now this job doesn't matter anymore. And it's going to make it all the more difficult for you to.
[00:36:26] Speaker B: Certifications, licenses.
[00:36:27] Speaker A: Right. And I'll be super clear if anyone's like going to go out there and try to troll me on this. Like, I'm not saying you shouldn't be licensed if you want to be a goddamn doctor. Yes, there are certain roles that need lots of license, but maybe not really correct, maybe not.
[00:36:41] Speaker B: Real estate agent.
[00:36:42] Speaker A: I would agree that real estate is way too many regulations. I don't think it's appropriate or necessary to have. Maybe there's some kind of certification. The ultimate example is like you need all kinds of like regulations for all kinds of menial jobs. Like, there's all kinds of certifications that are required. But like, we just like throw people in the military and send off, like shoot people. So I don't, I'm like, I think there's a lot of regulation out there around this stuff that like perhaps isn't.
Anyway, it makes the job market less flexible. That's the point. Right. And that if people want to do different things and like will desire different things and the labor market will evolve so good.
[00:37:18] Speaker B: Does the article kind of break down where some of the new opportunities are, let's say for the people that may or may not have been looking for jobs?
[00:37:27] Speaker A: It doesn't, it doesn't really go into, well, it doesn't go into like new, trying to look here, new jobs. It stays pretty high level.
But it talks about like, you know, when like a third of the labor market, United States was, was, were farmers. And then we invented tractors.
Right? We invented tractors. We've entered all these, like, they're all kinds of machinery, that automated farming. Now farm labor is like 1% or 2% of the. Of the labor market. So obviously we evolved and all the. And there's new jobs in my. And this is an example that's it's brought up time and time again. Like it's a super clear example where all these jobs went away. Obviously I wasn't around back then, but there was pain. There are people like couldn't get employment anymore, unlike being a farmhand. And I don't know what they did. It might have been a difficult time for them, but some people changed, some people evolved. Some kids went and learned different things and got into other areas. Maybe they became mechanics and they started bicycle companies and they built airplanes and all kinds of stuff happened in the industrial revolution that took people off farms and put them in other roles. Right. So the economy did.
Did evolve. So we have a clear example of this. There's lots of like, arguments over the nuances about, like, what enabled that. And I think those are actually really important. And they're complicated, like schooling and education. And there's a lot of things that are at play. But I still think there's a lot of jobs out there that don't require a lot of those things. And the economy just evolves. Right. Like when I was younger, like, whoever thought like yoga instruction was like a job, there's all kinds of people that are employed as like personal trainers and in the whole fitness industry, they become hermosi and then they become sales gurus. That guy is a talented salesperson.
[00:39:07] Speaker B: Yeah. Social media influencers. None of that would have been possible.
[00:39:11] Speaker A: Yeah. And people like to hate on it. But like, being a social media influencer is cool. I was watching some podcasts my wife last night. We had never what we stumbled on. It was really entertaining.
[00:39:18] Speaker B: It's super cool. Yeah, it was great.
[00:39:20] Speaker A: I think it's fantastic. Yeah, I do think people will be influencers and I think there'll be more of those in the future. And I think all other kinds of jobs will emerge.
[00:39:28] Speaker B: Yeah, we. We've kind of like said this before. I think it is just coming off of the ZURP era. Coronavirus over hiring. A lot of companies are just, you know, letting people go. They're resetting.
[00:39:38] Speaker A: Oh yeah, dude. So we've. We. So okay, so there's this big debate, like what's happening in tech right now. So there's charts that show like, software engineering is actually hot job area. Right There are hiring. It's like the area is evolving and the skills required to be successful Are changing.
[00:39:54] Speaker B: Yeah, correct, Correct, Correct. It's. So would. Okay, question to you, right? Would you bucket something like a market marketing engineer under software?
[00:40:04] Speaker A: I would say it's a tech job. How's that?
[00:40:06] Speaker B: It would be a tech job, right?
[00:40:07] Speaker A: I would say so if the job
[00:40:09] Speaker B: description says that you need to know how to code in Python, but your core job is like as a growth marketer or marketing engineer, maybe in some metrics that's considered a software developer these days, but fundamentally you're doing marketing work. So, like, I think, like, things are just bleeding together, right? Like some developers are. Has to have design or product management experience. I think, like a lot of these stats are just measuring the wrong things. Overall, I don't think AI is removing jobs. I think AI is just massively displacing jobs. So if you're kind of like stuck in the 2015 definition of whatever your job is, I think you're behind.
[00:40:49] Speaker A: Oh, so there was some, there was some cool, it was a cool charts in here. Like, they, they. Okay, they talked about travel agencies actually know this industry really well that like, other people commented on Twitter this morning saying that like the, in, like in a. Let's say automation and the Internet and what, what are called online travel agencies. So Expedia, those types of things, like, eliminated a lot of travel agencies, but the, the actual number travel agents are only down 50% since like 1995 or something. I think it's not a lot, is what I was saying that, like, still 50%, still 50% survived. Like, you go on and book all your travel, there's still a market for it. Right?
[00:41:27] Speaker B: Right.
How about, how about this, let's move on to the Coinbase layoff and then we can, we can talk a little bit about that tweet, come back to that. Okay, yeah, because that's actually a really good point and I want to use that as a segue to the.
[00:41:40] Speaker A: All right, you get Coinbase one. This one's a long one.
[00:41:42] Speaker B: So this one was a long one. The guy came out, the Coinbase CEO kind of came out with a tweet just saying that he's about to let go 14% of the Coinbase. And he kind of rolled this whole entire memo on a couple of reasons why he's laying it off. He's restructuring the company around the concept of like a single person team product team, where one person might be simultaneously the developer, the designer and the product manager. And of course, what everybody says in the comments are one, super negative. Two, they think that, like, what we just Said of course they overhiered and the market sucks, especially in crypto. Crypto is not doing well right now. And then he kind of like coined a few things that says they're going to build a intelligence with humans around the edge aligning it. They're going to introduce a player, coach managers with hypothetically 15 plus direct reports, one person teams. And then he also says something called non technical teams that are shipping production code. Of course the Internet X went, went a little bit crazy on everything over here. I mean I think that's kind of like the general dude.
[00:42:55] Speaker A: It's, it's like for me, for me it's all like just deja vu. Like it's amazing to see how things work in cycles. So basically what I feel like it's happening is like the dot com. So I was really young and I got a dot com. I was like hacking stuff together. I was like writing really bad Pearl scripts and stuff and kind of like the way things are working. It was, there was so much change happening and there was so many new tools that there was like all these kind of like hybrid roles. And then eventually what happened was like that all went away. Eventually.
Yes. Which is the same thing will happen. Like right now there's all this enthusiasm and excitement for all these new stuff and like some people know how to utilize it and like there's all these like, there's all these like hybrid roles that are emerging. So it's just a, it's a transition period to a place where it'll all become more specialized and.
[00:43:42] Speaker B: Exactly. Because we got a, we got a new tool, you got a new hammer and people are hammering everything with it.
[00:43:47] Speaker A: And so there's also, also times in Silicon Valley in particular where people try to reinvent the office. There was a lot of talk about this. Yeah. Like so that's, I see some of that too. Right. With like one person teams and non technical people shipping code. Like there's all these like experimentation right now I think is the right term for what's happening at like tech firms. They're re the reinventing how teams work which I actually think is good because I don't think anyone knows how to build an at scale product organization that utilize. That's AI first.
[00:44:19] Speaker B: Yeah. Okay. 100% agree on that. And then the thing that I want to tie to what you said beforehand is like so a lot of the talks, if you read through the comments are focused on the people who got laid off. Right. Like we're always looking at the people who got laid off trying to figure out where are they going to go. Like, are they being completely made redundant or are they going to be displaced into some new role? What I actually want to know is for the people who left behind, so the people who didn't get left laid off, how is, are their jobs changing?
Because like say you're laying off like 20, 30% of the workforce, right? Of course your role is now being compressed. How long is that going to last?
Like can you realistically just tell, let's say a software developer, hey, you're, you're, you're also now a pm. You're going to have to like become a one person team who's going to go from concept all the way to development, all the way to testing and production.
[00:45:16] Speaker A: All right, you want, you know what I think, dude? I think these companies employ like so many people that people were not working that hard and there's like whole teams that are being eliminated that, that didn't do.
[00:45:27] Speaker B: Okay.
[00:45:28] Speaker A: I don't want to put anyone down but like it wasn't core to the operations or success of the company. That would be the right way to frame it.
[00:45:36] Speaker B: So. Okay, so that's interesting. Right? So they, that basically means that they laid off, let's say the most, the bottom of the, I mean how would
[00:45:45] Speaker A: you, how else would you do it? Right? You're like, oh, I have a team that, that creates internal GTM tools. Remember we talked about the guy? Like I, I, my job is to like track all of the internal vibe coded tools and then put them in a list. Like do they really need to do that? Like it's not, it's not, it's not a hundred percent necessary to their survival. Right. Okay, here's another area where like if you're, if there's just less hiring. A lot of these companies have like way too many people in the recruiting area.
[00:46:13] Speaker B: Fine. So that's, that's kind of like, that's kind of fine. Right, well, let's kind of like project this 12 month into the future because this is really interesting. Okay, let's say they laid off, let's say 15% today. That's. You're going to feel that little bit regardless, right? Because of course these things are going to.
[00:46:30] Speaker A: Yeah, whatever.
My buddy got laid off. I get coffee with. I get it. It's not fun. I don't want anyone to get laid off.
[00:46:37] Speaker B: It's not.
[00:46:38] Speaker A: Fine, I understand.
[00:46:38] Speaker B: And then like of course this 15% people's, whatever they were doing gets rolled up into the intake or it stops getting done or Stop getting done. Something will happen maybe in 12 months. Like whoever gets the short end of the stick that gets. Have to. Have to do more work, let's say voluntarily leaves. Right. Buckles under the pressure. My question is like, are the people who leaves afterwards, are they the bottom people or are they the top people? Like, what if it's the senior people who's now doing a lot? I don't know.
[00:47:11] Speaker A: It's a mix, It's a mix, it's a mix.
[00:47:13] Speaker B: But this is dangerous, right? Like, we don't know what's going to happen.
[00:47:15] Speaker A: Dangerous from who's. What do you mean, dangerous for. What's the danger?
[00:47:18] Speaker B: Well, the danger is like, from the company perspective, the employees perspective, the shareholders perspective, I would say, like, CoinB is a public company. This is one of the.
So, like, it makes it interesting because this is the first time that the CEO came out.
[00:47:34] Speaker A: The stock's up, right? The stock's up, isn't it? I don't know. I mean, we have the stock they reported today, right?
[00:47:39] Speaker B: No, it's down.
[00:47:40] Speaker A: It's down, right.
[00:47:41] Speaker B: It's flat over the last five days, up 10% in the last month. But the entire market is up since the last month.
[00:47:48] Speaker A: So I think, I think, I think it's. I think it's. I think it's. I think it's company to company. I don't think there's any like, blanket way to like, determine. Yeah. I could tell you what I, what I think about Coinbase. I sold the stock this morning.
[00:47:58] Speaker B: There you go. There you go.
[00:48:00] Speaker A: Because I like, the announcement came out like Tuesday, right?
[00:48:04] Speaker B: Yeah.
[00:48:05] Speaker A: Or the tweet or whatever he posted. And like, earnings, I think are today.
And I was like, why would you say this before the earnings? Like, that is, in my opinion, not good. And so I was like, I hop on Coinbase, I didn't own a lot of it, I made a profit and I'm like, I'm out. Like, I'm not going to stick around and see what's going to happen on, on earnings. Like if you, if you drop the, like layoffs before earnings, it would seem to indicate that the earnings were not going to be getting people excited on Wall Street. And then the thing is down there, I think earnings might be after the call. So people are not expecting. I mean, bitcoin is down still like 25%.
[00:48:41] Speaker B: Yeah. The entire entire crypto market is down. Yeah. And then like all time, bitcoin is down.
[00:48:47] Speaker A: I think your, your question is valid. It's case to case. Like they, they have Too many people, they need to like make shareholders happy. There's probably divisions and teams that are doing things that they don't have to do that they can like rethink there's products they can probably sunset. So there's tons of things that they can do that like. So I think to answer your core question, if it's done correctly, they should be able to not disrupt the core function of the company that earns money. Right? That would be like any CEOs like Mandate, like we have to have layoffs for whatever reason, we need to like not disrupt the company. But now you're saying like once layoffs happen, it's like like really negative. Right? That's why like the recommendations always just like cut deep once, just fire a lot of people. So that way everyone who survives is like committed and excited and you don't have to like keep cutting. It's the cuts that happen multiple times that create like a lot of animosity and negativity and people just kind of give up in their job. They will always get caught in the next one. So like whatever. And that's just like not a great place to be. But dude, some of these companies employ so many people, I don't know if it even matters.
[00:49:53] Speaker B: I don't know either. Let's, let's move on. Okay, we're almost out of time, so. Okay, what's next? SpaceX and.
Oh, this was.
[00:50:01] Speaker A: Dude, this was crazy. So like. And I sent you some additional information on this thing. Oh, it's noon already, so. So let's keep going. This one's really interesting. So Claude and SpaceX made a deal for Claude to assume a whole bunch of GPUs. And yeah, Colossus 1 capacity that X had built out. And someone wrote like a really in depth description of what is going on here. I didn't realize the extent of this. So Claude is assuming like essentially whatever extra capacity had, which was Colossus 1. And I guess X is building Colossus 2 somewhere. So they took the entire thing. It's like a $900 million deal. And now like it's gone from on X's balance sheet from like they had this ability where they'd spend all this money as hardware to now they have like a revenue stream where now they can say, oh, all that hardware, guess what? That's not a, that's not a liability anymore. That is like a core asset that is earning us cash. It's wild. Like the way that this worked, I was like, this is fascinating. So let me Just say this one piece. Let me say one thing, the most important thing because I was debating with someone about this specific thing on Twitter this morning. I think it proves that the AI buildout in the hardware side is now based on this, this event. Not a waste. Like there's a lot of people debating like, is all this hardware valuable? Like are we going to utilize it all? There's going to be too much dark capacity, all that stuff. According to this, not yet. There's so much demand and we literally see competitors like selling each other shovels they have.
[00:51:40] Speaker B: Correct.
[00:51:41] Speaker A: Because there's so much demand for it.
So I think that the hardware aspect of all this, at least at this stage is is there's more demand than there is supply.
[00:51:53] Speaker B: Correct. And then it seems like the older generations are not as big of a capex head as people expected. Right. Like, so a lot of the foundational model companies, one of their, one of the scariest thing is you invest billions into buying hardware and it's made redundant in let's say like less than 12 months. But it seems like this is not the case. And the fact that Elon, I think he ridiculously or famously spun up process one in like 180 days, something ridiculous, like less than six months of time. So that is crazy amount of infrastructure build and it's crazy, dude.
[00:52:31] Speaker A: The other number in here was like, I can't believe this is real. Like Claude's revenue, like actual like yeah, their internal plan. The internal plan was 10x.
[00:52:41] Speaker B: 10x?
[00:52:42] Speaker A: Yeah, the internal plan was 10x. But the actuals are like 80 something ridiculous. Yeah, something right.
[00:52:48] Speaker B: Absolutely ridiculous. So they grew really, really fast and of course they're running out of capacity and they're looking everywhere for you. Right? I mean like the fact that like it's crazy.
[00:52:58] Speaker A: It's crazy. So for me it's like because there's all these people like oh, they're building on stuff is like the demand is there and oh, and you've got cursor in the mix. Look like the AI coding use case is a hundred percent proven business in my opinion. Like there is no doubt, like me making memes with AI, like I don't know if that's going to be a business, but AI coding, cursor, anthropic, chatgpt codex, like these things prove without a doubt, like there is a big market for this stuff and all this hardware is going to be utilized. It's, it's a real business. So like that was my takeaway from all of this. I think that that's the market implication? Because I think people have been waiting for where's the sign? Like, is this all going to play out? Like they're spending all this money? Those are good questions to ask, and I think we know the answer at this stage. Like the, the build out is real and it's valuable.
[00:53:51] Speaker B: I 100% agree. Shall we wrap it up?
[00:53:54] Speaker A: Okay. Okay. What do you want to say about the trial? This is the trial of the century between Elon Musk and Sam Altman over the conversion of OpenAI to A for profit company.
[00:54:06] Speaker B: I think we can do a whole show on this.
[00:54:08] Speaker A: Oh, my God.
[00:54:10] Speaker B: The spicy tweets.
[00:54:11] Speaker A: You saw the tweets or the text?
[00:54:13] Speaker B: Spicy tweets. The, the Brockman's journal back and forth on the day that Sam Albin got laid off or fired as the CEO. There's lots of stuff to unpack there. Um, I, I, I would love to wait until the results.
[00:54:29] Speaker A: I think the damage is already done. Hold on. The verdict drops May 21st.
[00:54:35] Speaker B: May 21st. Yeah.
[00:54:36] Speaker A: So coming up.
[00:54:37] Speaker B: Couple weeks. Couple weeks.
[00:54:39] Speaker A: Oh, my God.
[00:54:40] Speaker B: I think the damage is already done.
[00:54:42] Speaker A: But the way damage for who?
[00:54:43] Speaker B: Who's.
[00:54:44] Speaker A: Damn. Who's. Dam. Who, who do you think is damaged? Who do you think is damaged more?
[00:54:47] Speaker B: Maybe Sam Albin. And then gre.
Kind of caught in the crossfire between the two titans. He's like, by all means. He's just like this guy that wants to.
[00:54:57] Speaker A: You think they're, you think they're gonna lose the lawsuit? Is that we think.
[00:55:00] Speaker B: I don't see how they're not gonna lose the lawsuit.
[00:55:02] Speaker A: I like the confidence. It's a hard one for me.
[00:55:04] Speaker B: I don't see how they cannot lose the confidence.
[00:55:07] Speaker A: I think people perhaps dislike Sam more than they dislike Elon is what I think. So I do think you're perhaps correct.
That'll be my way of interpreting all of this.
[00:55:15] Speaker B: It's just, yeah, it just like the way that he kind of took over the company, kind of kept Eli in the dark. Regardless if, like, regardless if the mission realignment misalignment of that.
[00:55:27] Speaker A: The text, the text messages are creepy too.
[00:55:29] Speaker B: They're like, creepy.
[00:55:31] Speaker A: Yeah. Very uncomfortable to read. They're like, oh, the company, the board doesn't want you. And he's like, oh, that's like.
[00:55:36] Speaker B: It was just like, oh, something was going on. Yeah, I don't know. I'm gonna, I'm gonna reserve my comments until.
[00:55:43] Speaker A: All right, the 21st. We'll do a special edition.
Yeah, I think we, I think we should.
[00:55:49] Speaker B: All right.
[00:55:49] Speaker A: Are we going to wrap for today.
[00:55:50] Speaker B: Let's wrap for today.
[00:55:51] Speaker A: That was fun.
[00:55:52] Speaker B: That was great.
[00:55:53] Speaker A: All right, Paul, next week.
