Big Technology Podcast - OpenAI’s Ghibli Moment, CoreWeave's IPO Letdown, End of Silicon Valley’s Monopoly?

Episode Date: March 28, 2025

Brian McCullough is the host of Techmeme Ride Home. He's back for our weekly discussion of the latest tech news. We cover: 1) Why everyone's using ChatGPT to make Ghibli art 2) What is Ghibli 3) OpenAI's product dominance stands out 4) The Studio Ghibli copyright question 5) The AI servers are at capacity 6) The AI datacenters are still probably built out too early 7) What's CoreWeave? 8) CoreWeave's IPO disappoints 9) OpenAI eyes $40 billion fundraise 10) Is Silicon Valley about to lose its monopoly on tech? --- Enjoying Big Technology Podcast? Please rate us five stars ⭐⭐⭐⭐⭐ in your podcast app of choice. For weekly updates on the show, sign up for the pod newsletter on LinkedIn: https://www.linkedin.com/newsletters/6901970121829801984/ Want a discount for Big Technology on Substack? Here’s 40% off for the first year: https://tinyurl.com/bigtechnology Questions? Feedback? Write to: [e-mail]

Open AI's new chat GPT update leaves the internet jiblified and asserts its dominance once again. The Corrieve IPO leaves some concerned that the AI run is ending and is Silicon Valley's monopoly coming to a close. We'll cover it all on a Big Technology Podcast Friday edition right after this. Welcome to Big Technology Podcast Friday edition where we break down the news in our traditional cool-headed and nuanced format. Ron John Roy is out today. He'll be back next week. We'll actually be on on Wednesday. And joining us this week, we have a special guest, Brian McCullough, the host of the tech meme ride home podcast is here for a great episode where we're going to cover everything from the new Ghibli era of OpenAI. Hopefully we'll cover the new Google Gemini model. And we'll also touch on

this CoreWeave IPO, which we're waiting for live. Great to see you again, Brian. And welcome to the show. Alex, thanks as ever for having me. It's great to have you. And you're coming at an auspicious week because we've definitely seen AI in the headlines, not for models, really for a consumer product. And that is the chat GPT update that did a bunch of things. We also know this week, there's better voice that's come out, but the real headline here, because it's what everybody has gravitated toward, is this new image generation within chat chip PT. This is from the New York Times. Open AI unveils new image generator for chat chipt. The story says chatbots were originally designed to chat, but they can generate images.

too. On Tuesday, opening I beefed up its chat chippy chat bot with new technology designed to generate images from detailed complex and unusual instructions. For instance, if you describe a four-panel comic strip, including the characters who appear in each panel and what they are saying to one another, the technology can instantly generate an elaborate cartoon. Actually, Brian, your tweets about this were some of the earliest that I saw of like the advanced technology that is embedded in this update, we went from really not being able to include text to being able to include text to create like pretty cool new images based off of original images. And of course, the cartoon stuff is really good. It's clearly just much more advanced and capable than previously. But then the

craziest thing happened and everybody just jiblified themselves. And by the way, what that is is a, I think it's a Japanese anime style that portrays the characters as pretty friendly. You are going to get comments about this because Miyazaki is beloved. He is basically the Walt Disney of Japan. My kids adore his movies. They are his, his cartoon style, once you have watched those movies, you'll know it instantaneously, which is part of this. This is the argument that is being made is that his style is so distinct and beloved that

that's what people are freaking out about, is that I can take a picture of my family, put it into the chat Ghibit, and I can get a studio Ghibli style portrait of my family. Describe what is the style? Describe the style. It's sort of anime, but I mean, to be honest with you, because I do find these movies fascinating. It's anime with a sort of dream quality that the aesthetics of it are just, you could take any frame of a Miyazaki movie and put in in an art gallery and be like, that is a masterpiece. And so, right, that's one of the things is Miyazaki is beloved, okay, and people are very protective of his work and his legacy. And then because it's so distinctive, like once you've seen the

movies, you will see a Studio Ghibli image. By the way, Studio Ghibli is the studio, like Walt Disney Studios was the umbrella under which Walt Disney operated, right? Okay. Once you have seen the Miyazaki slash studio Ghibli style, you will recognize it instantaneously. And so that's the other thing. It's not just that, oh, that kind of looks like a Wes Anderson movie or that kind of looks like a Picasso style or whatever. No, this is dead on, which suggests that they trained it possibly on millions of cells and frames of Miyazaki movies. And so then why do you think it took off the way it did? Is it just because that this image style, is so fun because let me just give it an explanation of how prevalent it's become. I'm just going to read right from TechCrunch.

It's only been a day since ChatchipT's new image went live and social media feeds are already flooded with AI-generated memes in the style of Studio Ghibli. In the last 24 hours, we've seen AI-generated images representing versions of Elon Musk, the Lord of the Rings, Lord of the Rings, President Donald Trump, Open AI CEO Sam Ullman, even seems to have made his new profile picture into a studio Ghibli style image. Let me ask you this. I mean, is it that these images are so beloved or is it that open AI sort of seeded this style and people and it's caught on contagiously from them?

Because they did make a comment that they are very intentional about the type of images they release at the very beginning. And it does seem that if this was a top-down choice, it was a very good one from that. Well, put a pin in that because I do want to come back to that. I did a segment on the show today that talks directly to that. But to answer your question, I think Sam changing his profile picture to an obvious Studio Ghibli style made everyone be like, oh, wait, that's what we can do with this. So for sure.

And again, I think that they are, this is not accidental. They are writing a line here of they know what can go viral. and they know it will get attention. And even the questions and the controversy about the legality of it and is this where it's going, like, I think that they know that that benefits them at this point. I would say that one of, you know, the debate would be, Miyazaki has come out. He has famous quotes where early on in like 2016, he was shown using AI to create animation.

and he was extremely disgusted by it. If you have the quote in front of you, jump in right now and read it. Yeah, he says, I am utterly disgusted. I would never wish to incorporate this technology into my work at all. I strongly feel that this is an insult to life itself. So this is the creator of the studio Jibli art, basically saying that the AI is an insult to life itself. And he's like an 80-something-year-old man.

We know that probably the last movie that just came out, The Boy in the Haren, might be his last one. So put all these things together. A beloved artist. When I say artists, people think of him on the level of one of the great visual artists of the last hundred years. Like, he's that highly regarded. Number two, an old man that we know is coming to the end of his creative output. And then so people feeling like, oh, look at how technology is now robbing and cheapening.

By the way, I'm not, I don't necessarily. feel this way about it, but the argument would be robbing and cheapening what is the work of a genuine genius, right? It is now becoming commoditized, memeified, and so people feel like that that's cheapening it. The other level of it is, is that if you're a fan of Miyazaki, like, I did put a picture of my family into this and Miyazaki did because that's cool. And maybe I would print that out and frame it. I think every open AI user did that. I spent a good chunk of my day yesterday just putting friends and family into this thing

and converting them into Jeebly style images. Well, so I said put a pin in the thing about how this is obviously a, this was a choice by open AIs. So Joanne Jang, I did this on the show today, has a blog post up where she specifically says in regards to this image stuff that, open AI is, quote, shifting from blanket refusals in sensitive areas to a more precise approach focused on preventing real world, real world harm. Essentially what she's saying is that, yes, we are getting a little more permissive when it comes to what we're going to allow users to do.

She says, AI lab employees should not be the arbiters of what people should and shouldn't be allowed to create. We're always humbled after launch discovering use cases we never imagined. Towards the end of the piece, she says, my colleague Jason Kwan, once passed on to me, ships are safest in the harbor. The safest model is the one that refuses everything, but that's not what ships or models are for. So, and I'm, uh, hopefully you can link to the piece or whatever, because she goes into greater depth about how they're not just saying, oh, we're opening the floodgates. It's a, it's a, we've, we've thought this through and we want to be more permissive, but we're also still worried about, obviously, the, the, the worst case

scenarios of, of new stuff that we put out. But we also think that if we don't put stuff out that we're, she says something like it's the graveyard of things that, invisible graveyards of possibilities that haven't been imagined, right? So they're saying we're trying to strike a new balance between preventing harm and also being permissive and allowing a thousand flowers to bloom with the release of new technologies. So yeah, so like maybe this is, I'm thinking through why this took off the way it did and here's a couple ideas. One is because the technology is good and they picked the right artist. Two is Open AI.

One, seven, they picked the right artist, seeded it through places like Sam Altman's profile, Altman's profile picture. Three is that they became more permissive. And one of the things important that we should note in this update is that you were able to start to put well-known public figures or basically anybody, and they wouldn't refuse, like, putting, transforming a person.

And that's why I read a couple examples. You saw Elon Musk. he saw Donald Trump, basically every famous historical scene, including ones that are somewhat disturbing to have seen to have been jiblified, have been, including 9-11 scenes. I want to point out, Mid-Journey has allowed you to do famous figures for a while now, because I've done it. But they're just not good at it. I mean, I've done images of Zuckerberg and Tim Cook within Mid-Journey, and it spit out

like two men that look like a blend of both CEOs. And I think, yeah, it is, again, going back to competency or opening. I was able to do this better than anybody else. And there's a technical reason. And there's a, I think it was the verge piece when it came out, that instead of, like, if you do mid-journey, mid-journey does the entire image coming together at once and then slowly, like, comes into focus. And if you do the chat GPT now, it's going line by line, drawing it, sort of like it used

to be in the dial-up modem days of trying to get a picture down the end. Yeah. So apparently for some technical reason, that's one of the advances of it being more accurate is that it's literally drawing it line by line. But yeah, so number one, they're allowing public figures. Number two, I did Jack Dorsey as if he was on the poster of a Wes Anderson movie. You can do, you know, 1960s Marvel comic style. You can do styles that people would recognize.

And at least Open AI has not allowed that before. And so they are clearly taking some of the handbrakes off. Yeah. And I'm going to kind of push back on this, the ship in the harbor doesn't do much point from OpenAI. It's not that this is like what models do. And that's why they're like taking the permissiveness off. I think they do feel pressure from places like Deepseek from other open source AI companies. And they know that if they're slow, they're going to be left behind.

And this is sort of where like the safety mess. to me falls apart. Open AI, I think above safety, wants to win. And this is, again, their attempt to win. And it is interesting to me that they have done this. And we're at a moment where we're talking, every week we talk about how models have commoditized. And when models commoditize, what you really need is hits. You're in the hits business. And I think OpenAI knows this better than everyone else. They have chat GPT, which is a hit, 400 million users. Nobody else comes close and they're pushing forward. And by the way, it's no coincidence

that this all comes within chat chipped and not within Dolly. And every time they do a step forward in multimodality, we know that voice was big in terms of them being able to increase their user base. I mean, you look at Sam Altman's her tweet and it's basically an inflection point where Open AI goes from 100 million to 400 million users

once people realize that they can talk to these things. And this is them again putting the gas pedal on trying to win that product with yet another hit. And you're right in the sense that, especially in the last few months, as, you know, GPT-5 did not come around, which people were waiting for, waiting for, at least their thunder had been stolen by Deep Seek, by others, especially in the AI and tech communities, other model makers started to, people started to feel like Open AI was falling behind. But you're right. What Open AI does have is the normie brand recognition. And so this is a pretty smart move. I'm assuming that it's conscious and on purpose, and it has to be.

So this was a genius move to, again, sort of grab back the mainstream sort of branding of, yeah, we're still the cutting edge. And that's what people, especially people deep in the tech, have been saying for months now that Open AI is losing it and people are racing ahead of them. Yeah. And I think this is a really important point to stop because we also saw a release from Google. We also saw a DeepSeek release this week. No one's talking about that. And in fact, M.G. Seagler had a great piece about this. I'm just going to pull it up. It's from Spyglass. He said Google and Microsoft two multi-trillion dollar companies also had AI updates to share. yesterday, seemingly no one cared because neither could make me look like a studio Ghibli character. Open AI is on a role despite all turmoil and turbulence they've gone through internally and externally. They just keep killing it with all these product releases to the point where they again don't even have to have a technically, they don't even have to technically roll out a new product that can just drop what they consider to be a feature update

and it spreads like wildfire. I'm just going to point something out here. So Google did have this release of Gemini 2.5. We may or may not talk about it because we're getting into it now. We may or may not talk about it later on. But I went through the document and I like looked through the naming conventions where this is Gemini 2.5 and there's flash thinking and there's and then I saw all my Twitter feed like people upset that Gemini wasn't getting recognition for also being able to jibble yourself. And I just like looked at this and I was like, you know, maybe the results are are at par. Or maybe open AI. a little better, but Google can still do this. And the branding and the simplicity of this really

does matter. And the fact that this is all within chat GPT, chat TPT is now a verb like Google or, you know, basically people call it chat now. That is just this huge compounding advantage that Open AI is going to have. And releases like this just help it push even further forward. Right, because they did just drop it into the free tier and with rate limits. And the $20, a month tier, which is what I'm on. I'm not paying the $200 a month pro thing. So you're right. Again, as opposed to creating a new brand for it, as opposed to, hey, this is additional dollars per month to do this. It's still great, but it's more money. They're just throwing it into the thing that people are already familiar with. I'll tell you the other thing about the, you know,

because was it Microsoft also had like there was like a deep learning model that came out this this week and you know I reported on it and stuff like that and I'm sure that if you're in the weeds of this like maybe advances are being made but again as a quasi normie I don't really know even if you read the headlines about like how Gemini 2.5 is better I don't really know what that means how it's better why except if I'm not using it to call to APIs and stuff like that. But I sure as hell know that, man, that blew my mind the pictures that I can create now. I don't want to spend too much time on this, but let's briefly hop into this copyright conversation and see, you know, we're not going to debate the legal side of it, but we're going to debate like, does it actually benefit? Studio Ghibli here. So you have two sides of this conversation play out, you know, in the main discourse over the past

couple days. The first is represented pretty well by Brian Merchant, who's blood in the machine substack, says, Open AI Studio Ghibli meme factory is an insult to art itself. And he quotes that Miyazaki quote that we just talked about previously. Miyazaki says, every morning, not in recent days, I see my friend who has a disability. It's so hard for him just to do a high five. His arm with stiff muscle can't reach out to my hand. Now thinking of him, I can't watch this stuff and find it interesting. Whoever creates this stuff, he's talking about AI, has no idea what pain is. And then he says, I am utterly disgusted.

I would never wish to incorporate this technology into my work at all. I strongly feel that this is an insult to life itself, which we quoted. Here's Brian's analysis. This issue here, the issue here should be obvious. The man on record with likely the strongest, with likely the strongest, and bluntest disavowal of using AI tools for art, is now the man who is notoriously painstakingly handcrafted art is being given. Italy automated by chat GPT users for what amounts to a promotional campaign for a tech company that's on the verge of being valued by $300 billion.

Just about everyone in the AI world knows Miyazaki is adamantly against AI, and they're doing these memes anyway or worse because they know he'd hated it. So that's one side of the argument. Now the other side of the argument is, look at all this. I didn't know what Ghibli was before this week. Now I know about it. And that's not uncommon. Here's some people making that argument that's a benefit for them.

It's from Shantanu Goel. I don't care at all whether people knew about Ghibli before today or not. I'm glad that more people know about it now than before. Here's another one on the Twitter user possibly result. Studio Ghibli made millions today through reaching massive new audiences for free. Wall Gardens aren't always best for business. Just curious, Brian, which side of this debate do you fall on? I'm going to, you're, this is a, you'll think it's a cop-out, but it's 100% true.

I'm 100% on both sides. And I'll tell you why. That is a cop-out. No, it's not. Because I'll tell you, you can hold two ideas in your head at the same time, Alex. I deny that. Okay, I'm going to stipulate that Miyazaki is one of the greatest artists of the last hundred years, because I believe it. And his genius is something that a computer or an AI could not do.

Number two, I'm a kid of the 90s. I'm a kid of the Napster era. You know, Tribe Called Quest, what's that song? I can't remember. They never made a dime from one of their biggest hits because it was, they had to, it was a sample from Lou Reed that they had to pay Lou Reed for. You know, I'm, I'm a kid of the mashup era. And so I actually, it's not just, I get the idea that it's,

I can see that memeifying stuff is cheapening it on some level. But on the other hand, we are living in a world where memes are a means of communication and cultural discourse. And so it is, I fall back on the Napster era argument, the mashup era argument of new art and new means of expression. for all of humanity are possible if you embrace this technology. I guess where I would come down on the other side a little more is there has to be some way for artists to be able to opt out.

I totally agree. I mean, that opt out is so important. And you have to be able to, if you're an artist, let's say you're a studio Ghibli and you don't want to participate in this, like it should be their decision, not open AI's decision to make for them, not the internet's decision to make. for them. Just because we enjoy it doesn't mean that, like, you can just basically train or emulate this style, which is clearly their distinctive style and do it because you want to. So we're going to definitely see this play out in the courts, you know, over the next couple

months without a doubt and years probably, actually, now that I think about it. But what about those that want to use this in the above, the above ground way, in the non-sketchy way? and it just brings me to this tweet that I saw from Derek Thompson or a set of tweets from him that I thought were really interesting because what does this do to Hollywood? He says the tension I'm trying to work out right now is one hearing Matt Bellany on The Town saying Disney can't make an animated feature for less than 200 million and then two realizing that image gen could make a full animated film and yeah a full animated film for $200 in like a year or two and he's saying the point of course is isn't that these personal

will be anything like Pixar quality, but rather that by reducing the cost of animation rendering and by expanding the supply of animated films on the internet, there's a potential two-front disruption, both of the cost of production and the market for animation. I think that's such a good point. Basically, what he's talking about is the barrier to create animated films is just going to drop.

There was someone who took, I think maybe they used SORA and they took Lord of the Rings and they jiblufied the first two minutes and it's actually pretty insane. It's not perfect. But the fact that like Derek is saying, it takes $200 million to make an animated film and you can use this technology and do that for much less

is going to be, I think, a disruption to the animation houses and also just like potentially an explosion of creativity. Again, talking about double-edged swords. Yeah, and by the way, we're talking about animation houses now, but give it six months, 18 months, and we're talking about actual video. We know how fast this is going.

So all of Hollywood creativity writ large, really. And then an interesting thing happened, whereas so many people started using this, they quote unquote melted the GPUs at OpenAI. So this is from Sam Altman. It's super fun seeing people love images in chat GPT, but our GPUs are melting. We're going to temporarily introduce some rate limits while we work on making it more efficient. Hopefully that won't be too long.

chat GPT free tier, we'll get three generations per day soon. And I think that's already happened. And then with Gemini 2.5 Pro, the Google release, you saw something similar happen. This is from Logan Kilpatrick, formerly open AI dev relations now at Google. He says, we're seeing a huge amount of demand for Gemini 2.5 pro right now and are laser focused on getting higher rate limits into the hands of developers ASAP. That's the number one priority right now. Stay tuned.

Well, Aaron Levy. Read. Oh, that's Aaron. Yeah, I was going to say read that one too right above. Sorry. I mean, Aaron Levy, the CEO of Box, friend of the show, he basically quotes tweets or screenshots both of these. And he says, the two biggest launches in AI in the past month are now constrained by capacity. This is what we meant by Jevin's paradox. And it is interesting. I mean, you know, we saw that, you know, the models had become more efficient. And then, you know, we had this deep seek moment. They became way more efficient. People dumped in Vida. But, but. both Google and OpenAI are telling us in the middle of launches that they are just constrained by GPUs. So do you think that the sort of the run or the runaway from GPU stocks was kind of overblown, given that we've now seen Open AI, Google, pretty sure Amazon has also said that they'll take as many GPUs as possible. And Open AI recently also said that its GPUs were maxed out. Maybe the GPU companies are in better.

shape than we thought, even as these models get more efficient. Alex, I'm going to do it again. Two things can be true at once. There can be an overbuild and overcapacity, and there can also be an infinite demand eventually, eventually being the key word, because what a bubble is oftentimes is an excitement in a market that allows for people to overinvest, overbuild, and the timing's not quite right yet. So, yes, fine. We are seeing Microsoft walk away from like a full, whatever, gigawatts thing of different places in Europe.

And there was an article that I did this week about how in China something like 80% of the buildout of their data centers for AI is currently unused. So we saw the same thing happen in the dot com era. No one was wrong if they bet that we've got to build out fiber. We've got to build out for this internet thing. This internet thing is going to be huge. They were right. but sometimes you can be right too early but that's not going to matter if you're an investor

and you bet at the wrong time but again two things can be true at once eventually the demand can be infinite but the demand could be misallocated right now but here's the thing I mean we have these companies it's not like they can't use the GPUs right I think that was the concern when these models were getting more efficient is that they would just have GPUs sitting there unused

what we're hearing from them is the opposite, that the GPUs are melting and they are at capacity. I mean, Open AI, a couple times in a month, said if we can't do any more. Now, the thing is, they are losing money. That's what I was just going to say. Go ahead. So is that the misallocation? Well, again, it's the mistiming, and it is the dot-com bubble lesson from that era of the Internet is that, you know, people weren't making money until Google. people, you know, Amazon wasn't making money forever and forever and forever.

So, right, everyone's using this to create Studio Shibbley pictures, but yes, people are paying Open AI to do it, but the economics still hasn't like exactly lined up yet. So people can be using this and the companies, you know, providing it cannot be making money. Again, both things can be true. It's a matter of when the timing lines up the right way. Yeah. I mean, this is from Joe Tsai, the Alibaba chairman. This is in Bloomberg. He warned of a, this is this week, he warned of a potential bubble forming in data center construction, arguing that the pace of the buildout might outstrip initial demand for AI services. A rush of big tech firms, investment funds, and other entities are rushing to erect servers based in the U.S. and Asia. And he's saying it's starting to look indiscriminate. Many of these projects are built without clear customers. in mind. I start to see the beginning of some kind of bubble. I start to get worried when people are building data centers on spec. There are a number of people coming up, funds coming out to raise

billions or millions of capital. I'm astounded by the type of numbers that's being thrown around in the United States. People are literally, literally talking about $500 billion, several $100 billion. I don't think that's entirely necessary. I think, by the way, people are investing ahead of the demand that they're seeing today, but they are projecting much bigger demand. So there's two sides to this, I would say. One, he's absolutely right that people are investing ahead of the demand that will get an ROI. Two is he's wrong in that these servers are being used without customers. They clearly have customers.

There is use. And that's like very important, I would say, because if you don't have people adopting this technology, you're not going to get anywhere. Now, we can say that the use is not for the best financial purposes. Like, I remember when I had access to Facebook's M, which was an early version of their assistant, which had humans, basically on the other side of it, we asked it to draw pictures all day and basically maxed out its capacity because it was just people drawing pictures all day. And so that's what's happening effectively with these servers, just it's the AIs that are doing it.

I would say to crystallize his argument, there's this belief that the next generation of AI is going to be really financially renumerative because it's going to do things like automate all coding. And I think that his argument would be stronger if he says that, that the speculation that they're being built on is that AI is going to do things that we don't know if it will be able to do within the next two, five, or even 10 years. and that is where the problems are. And so even if OpenAI is melting its servers today, even if Google are melting their servers today, if there isn't a return on investment with these higher value propositions in the next few years, there's going to be a timeline mismatch, and that's going to be bad.

Web 2.0 happened because there was so much overbuildout of fiber capacity that it was incredibly cheap for someone to create a Facebook in a dorm room. And so a lot of investors got wiped out in the dot-com bubble bursting, but then a thousand flowers bloomed and Mark Zuckerberg became a set a billionaire, yes. Yeah. So that brings us to CoreWeef, which as of now has not yet IPOed. CoreWeave, for those wondering, I'm just going to read from CNBC, it provides access to NVIDIA GPUs for artificial intelligence training and workloads.

And it counts Microsoft as its biggest customer by far. Other clients include META, IBM, and cohere. Its revenue soared more than 700 last year to almost $2 billion, but the company recorded a net loss of $863 million. Their IPO is going to be a big disappointment. They initially priced at $40 per share. That valuation will be $23 billion. billion dollars. It's way down from the $32 billion that bankers had been floating in recent weeks,

so almost $10 billion less that they anticipated their valuation would be. And they're going, I think, down from $2 billion raising in this IPO to $1.5 billion. And that is going to be, you know, are they going to be able to cover even their losses with that type of money for a year? Also, I don't know if you know this, but part of what they're raising that much money for is they're retiring a ton of debt because, again, they're extremely capital intensive as a business. And so when you see the headline that they potentially will raise one and a half billion dollars, not all of that's going to their bottom line because they're immediately going to be attempting to retire debt. By the way, just for my listeners of the show, since we're going

to cross post this on, I screwed up today and I thought that the IPO had happened. But Alex, the fact that we are talking at almost 1.30 Eastern time, and it hasn't floated. The fact that they had to bring it down from the range that they wanted and the fact that it's not, then this is bad news. Oh yeah. So like they just opened and they're opening at $39 per share. Okay. So under even 40, which is what they priced at. Under 40, which is really bad. I mean, you're going underneath what you, they were indicating to the market that they might come out. Like usually after the IPO, there's this little pop, which is basically just people buying. It means that the IPO was sort of mispriced because everybody that gets in or,

you know, they ended up spending, they end up, yeah, getting less for their money because the market is willing to give it more. In this case, there's clear bearishness in the market right away. It's already underneath its IPO price of $39. Very bad. And then if it, not a good situation for Corleaf. If it closes today, like let's call it like a 37 or worse or something, that's that's extremely bad news but the damage the damage was done basically in the IPO itself where they just weren't able to raise the amount of money that they were and now now the now we see the public also turning on them which is which is pretty interesting not in a big way but enough to be like what the hell is happening let's contextualize this because I called it on my show

the first IPO of this AI era but they're not an AI maker they're a different sort of company they're a company that provides the chips right so one One of the context here is the first gold rush in this AI moment was, as we've been discussing, we got to get our hands on these Nvidia chips. Like, this is the, this is the oil. This is the commodity that we need to make this revolution happen. So essentially, the argument would be Corweave is a Picks and Shovels company, although they lend you the Picks and Shovels.

And they don't even make, yeah. They buy the Picks and Shovels from somebody else and lend them to Microsoft. Basically is what they do. So, number one, everyone's looking. at this as tech in general hasn't had a lot of big IPOs for a few years. So this is a big one. But number two, this is the first of the AI era? So is this a bad sign for the appetite in the public markets for AI companies? But is it really an AI company? And then is this company specific? I believe that you spoke to folks at the information that have a very specific

analysis of this. Like CoreWeave has something like 70% of its business is one customer. And also if you're Microsoft, right. And if what your concern is, is that again, Jevin's paradox isn't working out, then things will get cheaper and cheaper and cheaper. And so you don't need a CoreWeave to get you your chips because you can do it cheaper and all that other stuff. So is this a company specific issue where they don't have a moat? I think it's company specific and I also think that it's still on a trend, right? I'm going to go with your line, Brian. Both can be true at the same time, right? It's not a strong company in terms of what you're looking for if you want to get AI value. And AI has been on a downturn at the beginning

of this year and it's not good for a company that wants to ride that wave. And you're right, this is from the information. So this is from Corey Weinberg. He says, As tech investors, and this is making the point exactly, seem a bit overstuffed on AI stocks, Oracle and Nvidia, two public companies, investors might compare to CoreWeave, are down 12% and 19% respectively on the year. It's hard to ask investors to pay up for a new AI firm when they're worried about their existing portfolios. Investors also worry how much the business is tied to Microsoft or Nvidia.

They worried that this is important. The CoreWeave founders sold so much of their stock already. I think they sold something like $500 million. They worry how much cash the company expects to burn, which is a lot. And this is more from Corey. A bank anonymously surveyed 135 investors, including hedge funds and long only stock pickers. A whopping 90% of the participants said they didn't think Cori we've had a sustainable moat. Essentially meeting it wasn't really a good long-term investment.

Here's the money quote. Someone said it's radioactive and I think every investor knows that. So it's not a strong business at a moment where, well, it might be a strong business. I don't know. But it's not a strong long-term investment, at least according to these bankers, at a moment when people are kind of pulling back on AI. Alex, you know what this is, and we haven't had one of these in a long time. This is a perfectly timed IPO.

And I'm not casting aspersions. You mentioned that the founders have cashed out a lot. But sometimes companies ride a wave and they're private and you try to get to, to public markets before that wave dissipates. And I think that we're watching that happen right now. Wouldn't it be great though if they went like a year ago? It would be, it would have been better for the people that invested in this IPO.

They could raise way more money. I mean, without a doubt. There was a reason they were, they were hitting a certain valuation then. Yeah, but they also, you know, their, their revenue was up 700% I think it was over the last 12 months. So they had,

it's all again it's all timing Alex so you you had to wait until you could show that level of revenue growth um to to achieve you know if they didn't have 700% revenue growth over the last 12 months they wouldn't have been able to get 10 million dollars 10 billion dollars uh in in public markets for sure but talking on timing I do think what this is also showing is that the euphoria and the unquestioned money spigot for AI is over And I texted Corey, who wrote the story. I said, what's going on here? And he said, I think people are just stopping.

This is what he wrote back. I think people are just stopping and thinking about how companies are actually going to make money in AI and who's going to be left holding the bag. I don't think that was the case a couple months ago. This is definitely AI entering this new era where people are thinking about, yes, the ROI, the sustainability of the businesses. And I don't know. I think that's going to make the AI industry make everything more difficult.

for the AI industry than it has been up until this point agreed but it's interesting to me that you're willing to make a state you're willing to make a claim there um because again apologies i i wrote a book about the history of the internet and the the dot com bubble was not a straight up rocket ship you had you know the the Asian flu crisis you had the long-term capital management crisis like there were times when people were like okay the boom is over the boom is over the boom is over these things move fast. And we could be talking again in three months and there's more euphoria because open AI lists to go public or something like that. And so I like the call. I agree with the reasons that you're making that. But I would also say, talk to me in three months and we could be talking

about something completely different. Yeah, it's a great point, Brian. And I wrote this down in our show doc, which was just sort of this thing that I've been scratching my head about. And I'll just read I don't get how the market is pulling back on AI because the technology is as cool as it's ever been. It's getting more efficient. And we're also seeing massive data center buildouts and chip buyers out of chips. So what's happening? I mean, to me, and this is basically to crystallize the point, the technology like we saw

with the Ghibli statement we did at the beginning of the show, the technology is absolutely cooler than ever. It's doing all the things that, well, not maybe not all the things, but it's doing many of the things. But is it profitable? Well, it's not profitable. But it is, it does seem to be on this path to be able to do. I mean, you could run, I don't know, old chat GPT and run it profitably, but the losses are coming to try to build what's next. And then that will become more efficient. So it is interesting to me that the market is not turning on this technology, but cooling on this technology as it starts to, as it continues to advance more and more and gather more steam

among people. Let me ask you this real quick, um, from your perspective, because I'm, I'm interested in talking to you because you're deep in this stuff. If we get a lot of signs like we've been getting recently that there is a KAPX pullback on data set or spend, is that bullish or bearish for AI because obviously it's bearish for certain segments of the AI market because that spend that they were planning on might not come or is it actually bullish? Did I say bearish last time? Anyway, is it actually bullish because that actually means that things are

coming down in terms of cost in sort of like the Moore's Law sort of way that we're used to with technology so that now if it's cheaper, then it's going to happen more and there will be more chances for profitable companies to be created. I will say I don't think it's going to happen because the CAPEX pullback. The CAPEX pullback. I don't see it happening. We have big tech that's going to spend 300 billion on CAPEX this year plus. Much of that is going to AI. Open AI, there's rumors now that or reports really that they're about to raise 40. billion dollars. They raised 6.6 billion last year and it was the biggest VC round ever. Or could you imagine them raising 40? That's crazy. Well, Masa likes, Masa likes to do things big.

Yeah, that's money. That's money. So I don't see a pullback coming. I think the Microsoft headlines were largely Microsoft saying we had this dedication to open AI. We were trying to build out for them. They're going to go do it on their own now or with other partners or with the diversity of partners, we don't need to do that. We're seeing everybody out of GPUs. We're seeing all those GPUs being used. We're seeing bigger training runs happen. I don't see the pullback happening.

And if we do see a pullback happening, I will view that as a bear sign here. Because I just think that like this stuff is going to be very expensive and they're going to need all the GPUs they can find to serve it. And then we'll once they do that, we'll then see basically improvements of models. then we'll try to, they'll try to do the next step change. Yeah, but that's my argument, and we don't have to go on this, that if, if the cost do come down, that's actually bullish for the AI as a technology becoming pervasive across all of tech and society or whatever. Right, but we're going to have, we're going to have Dylan Patel from semi-analysis come on

the show in a couple weeks, maybe next week. No, definitely in a couple weeks. And what he basically said, I'll just preview it. What he says is basically it's this step thing where, like, you advance your capabilities, and then you find ways to make that more efficient. You advance your capabilities and you find ways to make that more efficient. And so it's a step,

but that always will require spending a lot more money to advance your capabilities. And then you can use that stuff in a much cheaper way. So yeah, Brian, we should definitely talk again so in three months and we can see where this thing is going. Yeah, yeah. But before we go, I definitely want to get to your story about the future of Silicon Valley.

You have a pretty provocative piece on YouTube. And I want to hear your thoughts on it. The title is, is Silicon Valley about to lose its monopoly on tech? My thoughts on sovereign tech stacks. So can you run us through the argument? And then I'm going to, I think I'll debate you about it a little bit. Yeah, yeah, real quick. So this was something that I, I did a bunch of stories this week about how Europe is maybe going to pull back on buying US cloud infrastructure. There's been a bunch of stories like that about recently, like the people don't want to buy American because they're concerned about, you know, here's the thing. What if the rest of the world

did to us what we've been doing to China in terms of like import bans, export bans, things like that where we don't feel like your tech is safe to use internally? The, you know, the argument that has been made that we've been covering for years now is you have to onshore chip development, silicon development because if you lose access to chips and China invades Taiwan, your economy is screwed in the same way that if you lose access to oil, your economy is screwed. The term that is being banding about in recent weeks is tech sovereignty. So the argument is that if people felt that way about silicon, about the chips themselves, now it's moving to the entire tech stack. If you are a country that feels like that feels like that.

like, oh, we could lose access to databases and cloud computing and even social media, because social media is the modern communication network in the same way that you would fear in a war that an adversary would take down your telephone network or whatever. So people around the world, and in my piece, we're talking a lot about Europe, are starting to pull back and say, it's not just chips, it's everything, it's the entire tech stack. So the argument that I'm making is that Silicon Valley for at least 30 years, post-Cold War, has had essentially a monopoly on the tech stack, right? And also we made the best stuff and brilliant innovation and things like that,

but we were the default option. No one ever had a motivation to choose other than Silicon Valley, right? So now if for geopolitical imperatives, for cultural reasons, people are starting to reconsider that, Silicon Valley for the first time, is not the default, does not have the monopoly on the tech stack. And the example I use is Pallant, or Arindrol, Palmer Lucky's. Andro. Andrel, sorry. I never can pronounce that right. It is tough.

The defense tech startup. And right now, Europe is rearming. They should be, it should be boom times for them, except Europe's not going to buy from an American tech company because reasons that we don't necessarily have to go deep into, but they wouldn't trust American tech. And it's not just, oh, maybe they'll spy on us. Maybe they'll do a kill switch and stop our drones from flying or something. They're afraid of tariffs, export bans where all the sudden you live.

lose access to the supply chain, right? Extrapolate out beyond that, and you have things like Europe saying, we don't, we need local cloud because we don't know that if we, if a trade war happens that maybe as part of the trade war, tit for tat, we lose access to our data. And then you layer on top of that what we've been talking about with AI. And I make the point in the piece that AI is the first new technology in 30 years. It was born in Silicon Valley. Silicon Valley is still the leader of it. But for the first time, at the nascent stages of this technology, it's not a Silicon Valley exclusive. It's not a monopoly. There is Chinese tech. There's Chinese AI, Middle Eastern AI. There's different flavors of this. And so I, my argument is, does the business model of Silicon Valley, is it prepared for the fact a reality where instead of addressing 90% of, of the global market. The market is bifurcating. The market is fragmenting and scale no longer

means potentially every human being on the planet. Scale means maybe you can only sell to certain markets and maybe your own local market. So let me, I'd love to hear your thoughts on that because I value your opinion on stuff like this. Yeah. So when I read your story, I was like, all right, well, let's see how much like losing a market like the European market might hurt for the tech giants. Actually, even a small decline in Europe would hurt tremendously. Apple sales in Europe, 26% of total revenue. Meta's ad sales in Europe, 23% total revenue.

Amazon's international sales, 23% of total revenue. And let me interrupt real quick, because I want to say, another thing that I say in the piece is someone could do to meta what we have attempted to do to TikTok, which is to say, we're going to ban Instagram, right? So it's not just like hard tech or software. also tech, tech because, you know, social media is media and its communications. I'm sorry, go on. Yeah. So that being said, I don't see Europe banning these software platforms. I just think that you make your citizens really angry. They're restricting, you know,

future development of it. You see Apple, you see Facebook say we're just not going to roll out our AI products within Europe. But to take something away like this, you saw how hard it is for the U.S. to do it with TikTok. I'm pretty sure TikTok's still working. I was on it this morning, even though the U.S. really wanted to ban it. I don't see them taking it away. Let's pull back from a ban because the better example is the piece from, was it wired, where that European companies are starting to say, we need the cloud onshore. The idea is onshoreing for things like the cloud, for things like software in the same way that onshoreing or reshoring for chips, becomes a geopolitical necessity, a sovereign imperative, essentially.

So, like, if you were, again, I don't have an opinion on this current administration's policies, but if you're in Denmark and you are going to buy a cloud services and storage and things like that, maybe you do want to consider having your cloud be local, right? What would you say to that? I think you're going to go with Best and Breed. I mean, I think you'd have to anticipate that the risk of, like, true geopolitical fissure between the U.S. and Europe is present. And even though, like, it might be in a rocky place right now, I just don't see that full separation happening. I'll tell you why.

So I would still go best in breed if I was building. I mean, a world where things get so bad between the U.S. and Europe that your AWS stack is at risk is to, I mean, maybe I'm not imaginative enough, but is to me so far-fetched that I can't fully, I wouldn't say a company would begin to change its strategy on it. But that being said, there was a piece, part of your piece that I think is totally spot on and actually bears paying attention to. And that is that basically the U.S. has had a monopoly on tech for a long time. I mean, of course, we know there are, I mean, Spotify, for instance, which we're broadcast

on today is a European platform. So not completely, not a complete monopoly, although certainly U.S. companies have tried to squeeze these companies. I mean, Spotify is one of Apple's biggest opponents because of app store revenue. But AI is, you know, however cliche it sounds, it's a democratized technology. I mean, we're seeing models come out of really China. I mean, Deep Seek, and we're going have a China episode coming up as well, going into all the different AI applications in China. And the Middle East, because you're right. I mean, basically, I'm going to give you a moment to talk about the Middle East, but basically, if this is a technology that requires scaling infrastructure, and if you have unlimited energy for sure in the Middle East and resources, you can compete and they can

compete. Yeah, I don't want to talk about it in the Middle East, but I'll give you one more wrinkle to the argument that I'm making, which is that. Essentially, Silicon Valley Tech has been the default. There's never been a motivation to choose otherwise. There is a motivation for the first time. And I agree with you. Listen, tomorrow this administration's positions may change. There could be another administration in a few years.

But I think the genie is out of the bottle in the sense that if people start to realize the geopolitical necessity of the tech stack as, being existential. Even if everything goes back to, we're all friends and kumbaya, whatever, people have learned this lesson or have gotten this fear in their hearts. And I think that that is for the first time changing how, why people would make decisions. And it only takes the genie being out of the bottle for there to be, well, okay, maybe there's a market now for just a European cloud. Maybe there's a market. We keep using Europe. For India, you know, if Europe doesn't want to buy U.S. defense tech, does India?

Does Brazil? And so go down the line in terms of cloud or every product is a tech product now. So I can't buy a Chinese car right now. Like, what if India doesn't allow U.S. smart cars and smart TVs to be sold there? So my point is, is that one of the things that Silicon Valley has assumed for 30 years is that we can sell our products to 90% of the planet. But what if the global market starts to fracture and become localized and localized? And then the last point to layer on top of that is, is what does that do to the talent?

Because the talent for the last 30 years around the world has come to Silicon Valley, because that's where you go to achieve scale to reach 90% of the planet. But if you can stay home and have an addressable market that's India, an addressable market that's Latin America or whatever, and it is a fractured world, then the talent doesn't come to Silicon Valley. And that's why, listen, I say in the piece, people have been saying since I got into tech, chicken little stuff, Silicon Valley is about to die or whatever,

and it's never come true. And I've never believed it. But I see the angle for the first time where I'm worried, and I love Silicon Valley and I'm an American, and that's from my perspective. So that's why I wrote that piece. Yeah. I mean, I think that, yeah,

if there is a separation, it's going to be. If the separation of this magnitude happens, it's going to be worse than just like Silicon Valley losing its dominance. But it is challenging. It is challenging, I think, for companies outside of Silicon Valley to build in the same way, it doesn't mean it can't be done. But the concentration of talent and capital, and you know this because you wrote the history. And the tolerance for risk is really unmatched outside of that, outside of that area, really. and it's going to be, it'll be tough for the rest of the world to catch up or displace. But this is a thing. Think about it. Let's end here because you mentioned you can't buy a Chinese car.

You can, I think, but it's just a 100% tariff. I can't buy Chinese phones. Think about, think about, we could, we don't have to get all the way there. We get a little displacement. Think about what's happened to Apple in China. It became a point of national pride to not have an iPhone and to have a,

Huawei phone home, Huawei Mate. And Apple is still 15% of its revenue in Q4 was coming from China, but that wasn't the 18%. And that has dogged Apple revenue for quarters, the fact that their revenue isn't up to China standards. So it doesn't have to be an earthquake. It can be a ripple, and that could be quite destructive. Right. And talking to Mr. Big Tech himself, is that that's one of the concerns for me is in a fractured market, If an AWS has to go market to market as, you know, the phrase is always, write the code once, sell it everywhere. If that world changes, my only point was, is I don't think people are talking enough about what would happen if the world changes in that way where you have to, you know, oh, Europe is regulating us, so maybe we have to change our software for the European market. But what if every market is different and you have to go bespoke market by market?

If that happens, that is very different than what Silicon Valley has been used to for 30 years. Definitely. And another thing that Silicon Valley has not been used to is dud IPOs, especially bad AI trading. And Corweave is now down under $39. Down 3.77% on IPO day. That will change before the market closes, but not exactly what they've been hoping for. And certainly a day that we'll come back to and maybe in our three-month recap, Brian, We'll see, was that at just a data point or was that a sign?

I'm calling it. If it's under 37, that is very bad news by close. All right. Brian McCull of the TechMeme Ride Home podcast has been here with us. You can find the show in your podcast app of choice. Just type in TechMeme right home. A really nice compliment to big technology. You can come to us on Wednesdays and Fridays for our interviews and news recaps and go to TechMeme every day for the latest in what's happening in AI and the rest of the tech world.

So, Brian, always great to have you on. Thanks for coming on the show. Alex, one of the funnest conversations I've had in a long time. Definitely. All right, everybody. We'll be back on Wednesday with Ron John Roy. More AI news coming your way. And then we'll have a special guest coming and joining us next Friday.

So hit subscribe. If you're not subscribed already, you're not going to want to miss our show coming on Wednesday and then on Friday. Thanks for listening. And we'll see you next time on Big Technology Podcast.