NVIDIAdrome with Steve Burke of GamersNexus

00:00:02
Speaker 1: Also media.

00:00:05
Speaker 2: Hello and welcome to this week's Better Offline. I'm your host ed Zetron. Subscribe to the newsletter, buy one of our beautiful new fuck Data Center t shirts and join me and welcoming the wonderful and incredible Steve Burke of Gamers Nextus, who joins us today to talk about how great everything is.

00:00:33
Speaker 1: I didn't know there was a fuck Data Center shirt.

00:00:36
Speaker 2: Yeah, man, we got fuck datas that. You've got this shirt with like a skull, whether you is so that you can wear it out you're not technically swearing.

00:00:44
Speaker 1: Yeah, I might buy one. We have the we have the new micro slop shirt. Oh yeah, between the two of us, as we're making the world the better plays just like Silicon Valley.

00:00:55
Speaker 2: Yeah, I'm just worried that, like someone's gonna be walking around and do something wearing one of mine. But you know it's not my not my fault. Yeah, it's so this this year's GtC was It kind of reminded me of this year's CES in the sense that it felt like a couple doing stuff to pretend that they were doing well, like being like, yeah, you know, we're great, we're going We're going to Puerto Rico going on vacation, We're feeling great, we love each other, we're only in couples counseling three times a week. It was so weird.

00:01:25
Speaker 1: Yeah, it was. I gotta say the DLSS appearance, the gaming appearance, I wasn't expecting just because.

00:01:34
Speaker 2: What happened there, what happened, Well, why was everyone upset? This part just completely passed me by.

00:01:40
Speaker 1: I So DLSS is a gaming thing, and one thing I wonder is that there's a huge amount of backlash towards it from just everybody. I mean, it's like one of the most hated dance I've ever seen it now, so I'll get into what it is. But what I thought was interesting was, uh, I don't know, just like Jensen's reaction afterward, where they had multiple statements and reactions, and I almost wonder, if you're Jensen, if if you look at that and you're like, okay, fuck gamers, like.

00:02:09
Speaker 2: I mean that kind of already doing that already.

00:02:12
Speaker 1: They are, But I think I mean in like a more direct less of a shrewd business decision and more of a no, seriously.

00:02:20
Speaker 2: No we have we actually don't. Yeah, we're just saying it out loud.

00:02:24
Speaker 1: Now, yeah, so yeah, it's it's basically a filter. It's an AI slop filter more or less. They said it's not a filter, and then they put out multiple different public statements that are like paragraphs long when you read all of them, and they they say it's not a filter, and then they proceed to describe a filter.

00:02:42
Speaker 2: So, but isn't the LSS generally good and that it adds frames like that's a good thing.

00:02:48
Speaker 1: DLSS has has some things it's done pretty well over time. So it starts as basically an upscaler, and then it progressed to where there's been times where DLSS can be better than native while improving the performance just because of the way games render. Right, Yeah, different things. So, I mean there's certain types of for example, fence posts or grates or things that have a lot of repeating patterns straight lines. Sometimes you have trouble with that at low resolution. So DLSS to an upscale and sometimes even repair these things they rendered in a way that's just not good. But they've now attached this this basically Instagram twenty nineteen face filter to DLSS and kind of tarnets the name of something totally unrelated. It looks like they're trying to say, look, AI it has uses, please kill the bubble.

00:03:45
Speaker 2: And so it seems like it's not just like adding frames, it's attempting to fill in details. It's almost like yasifying and yes, this is really horrible.

00:03:55
Speaker 1: Yeah it so it basically what it's doing. So they originally said that it changes it uses color and some vectors on the screen to then improve things by using AI generation to do whatever it is they think is improving it. What they're actually doing, yeah, they're changing the entire feel of the game. And I mean you know this where AI generated images, especially when you ask them to be photorealistic, they all pretty much look the same, right, like, yeah, yeah, there's no style, there's no character to it. And whether or not it's useful for people, I don't even care. I think what most people cared about was you're stripping all of the character out of the artwork. And the you know, part of the art for video games is you have to work within these strict confines to make it regnerable. Yeah, and they strip it out and they replace it with something that just looks like the same slop everywhere. And Video's initial defense was well, but the artists can control it. And I think the answer to that is, I find it very unlikely that the artists make these decisions. I think this is probably an executive level corporate decision of we're gonna partner with Nvidia to use this thing, and if you're an.

00:05:15
Speaker 2: Artist, fire artist, so yeah, yeah, like yeah yeah. And also wouldn't it add a bunch of work to an artist anyway, because you'd have to tweak this completely separate thing that also is not consistent and is generative.

00:05:29
Speaker 1: Yeah, and I mean it. They didn't even show any like actual motion other than one game, and it looked awful. And I think it's likely that the face details slightly change, you know, from scene to scene, because it's some form of generative, so it's it's not going to be perfectly consistent. So anyway, they're they're trying to shove AI slop into real time rendering for games now in a way that I will say I was, I guess proud of the gaming can unity to see. The amount of pushback was unbelievable, Like these large companies have done far more corrupt, terrible things in the real world with governments and data centers whatever, and this got more pushback than that stuff.

00:06:17
Speaker 2: Yeah, I think everyone's just kind of fucking tired of Nvidia at this point, because it it isn't really clear what Invidia is doing other than selling more stuff, and it isn't clear who is actually using that stuff. I mean, I just the the indignity of watching a company with like a three four trillion dollar market cap talking about open claw. Truly, like, to this day, I cannot get a straight answer as to why you should use an open claw.

00:06:47
Speaker 1: I liked his uh in their pregame for GtC where he made the comments to the other I don't know the podcasters about I think AI is going to make us all busier. Isn't it making you busier? Ha ha?

00:07:03
Speaker 2: What do you do all day? Jensen? Jensen, You're not that busy. I refuse to believe any of these people are busy at all. In fact, I think that they're just kind of fucking sitting around coming up with new ideas of stuff they can say at GtC, and then they go to sleep for three months.

00:07:18
Speaker 1: Yeah, because I'm not even sure. It's just not even a good I don't know, it's it's not news that these executives are toned aff obviously been disconnected from reality, but it is a very disconnected from reality thing to say to make the cell of AI is going to reduce the workforce for the executives, and then also say that it's making people busier. And it's just like these, this isn't the cell you think it is.

00:07:44
Speaker 2: Also, these aren't thoughts that connect together. If you say them both in a sentence, you sound insane. I just this. I'm sure you haven't really explored open claw march. I certainly haven't wanted to. But it appears to just be what if I turned over my compute to a large language model and nothing really changed. That's long and short of it. So, which is what makes it so insane that in videos, Look, GtC used to be a place they announced hardware and new technology, not hey guys, the latest open source thing you're gonna forget about in six months that is now the most important thing in videos stack.

00:08:25
Speaker 1: It's it's really weird because we've covered on video for so long that I've met a lot of the engineers there. Yeah, and the people who are running these AI departments now are at hardware and software levels are people that I worked with on the gaming side, you know, anywhere from fifteen years ago until even more recently, like three years ago, and it's a little sad for me seeing these people move over to purely AI in these big important roles, and it's just like, I don't know, they're smart people, and I have nothing I respect about their work anymore for the most part because of the just I think, sort of feckless rollout and widespread damage that it's causing. And it's just it's interesting where you see on video moving their pieces around like a chess board, where they're going, Okay, this guy is really good at this specific type of research. Let's strip him out of the thing that he's been doing really well for ten years in gaming or whatever and put them into AI, you know, and then outcomes like the DLS.

00:09:33
Speaker 2: Nemo Claw, Nemo Claw, Nemo Claw. This leaked to like four months ago. No one knew what the fuck this was, Like they the guy who made open claw sold hisself to open AI. Cool for him, I guess, And now this is like, now this is the most important thing to do in video now, Like what's It's just it's so strange. It's it's just very strange. They did it song. At the end of GtC as well, they had a song a horrible altar I saw, take him. I was a disgrace to journalism. Sorry Tay, who was like, not gonna lie, it's just kind of a banger.

00:10:13
Speaker 1: Maybe go to the song as a banner.

00:10:15
Speaker 2: Yeah, banger. Yes, he did go to the doctor. That's my first thought is like, you need to go to the doctor. Man, you have like I said, to hear it and it was not good. But I heard it and it made me angry.

00:10:28
Speaker 1: Yeah, yeah, I was thinking about this when AI music covers first started to really propagate on YouTube. And yeah, I remember the first one I ran into. It was like Slipknot reimagined as soul like blues. Yeah, and I did. Actually, that was the first one I saw. I listened to it and I was like, damn, this is like not bad. And it was only you know, thirty seconds into it where you realize the horrors of music and entertainment. The entire currency of it is attention, and you only have so much time you can give to listening to music in a day. And for each of these AI songs that someone's going to listen to, and even worse for the real artists. If people start liking them, that's one less real artist who may already be like just a starving artist archetype who doesn't get that view or that listen.

00:11:29
Speaker 2: Yeah, and I've heard some of those AI thinks, and it's always like the first ten to twenty seconds you're like, okay, this is and then nothing changes. It's just very repetitious. But you're right, it only exists as a kind of phishing law to get people in just kind of hear in the background. It's just patently evil. It's just wafting evil at the moment, and even by the standards of regular software, it's not really clear what any of these companies are doing anymore. Like I GtC, I don't know if i'd describe it as exciting, but it felt like they were selling real things every year. Yeah, and now it's just you want an open claw. Oh, actually, I forgot my favorite part. So Jensen went on the all in podcast, you know, the classic guys we don't think of anything happening too, and he went on and he was like, what, so, what should happen now is that you should give engineers when they join your company two hundred and fifty thousand dollars worth of tokens a year and that's part of their compackage. It's like, oh, I'm sure, I'm sure he would love that. I'm sure he would love that. I'm sure. Like so much of the so much of this is just Jensen going out there and being like, it'd be really good if you use my stuff.

00:12:43
Speaker 1: That's not even camp though, right, Like, but isn't it for work? It's literally just budget.

00:12:49
Speaker 2: Yeah, it's it's it's your compensation so that you can do your job. Like, it's just it's really weird. It it's really weird. The more of these we see, the weirder it gets, Like it says on the website. Now and video is opening in the Agentic Frontier with Sorry, and Vidia launches Veros CPU's purpose built for Agentic. Ai'm not sure that that's really true considering GPUs run it. Yeah, large language models and something about Vera Rubin and Agentic as well. It's just it feels like mad lips at this point.

00:13:22
Speaker 1: I mean together going to old gtcs, like some of the really early ones and announcements, the exciting announcements where you know they'd have the shield like handheld gaming device, or they'd have the Shield TV if you prefer that. They sometimes some GPUs, sometimes it would be Quadro or or professional class have a fun gaming class. Yeah, and it was pretty interesting and they had servers too a lot of the time, and they were also interested in It wasn't pitched the same way. But the biggest thing about it, and I actually really did like the show, was there are a lot of panels where you could do really technical discussions, and they still have those, but it's just completely transformed into now it's the AI Show and there's like almost nothing else there. So we don't even bother going anymore. Total company shift. I mean they are they're really all in on this. I speaking of things that Jensen though has studies all in on. I don't know if you even want to get into this one, but he made the comment in the in the all in I think it was podcast about AI uses in the Middle East given the current situation.

00:14:29
Speaker 2: Oh god, what what do you what? Why is Jensen Jensen? You run a you run a GPU company? Come on, man.

00:14:37
Speaker 1: He said that he believes there's a paraphrase and he believes there's a reason for the war with Iran and talked about potentially use cases in the Middle East, and I don't really he said. He did say I'm one hundred percent in on that when he was talking about the subject, which is just like, how did we get from You're on stage talking about like doom and quake and making jokes about crisis. You know, they like judging the reasons for various wars. I don't know. You just you wonder like they've been angling for this military industrial side for a while now, and that to me, it seems kind of like more of a signal to the US government than anything. I'm like, hey, we want to sell you stuff, But I just.

00:15:23
Speaker 2: I think it just feels that it's kind of my thing to say. But it's the end of the hypergrowth era in the sense that it's just everyone going like, oh, what do you you want to do war with the GPUs. I guess you want to, Yeah, what do you think if the do you want to blow stuff up? It's not really what we do well, I mean, actually I take that back. Doesn't in video also already involve itself with what's it called the robot platform.

00:15:47
Speaker 1: They're definitely involved with Palette here, which is involved in war.

00:15:52
Speaker 2: It's not Isaac, there's they have a specific Jetson Oh ye, jets Jetson's already involved in I can't say specific leagues truly done, but I swear to god I heard it was involved in some sort of war related censors that I'm not.

00:16:05
Speaker 1: Sure if I should I shouldn't say it, but yeah, Jadson's we have photographs of things that they are in that have been used in uh in Ukraine. So I don't know if I couldn't say any more of that. That was a conversation with some congress people earlier.

00:16:25
Speaker 2: Yeah, that's the thing though, It's like they talk about the Nemo klor and they talk about doing war stuff, and they talk about all these things when the only ninety percent of their revenues from these fucking GPUs. It's not even it's not even a case of like, oh, they're making like fifty percent of revenue all pretty much all of their revenue comes from the GPUs. And just before we got on, I have I've been trying to answer a question of how many data centers have actually been built in the last year. It is a very difficult question, but from what I can tell based on like there's like four or five gigawats might have gone online, which means we're running at about it takes about a year may to install like half a year's worth of GPU seales. Oh wow, Okay, it's it's really I don't think people realize how slow this is. But I just can't work out at this point why anyone's buying these GPUs anymore. Like it's just just as if they stop doing it, it's not the bus from speed their company or shut down.

00:17:29
Speaker 1: Like yeah, I mean it might be like that. There might be a chain reaction, you know, of just.

00:17:37
Speaker 2: It's just so bizarre because Metaverse was killed last week. They killed it. They shot it with a gun. Horizon Worlds will no longer you can. But then they were like, oh, we're going to keep it running kind of, but it's not clear what that.

00:17:49
Speaker 1: Was stupid the beginning.

00:17:50
Speaker 2: It's so it's so cool, it's just.

00:17:52
Speaker 1: A shitty video game, Like it's not even you could literally just play a video game and it would be the better.

00:17:58
Speaker 2: It's just But the great thing is they no longer bringing it to Horizon Worlds to VR, but they are also bringing it back for an indeterminate amount of time I I just I don't think the tech industry knows what it's doing ever anymore. I think that they just wake up one then they're like, what what do we What do we do today? What we do? We fucking what? We buy? More? GPUs that? What do you like?

00:18:19
Speaker 3: That?

00:18:32
Speaker 1: It's validation seeking, right, It's almost like the whole industry is seeking validation. And I think some of that may mirror what we saw happen to Intel for for multiple CEOs running where it's uh, they eventually moved from engineering focused CEOs to accounting focused CEOs and marketing in one instance, you know, and that didn't work out very well for them. And I don't know, I wonder if maybe there's just so much money in everything now that it's become a game about and it's like cost engineering or it's stock engineering or whatever rather than engineering engineering. I think maybe that's some of where we're seeing that well.

00:19:16
Speaker 2: I think we're came out just before we we got on as well. We're finally getting Enron shit though, because open ai is trying to offer they're trying to do this spurious joint venture thing with private equity firms and they're like, well, what we'll do is we'll raise money and then the private equity firm will pay us in advance to install chat GPT across their portfolio companies. Right, and as part of this deal, open Ai is guaranteeing a minimum minimum return of seventeen point five percent. Now do you maybe think that's a good question? And also a seventeen zero point five percent return on what I've read the routter's opticle four or five times, like kind of work it out? I just this whole thing feels like an exploitation of ignorance. It's just like how many reports actually read the emails they get or when they write up a story, how hard do they think about it?

00:20:13
Speaker 1: I had I had a reporter from another from a business focused publication email and asked to talk about and video And basically the reporter was interested in learning from our perspective, why is it that gamers and why is it that consumers are so mad at nvideo? And the first question that was asked of me was, uh, did when did gamers for like, did did gamers notice or were they happy about Nvidia stock going up? When it started to go up? And I had to start It's like let's start from scratch. The consumers ten years ago weren't paying attention to Nvidio stock. You know, they wanted to buy gps to play video games. Like people weren't. You weren't interested in the GPU for the stock price.

00:21:03
Speaker 2: I don't get stock when you buy a graphics card. You don't get that.

00:21:07
Speaker 1: That is a good idea. Though, maybe that's how you can do next.

00:21:11
Speaker 2: Yeah, and they're just like, that's how they do stock buybacks. But each time they sell a GPU, they buy back. It's just I love it. But that's I actually think that that question is really indicative of the media problem with in video, where it's just like, oh, in video, right, so every single person who's ever used in video cares about the stock, right, That's that's not true, Like most people should not think about the stock market.

00:21:33
Speaker 1: I would say the I mean, for sure, I don't know exactly, but at least going back to the earlier days for end video, I can't say their stock price ever really came up in any discussions like in technical news, Like it definitely came up in news in general, but in our space, you know, it never came up.

00:21:53
Speaker 2: Why would it come up? Why would it come up. You are you are covering hardware, You cover hardware. Why would the stock come up?

00:22:00
Speaker 1: Yeah?

00:22:01
Speaker 2: Why? Why? And I think it's also because, and I mean this with no offense to the reporters that listen, I'm sorry, but these people did not know who in video was four years ago. They had no idea. They did not they had no idea. They were not thinking about in video they had they did not think about. It's kind of like how a much larger version of how I think. I now see seventy five different guys who are now an expert on the straight off Umus despite learning what it was around two weeks ago. Like it's that I'm now, Oh, look at the straight and did you see the venture capitalist whose solution to it was to just build a big road. It's just there's this little like blob, like the kind of the the tip of it to.

00:22:45
Speaker 1: See understand how health?

00:22:47
Speaker 2: No, No, it's fine. You just draw a giant road from like the coast of Iman to the coast of the UAE and you also this just.

00:22:58
Speaker 1: Then, structure infrastructure is famously quick to build.

00:23:01
Speaker 2: Exactly, especially during a massive assault from a large a nation the size of Alaska. Yeah, but on top of that, I assume he has some way around the giant rocks either side, and also a way for the boats to get in. It's just that easy. But I genuinely think that that's the era though, because people are like, yeah, you know, there's like two hundred and forty gigawatts of data center's being built right now, and it's great. Actually there's only about five gigawatts in America that's actually under construction.

00:23:34
Speaker 1: But you know, easy because if you look at maps of like announced or planned or hopeful data centers, there's thousands of them in the US.

00:23:42
Speaker 2: Yes, you know, yeah there's oh god, I had it. I was just working on it before it came on. But it's something like there's two hundred and forty gigawatts, but let's see. Yet, it's very very very strange that, you know, yeah, there's about five gigawats that are actually under construction, despite like sixty or seventy that's allegedly meant. There's meant to be like thirteen to fourteen gigawats that actually there's meant to be I think fourteen that comes on this year. Hmm. From yeah, Sideline, Sorry, I'm gonna get it. Right this tunge because I actually found it in my notes. I'm sorry everyone. Sixteen gigawats of capacity was slated to come online in twenty twenty six across one hundred and forty projects, but only five gigawats is currently under construction. And that's under construction. That's under construct, that's just being built. It doesn't mean they're actually going to be built. And the only reason this is happening is so that we can keep in video inflated.

00:24:37
Speaker 1: I people still I don't know who they're pulling the wool over the eyes of these days, but we always reading the following this news of some of the ones that were canceled, and you know, from pushback from the community and town council hearings and stuff, and it's interested in reading the comments where if they're even humans these days, I don't know anymore, but so the comments being about, uh, why are people proud about turning away jobs? And it's just like, do you know how many jobs this like million square foot facility would create because it's not a lot.

00:25:13
Speaker 2: And on top of that, even the jobs supposedly created by building them, they're all flown in like yeah, they're they're not local. It's not like you get local builders. Well it's start things well, Stargate ampline. They've just flown in people to just move in to town. So I guess like that helps the economy, and then when it turns on, which it never will, they'll be like seventy jobs.

00:25:38
Speaker 1: I mean that's the same, right, It's like it's the one it turns on where Okay, you get this transient burst, great, but this isn't It's not like you're I think the branding of data centers as AI factories, which is insane by the way, but I think the branding of it it's like blue collar washing or something. It's like they they they're trying to make it sound like a I don't know, like like the grit and the hard work and like the manpower you get out of a traditional factory. Except it's a bunch of computers in a room, yeah, you know, and there's like fifteen guys in the on the floor who are just keeping them turned on, which is a real job and can be an important job. But it's not a factory.

00:26:24
Speaker 2: Yes, it's not like bringing thousands of factory jobs from the bygone era, Like we're not doing like we're not going to repeal NAFTA, so we're not actually going to fix the problem here. But my favorite, my most favorite of all recent data centers are not being built story is from Sachin Kati, who is head of Infrastructure Open AI, March sixth, twenty twenty six. Construction is underware our site in Port Washington, Wisconsin. And there is just a single steel beam.

00:26:54
Speaker 1: That was just one beam. I saw your tweet that was like probably my favorite tweet a lot time, just.

00:27:01
Speaker 2: The single just and at this rate, great, this will be done in like twenty twenty nine.

00:27:08
Speaker 1: Maybe the beam was almost entirely in the ground.

00:27:11
Speaker 2: Yes, And it's like there's still there's like four guys, like three guys that sitting kind of standing around it like poking it, and one guy walking towards it. It's just it's just really what's great? But they only do this because none No, I don't think I know I sound conspiratorial, but it's like the mainstream media just doesn't say anything. Like to naturally move into my favorite story, super Micro, Yes with a Wale Lear who so to explain for the listeners. Super Micro is an AI server company. They two years ago had a massive accounting scandal where a bunch of people resigned. I forget exactly what happened in the scandal, but like they had to restay earnings, they had to delay earnings. But then things got back on track late twenty twenty four, twenty twenty, Sorry it's twenty eighteen when the scandal was. But twenty twenty four, twenty five, they got back on track. They were selling GPUs to Crusoe, they were selling GPUs to core Wave, and then very recently there the Apartment of Justice arrested three different guys, including co founder Wally yet Lao, for shipping like hundreds of millions of dollars of GPUs to China and like changing the serial numbers using a hairdryer. Yeah, which I fucking love it.

00:28:26
Speaker 1: They're removing the serial numbers and the stickers and then using a middle in logistics company to basically repackage and forward them.

00:28:33
Speaker 2: And this story, this story has just kind of disappeared already. Yeah, there was a picture where Jensen Hwang is standing next to co founder Charles Lang of super Micro, just stanning.

00:28:45
Speaker 1: My personal opinion and this is an opinion, but okay, my opinion is that there is no way that Nvideo couldn't have known this was happening. Yeah, I just think it's like actually impossible that they were not aware of GPU smuggling.

00:29:01
Speaker 2: But wouldn't they don't these GPUs have some mechanism of phoning home?

00:29:05
Speaker 1: Oh for sure they can be. They can be identified through the internet one hundred percent. I mean, yeah, they have. The firmware has identifying information in it for the GPUs. You don't need the serial number on the sticker to know what it is. You know, you don't have to physically see it to know what it is. And so yeah, if this is like connecting to some kind of server somewhere or whatever it may be doing, if in any way that's linked back to end Video, they would know about it. If there's ever a service request, I would assume at some point someone figures out and maybe it's a okay, well kind of you know, blind eye approach, But it's just there's I don't know. I really think there's no way that they were ignorant to this because, to quote one of the guys we spoke with when we did the Black Market GPU Smuggling documentary last year, the professor of a in Hong Kong. We spoke with. I asked him, I was like, do you think Nyvidyo knows? And yeah. His reaction was, well, how could they not? These are very expensive jeep us. You would think you would track them all extremely carefully.

00:30:12
Speaker 2: Yeah, I just And also it's the co founder of super Micro. And just to be clear, Wally Lear, the co founder, resigned in twenty eighteen, rejoined as a consultant a few years later, and then was just added back to the board. Yeah, and now he is in jail. And you're telling me that Charles Lang had no idea, no idea, what, no one an inventor, where where they think they were going? But this is like, this is yet another story that is just ignored. And I'm trying to work out if it is a journalistic competence issue or whether it's just a don't don't fuck without out Disneyland problem.

00:30:51
Speaker 1: I think there's yeah much there there. So, first of all, Land Video has a lot of money in different media publications, and then also the ones that are how do you mean, yeah, in the media publications that are involved in like stock coverage, they're all going to be biased towards protecting en Video In someway or another for the most part. But secondly, also I do think there's probably an issue of what a super micro for a lot of the reporters. And yeah, I.

00:31:20
Speaker 2: Just there's a picture of Wallie lear was at he was actually at GtC. There's a picture of him with an ice luge. There's a picture of him standing three people over from Jensen Huang. There's you can like Lambda, an AI compute company. He was. He was commenting on their LinkedIn very recently, being like super congress all. He's like clearly a dorky old man because he's like super congratulations, super great. I kind of I kind of love that, and I kind of loved the text he sent where somebody basically said you're going to jail and he just was spotted with three crying. Honestly, it was.

00:31:57
Speaker 1: Sad, but it was fine.

00:31:58
Speaker 2: Could be if I would do the same thing, I would be responding the same way if I was going to jail for like the most obvious crime. But it's so weird, like this super micro is the backbone of coal Wave like that.

00:32:15
Speaker 1: Also wonder if so some behind the scenes is after our black Market documentary last year, I went to Washington d C. And met with a number of congressional staff on one of the committees there, and it was they One of the things they told me was that they had some stuff in the works. There was some talk about, you know, maybe testifying or something, but they had stuff in the works and they were hoping they would materialize earlier this year. I didn't know what that was at the time, but I know they said it was related to the GP smuggling. They had a bunch of questions about it and how it works and whatever, and I think this is probably what they were working on. I think it was probably only the super Micro keys, I guess.

00:33:02
Speaker 2: But just the fact that super Micro is just walking around, still doing business and they weren't named in the indictment.

00:33:09
Speaker 1: Yeah, it's they weren't a defendant, right, it.

00:33:11
Speaker 2: Was the three ye Yeah, And it's just I don't know that doesn't see it doesn't seem very likely that nobody knew. But you know, it was just a few bad apples, right, just the co founder of the company. Yeah, just one guy. Couldn't be I just feel like by the end of this, we're going to have a few of those. I don't think you've got everything that you could possibly want from a scam For a scammer, you've got a vague technology that could do everything but doesn't really have to do anything, and a shit ton of construction.

00:33:44
Speaker 1: Just an absolute awes funny point on the construction, it's it's easy to hide money there.

00:33:49
Speaker 2: And just also let money just like marinate. The one thing that I can't find any way of like structuring this into an argument I could point out with data, but well, I've never seen a building project come in under budget, not once in my life. And you're telling me none of these companies need follow on funding. Yeah, none.

00:34:12
Speaker 1: I think the I think the construction alone is I don't know, it's it's all part of this interesting like the meme that's gone around of buying gps that don't exist yet to put in data centers that aren't built, you know, or it's yeah, the amount of promise, especially open AI versus actualized is there's a huge disparity there.

00:34:38
Speaker 2: My favorite thing is them claiming that the more compute they get, the more revenue they make, because that doesn't make sense at all.

00:34:45
Speaker 1: I mean, Jensen said that in one of the interviews where I think he went as high as ten x. He was like, if they had ten x were and they would have ten x the revenues, Like, I don't think so.

00:35:08
Speaker 2: Well. The thing with that I always say is where is the current state of affairs if that's the case, Like, where are the people that cannot do something because there's not enough compute? What is the thing being held back? Who is holding it back? Like are they sitting there and they're saying when this comes online, X bazillion dollars will slosh out of it, Like what is the what is not well? Like what's going on?

00:35:34
Speaker 1: In the previous CNBC interview, I think it was where Altman and Juan are with Brockman, who is a dollar store Jensen, Juan with his leather jackets. They were talking about that one hundred billion dollar investment back at the time, and Altman, to answer your question, his angle for the CNBC or whoever was audience was if you could choose, if you could choose between free education for everybody or curing cancer, And he pitched it as that those are your choices, but if they only had more compute, they could do.

00:36:09
Speaker 2: Both right as opposed to neither right, correct as opposed to nothing of the sort. Happening at all anywhere.

00:36:20
Speaker 1: But I do like that. That's it's I don't like the examples he chose. They're just so intentionally manipulative. I think you know where it's like, well, you could you could choose from preventing puppies from being kicked or from preventing cats from being homeless, just.

00:36:37
Speaker 2: Like you can only crush one animal. Like it's just very It's also the kind of thing that you come up with if you do not do any work, if you don't really know what happens in the real world, You're like, fuck, do yeah, what if education was free or what if cancer note no happen? Do you like this? Like?

00:36:58
Speaker 1: And of course CNBC is like, yep, can I have money now?

00:37:01
Speaker 2: Yeah? This is good? And that one hundred billion is very real that that's enough. This is the thing maybe conspiracy theorists right all along in the sense that it's like, yeah, there was a point last year where in video an open ai light one hundred billion dollar deal never happened. AMD and the open ai Light as well, just nobody talked about their guidance didn't change for a MD. They do not appear to have sold a single chip to open Ai Broadcom still nothing like three quarters in we don't have a single sale. What the fuck.

00:37:34
Speaker 1: Johnson was really defensive about that one. When the Taiwanese media asked, oh yeah, yeah, oh he there was never any commitment.

00:37:41
Speaker 2: He said, there's never got any commitment from any of these things, though. I just I think and I think at some point we're just going to find out that nothing happened, like I just did, nothing actually moved to anyone, Like we still haven't heard of in Vidia has sent a dime to open Ai. Soft Bank has to raise forty billion dollars to send thirty billion dollars to open Aye. Yeah, that hasn't happened yet.

00:38:06
Speaker 1: You see the game. So there's a game that came out this past week called Crimson Desert, and.

00:38:12
Speaker 2: I've been meaning to play it, but I hear it's some mess.

00:38:15
Speaker 1: Yeah, it is a It is a bit of a mess, I think. But the interesting thing was they had a bit of an unexpected scandal where amongst all the other various problems like blocking Intel from working, they had some users noticed that there was very clearly AI generated art in the Yeah hell yeah, yeah. So it's like you walk into the whatever the tavern and on the wall there's a painting handing and if you get close to it you can see it's like the early several years ago generative AI art. That's just fairy obvious, and so on Steam, you're supposed to disclose if there's AI art or assets in the game. They didn't do that. They're too big, so they're not gonna be punished for it. But they put out a statement apologizing saying these replace holders. They're supposed to be replaced. We're going back through and removing them all whatever. And the Mike Evara, who's currently the CEO of Prize Picks and previously was president of Blizzard and previously was also a VP of Microsoft Windows, he tweeted out asking why the studio was apologizing and basically saying this is going to be in everything, so you shouldn't apologize. And it's just it's crazy to me where the approach to a lot of the AI stuff. It feels so like this is inevitable, is what you're being told by the suits and the vps and the CEOs. I don't think they realize that. That's the part a lot of people don't like, is the part where it's being forced and you're told this is inevitable. Get used to it.

00:39:52
Speaker 2: Yeah, it doesn't have to be good or you don't have to like it, but you're going to have it because of us. We're gonna make you have it and that'll be good. Well it won't be good, but you'll have it, and we will. Ostensibly something will change. Like they're not even making I don't even think they're saving money doing this stuff. For if they are, it's the it's like very it's just firing three people, which is a lot for the people being fired, and their work would make the game better. But it's just a fucking excel spreadsy edge, right like it's for the for the larger companies. I mean, it's just so I think that this is also a mask off moment for so many of these companies. It's the ones that just don't give a rap fuck about anyone who just they don't make stuff, they don't contribute things. They just like, yeah, you know, man, you know the thing. You're like, it's gonna suck shit and we're gonna save like eight eight dollars.

00:40:46
Speaker 1: Yeah. No, it definitely enables drift in, right, Like the whole it's like you said earlier, you have the just the entire premise of the way the technology is set up and pitched and the lack of understand and technically for a lot of like media and coverage, it just it's set up where people can grift. And I think you saw that with crypto too, where there's a lot of blockchain, you know, is the big word everyone's thrown around suddenly all these people on TV who had no idea what that meant, but same type of thing, and it's I don't know the situation's right where maybe there's some real use cases for sure, but it's pretty easy to scam people right now.

00:41:29
Speaker 2: I just I also think that it's just showing how many dunces there are, how many people that don't actually want to learn anything. I mean, even the coding side. The little code I've learned so far has just made me aware that there are lots of engineers who don't read any of the code that comes out and they're like, it'll be fine, shit work fine.

00:41:48
Speaker 1: I think I had someone really pushing me to try claude where he's basically like no, no, like you don't the other ones. Because I've I've tested a few of the LM's and I'm not too impressed, but he was like, this one, this is the one, you know, that's unbelievable how good it is. And I guess my use case wasn't maybe the right fit or something.

00:42:13
Speaker 2: Yeah, it's just like.

00:42:19
Speaker 1: Like I tried it, and the thing I tried it for was I wanted to make a template for Adobe in Design, which I've never used, and you can go buy and download a template and whatever, but I wanted like a specific one made in the way I wanted to use it. And because in Design can accept code I might be XML, I don't remember, but it's got some kind of code it can accept, I thought, Okay, this should be able to write the code. It's not. It's not a design I'm asking for. It's code, and then I should be able to shove it into in Design and have a template. And I experimented with it for probably I got like really obsessed with when I'm trying to make software work. I spent like eight hours on it. Jeeuz it did this the same thing as chat GVT, where you know it gives you code. It didn't work. I feedback the error to it, and it's like, oh, I see what went wrong. It was this, and then it sends you more code and it doesn't work, and before you know, you've blown through all your tokens for the day, right, which seems like a great way to make more money. And I ended up after that experiment, I was like, let me just do this the old fashioned way, and I watched like three tutorials and within two and a half hours I had done what I wanted myself.

00:43:32
Speaker 2: You know, the model did not provide you with the thing you needed.

00:43:36
Speaker 1: It did not. And not only that, but there was a moment there where I was starting to feel that that disconnect with that like psychosis I think some people get as they use these tools more and more and rely on them, where it starts to feel like, well, I can't do this and this thing doesn't help in me. So there's that temptation of like I'll just give up, But then you know, you try to do it yourself for like five minutes, like, oh, this isn't that bad to learn new software. You know, I can handle this, And I just worry the people who get really reliant on this stuff. I don't know, it seems not good.

00:44:13
Speaker 2: Well that's the thing though, whenever I've because I have the times have experimented with like Claude and chat GPT, I've always thrown it stuff where I'm like, I've already finished a model, I've already finished at a paraff what do you think? And it will say something. I'll be like, okay, but what about this, and I'll just make up something. It's like, actually, there were eighty gigawatts added in the last quarter, what do you think about that? I'll be like, oh, my mistake.

00:44:37
Speaker 1: It's just like you're gaslighting.

00:44:38
Speaker 2: Now, I'm gaslighting you, like I am abusing the model. But that's the thing.

00:44:42
Speaker 1: You better be careful. That's gonna come back to you one day.

00:44:45
Speaker 2: Yeah, well, well Rococo's modern basilisk. Yeah. It's just it's just like the fact that they can just have their wrong twisted immediately makes them useless, because the whole point of artificial intelligence is it's meant to be in intelligent and kind of support you when you're like, not agree with you when you're wrong, but be like, hey, you're wrong here. But it's just very weird as well that every time I hear a story it's like yours, where it's like, yeah, I wanted it to do stuff, but it didn't. And do you talk to someone who's a booster and they go, well, what you should have done? It's like, so, so you're just gonna just let's start from the point of this thing doesn't work and you have to find a way to make it work. Well, it does work, you're just not using it right. Well why not? Yeah, well how did you prompt it? How did you exactly? But the thing is, you can hold something wrong. There's no way to use these right. It's just how how much can they gaslight you into believing that you're deficient and that you should say I think it work.

00:45:48
Speaker 1: It just seems like, at least for the task I was doing, the amount of time required to babystep it through the process is so much greater than just learning something new.

00:46:01
Speaker 2: Actually, I also don't know about you, but when I don't understand something, the more I don't understand it, the more worried I get, Like, I don't I can't sacrifice myself to a machine. I'll be like I will straight up just think like I don't get this this. I'm worried that I'm going to learn something wrong because I'm looking at this and it seems to be trying to gas me up when I do not understand what I'm talking. Well, I'm just like, I don't fucking know what you're talking about. Brother.

00:46:29
Speaker 1: You had a great quote when we met up when you came down here, and I've actually said it to writers on the team for training. But you know, one of the things I made a comment about how I needed to do a bunch of fact checking that day or something, and you made a comment about how you always double check your facts and if you want it to be right, you check it a third time.

00:46:52
Speaker 2: Yeah, which is brilliant.

00:46:54
Speaker 1: I think if you add AI to that, you know, it's like it's kind of that same feeling where it's like if anything comes out of an llm uh, the amount of anxiety I have mm hmmm, yep is not even worth whatever the answer is.

00:47:09
Speaker 2: And also, I so a while ago I had access to like a much bigger, more financial focused one and I used it to like look up a bunch of stuff, like do a report on this. Three or four of the links were for were just dead. One of them referred to something that just was an irrelevant link. I did look something up on Claude once and it gave me two different citations from Grokipedia. Are you you're not familiar with Grokipedia. I just well, that's the special groc Wikipedia that okay at xtually everything app.

00:47:48
Speaker 1: Ah, they have their own Wikipedia.

00:47:50
Speaker 2: Yep, and that's what Claude is citing from. There's the basic Wikipedia. It's so good. It's just I do think that there is also a degree here where AI is just a test of how much you're how willing you are to just be like, yeah, I get it. I guess like just quickly you can wave off something and be like I have yes, fine, I guess I'll mostly get it. This sounded good enough for me.

00:48:15
Speaker 1: The best use cases have still been for browsing Chinese Internet, or browsing the Internet generally in another language, and specifically it's for like when I've needed sources and research. So for the I don't know if I gave this example on your show before, but the one that came up previously was for the d ram Cartel Piece and for the Rise of Chinese Chinese Memory. We were looking at a bunch of different filings for these companies in China. I wasn't one hundred percent sure if they had like an SEC equivalent or something like that. And so using I don't know if I use I think I used GPT to try that. But using GPT or whatever it was at the time, it was useful for providing phrases where it's it's like, okay, here's the Chinese characters of what you should search for. And the reason Google Translate wasn't good enough on its own for that is because I didn't know the colloquialisms of what might they call these different things that I know exist in the US, but I don't know if they exist there, you know, And then I can take that and I can put it into Google Translate, which is a dumb translator that's just word for word, and just make sure it makes any sense and that it was useful for. But so far that's pretty much the only, you know, the only use case I've I've had.

00:49:31
Speaker 2: Yeah, and it's also it's the classic thing of maybe your own device model could do that. Yeah, I think that eventually that will be the case. And it's just is that really enough to substantiate nearly two hundred billion dollars worth of GPU and networking sales from Vidia a year. No dollars.

00:49:51
Speaker 1: Yeah, it's it's not enough to be worth the destruction of consumer hardware, you know. And that was constantly, like every time I try to poke at these things every now and then so that i'm at I want to make sure if I'm criticizing them, I've at least used them the same.

00:50:08
Speaker 2: Yeah, yeah, but it still feels bad. It's just and then you even then just having my Twitter feed up while we're talking, random ship pops up, such as SaaS CEOs. If your CFO hasn't used Lauter Excel by now, you need a new CFO. Only the curious will survive. Shut the fuck up. Just shut the fuck up, Like what are you talking about? Why? Why? What do you mean?

00:50:33
Speaker 1: That's so I feel like that's so dangerous, specifically for just responding with why feed all of your companies financial information to the cell?

00:50:44
Speaker 2: Also all what do you think? Like, what do you think a CFO does? Oh, they just they build models, I guess.

00:50:51
Speaker 3: Yeah.

00:50:51
Speaker 2: Well, the good idea for financial modeling is when you distance yourself from everything as far as but ideally you're so abstracted from the problem that you don't eat really understand it anymore. As we're finding out isn't a problem anywhere. It's just insane. It's I feel like we live in a truly insane period.

00:51:11
Speaker 1: And it's bad. I mean, the consumer space is just like last time I talked to you, similar topic, but more of the companies in the industry, computer industry that don't sell silicon products, you know, cases, cooler as, accessories, whatever, motherboards, we've had more of them now contacts basically saying hey, we're in big trouble and they want to talk to us about it for a video. But it's just when I say big trouble, I mean a lot of these companies are down like fifty plus percent from expectations, and that's because you can't build a computer as a consumer if you can't afford like five six hundred dollars minimum for basic memory now because of the data center demand for it.

00:51:57
Speaker 2: How does any of this turnaround?

00:51:59
Speaker 3: Though?

00:52:00
Speaker 2: How quickly? Is there any possibility that when the bubble busts anything returns to normal within gaming like you've seen I'm sure you've seen aras where they've oversold memory in the past.

00:52:10
Speaker 1: Yeah, not like this though, Yeah, I've never seen it like this. Memory is a really unique industry because of how it's been corrupt for forty years. And when I say maybe, well, yeah, So, like the Department of Justice under the Bush administration w busted the memory makers for what they called a cartel, and by definition that's what it was, price fixing cartel. And this has been happening like at least once a decade where in the late nineties early two thousands, the memory suppliers all got together and would agree on price ceilings and floors. So when they go and sell memory to Dell or Apple or HP, they've already spoken to their competitors and agreed, we're not going to go below this price.

00:53:06
Speaker 2: Cool.

00:53:07
Speaker 1: Yeah, and this has happened multiple times for that industry.

00:53:11
Speaker 2: Very cool.

00:53:12
Speaker 1: And you know what my favorite part of the story, and I didn't know this untill we did our report recently, was so there were there were several people that are like over one hundred people involved, I remember correctly, And if you look up the people who were charged with crime, including fined and even sentenced to prison for that for a price fixing cartel. If you look up what happened to them after a lot of them got promoted. Yep, one of them, his reward was to become CEO Jesus what company Samsung Europe?

00:53:47
Speaker 2: Very well, didn't someone from Samsung WANs to go to jail a few times? Like I swear to God like people from few times. Yeah, It's just it really does feel that by the end of this someone else is going to jail.

00:54:00
Speaker 1: There's just somewhat and they'll get out and they'll be promoted.

00:54:03
Speaker 2: Yeah. Well, the question is if it's clammy Sam Mortman though that I don't think they're gonna I don't know. I hear people being like someone was going to go to jail. I don't know if I buy it. But this private equity thing with the minimum returns maybe just to a classic Ponzi I hope.

00:54:18
Speaker 1: So yeah, classic Ponzi at least has some some history of meaningful busts and sentences, right, So, I don't know. But a lot of the white collar stuff, I feel like a lot of the time it's kind of a whatever you get. Like in the price fixing cartel, most of them got seven or eight months max.

00:54:36
Speaker 2: So I just America doesn't really have antitrust anymore, otherwise this would be I don't know. All all of this just feels like it's going to set back consumer hardware for a decade at this point.

00:54:50
Speaker 1: Oh yeah, yeah. And it's not just gaming either, like I think that's just what we specialize in. But yeah, it's it's more it's becoming increasingly viable I think for people on the surface to rent their computer or rent access to a computer in the cloud, and that's a bad thing.

00:55:08
Speaker 2: We don't want that. Is that really a likely outcome?

00:55:13
Speaker 1: I mean, the low end is drying up, and so if you can't really afford the cheapest, let's just keep it a gaming computer anymore. Your best option might be to go rent server access from Nvidia and use one of their thousand dollars GPUs for twenty dollars a month.

00:55:34
Speaker 2: God, what a fucking bleak, What a bleak environment. I just hope it. I hope it works out more that just these companies lose shit tons of money. I don't know if they because surely the memory prices would come down, in storage prices would come down as well, I would think, so, yeah, I would think. So. I think I have one involved with this should should suffer in ways I can't say on the show due to legal reasons, you know, just the thoughts that I have in sign in my head. All right, Steve, we're gonna wrap it there. You got anything you want to hype?

00:56:06
Speaker 1: Nah? I mean we've got a Yeah, I guess I'll throw one out there. We've got a big another documentary we're working on right now, but it's gonna be on the war on Huawei. So we went and dug through all of Huawei's history. It's pretty interesting and then interesting. Yeah, and that should be over an hour long and lots of They're a wild company and it's very politically intertwined with the US and China. So that'll be our next big one.

00:56:34
Speaker 2: Should be fun. And yeah, you know where to find me. The links will be in the show notes. Thank you as ever for listening to Better Offline. There will be a monologue this week. Love you all, goodbye.

00:56:51
Speaker 4: Thank you for listening to Better Offline. The editor and composer of the Better off Line theme song is Metosowski. You can check out more of his music and audio projects to selsk dot com, M A T T O s O W s KI dot com. You can email me at easy at better offline dot com or visit better Offline dot com to find more podcast links, and of course my newsletter. I also really recommend you go to chat dot where's youreaed dot at to visit the discord, and go to our slash.

00:57:19
Speaker 2: Better Offline to check out our reddit. Thank you so much for listening.

00:57:23
Speaker 3: Better Offline is a production of cool Zone Media. For more from cool Zone Media, visit our website cool Zonemedia dot com, or check us out on the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts.