WEBVTT - AI Versus The PC Industry With GamersNexus 0:00:02.800 --> 0:00:03.560 Zone Media. 0:00:05.440 --> 0:00:08.799 Hello and welcome to bear Offline. We are not a car. 0:00:09.280 --> 0:00:11.719 This is a podcast, not a car, and I am 0:00:11.840 --> 0:00:27.120 ed Zetra. Today we have returning champions Steve Burke of 0:00:27.200 --> 0:00:29.520 Gamers Nexus joining us to talk about a little company 0:00:29.520 --> 0:00:32.720 called Video. Hey doing Steve, I'm also not a car. 0:00:33.120 --> 0:00:36.040 Good stuff. I was just making sure, just checking are 0:00:36.080 --> 0:00:39.960 you triggered as well, because based on the Stanford video, 0:00:40.800 --> 0:00:43.919 there's an interview with Jensen woggot Stanford. He appears to 0:00:43.960 --> 0:00:46.640 have been both triggered but also not mad at the 0:00:46.680 --> 0:00:47.240 same time. 0:00:47.880 --> 0:00:54.360 Yeah, well, I mean he definitely seemed a little sensitive 0:00:54.600 --> 0:00:55.280 in that one. 0:00:55.360 --> 0:00:57.800 It's interesting, So just to be clear to the audience, 0:00:57.840 --> 0:01:00.480 there was an interview about a month ago or two 0:01:00.520 --> 0:01:03.560 ago with Jensen Huang on the Dwarkesh podcast, which is 0:01:03.600 --> 0:01:07.880 a podcast that's both very fluffy but also occasionally has 0:01:07.959 --> 0:01:11.480 sentient thoughts, and there was this bizarre bit where I 0:01:11.880 --> 0:01:14.360 forget what the question leading in was, It's something about 0:01:14.440 --> 0:01:17.080 China perhaps, and Jensen Wang just gets pissed on goes 0:01:17.200 --> 0:01:19.440 we are not a car. We are not a car, 0:01:19.720 --> 0:01:21.319 and it's just he. 0:01:21.800 --> 0:01:25.040 Also, I think that was the same podcast right where 0:01:25.040 --> 0:01:28.720 he was like, I'm not a loser. America is not 0:01:28.760 --> 0:01:31.880 a loser. We're not losers like he wound this. 0:01:31.880 --> 0:01:33.760 This is I say this as a I was a 0:01:33.840 --> 0:01:36.280 very fat child and I was bullied a great deal. 0:01:36.520 --> 0:01:38.800 Let me tell you something. When someone's saying they're not 0:01:39.000 --> 0:01:43.039 upset that they're not a loser, they are like they 0:01:43.440 --> 0:01:46.680 they are, just that's them saying it. Except I was. 0:01:47.440 --> 0:01:52.000 I was just large. I wasn't a five trillion dollar company. 0:01:52.120 --> 0:01:55.040 It really feels like Jensen is starting to go off 0:01:55.080 --> 0:01:55.880 the wheels a bit. 0:01:56.440 --> 0:01:59.160 Yeah, And I don't know if it's because it could 0:01:59.200 --> 0:02:00.960 I guess it could be all the things. It could 0:02:00.960 --> 0:02:04.360 be he's gotten to a size where he just doesn't 0:02:04.360 --> 0:02:07.160 care anymore. Maybe he is so important he's just dropped 0:02:07.240 --> 0:02:08.840 like that's just how he is, and he's kind of 0:02:08.880 --> 0:02:12.799 dropped the actor, or you know, maybe it's a top 0:02:12.880 --> 0:02:13.920 endicit or I don't know. 0:02:14.320 --> 0:02:16.720 I think it's more likely that because here's the thing. 0:02:16.800 --> 0:02:20.840 If we're talking about this. A few episodes back in 0:02:20.880 --> 0:02:24.360 twenty twenty five, he was signing Woman's Boobs. He was 0:02:24.720 --> 0:02:27.240 he was all swaggy. He'd go out on stage at 0:02:27.280 --> 0:02:29.600 GtC and just be like, the more you buy, the 0:02:29.600 --> 0:02:32.840 more you say we're the best, and everyone was cheering 0:02:32.919 --> 0:02:35.840 for him and he was like a rock star. Now 0:02:35.919 --> 0:02:39.760 he seems like Boakeshpottel is twenty three years old. If 0:02:39.760 --> 0:02:41.760 a twenty three year old was it was rude to me. 0:02:41.800 --> 0:02:44.640 Other than the numerous times they are online, I would 0:02:44.639 --> 0:02:47.960 I would just be like, Okay, it's not that not 0:02:48.000 --> 0:02:48.800 that big a deal. 0:02:48.880 --> 0:02:52.600 Especially if your companies worth trillions of dollars. 0:02:52.160 --> 0:02:54.360 And the quote is just a quote. We are not 0:02:54.800 --> 0:02:56.760 we are not a car. We are not a car. 0:02:56.880 --> 0:02:58.680 The fact that I can buy this car brand one 0:02:58.760 --> 0:03:01.000 day and use another car brand the other day easy 0:03:01.200 --> 0:03:04.280 computing is not like that. There's a reason why why 0:03:04.360 --> 0:03:06.919 the XA six deal exists. There's a reason why ARM 0:03:07.000 --> 0:03:10.880 is so sticky. These ecosystems are hard to replace. Oh no, 0:03:11.560 --> 0:03:12.920 if that, I mean, like. 0:03:13.000 --> 0:03:14.480 It's not really a clean analogy. 0:03:15.400 --> 0:03:18.560 It's also the kind of thing. It's also insane that 0:03:18.560 --> 0:03:20.880 they've got to this point without having a cleaner one 0:03:20.919 --> 0:03:21.200 than that. 0:03:22.639 --> 0:03:25.880 Yeah, I mean, it's that's easy to poke holes. And 0:03:25.919 --> 0:03:27.360 I get what he's trying to say. I get it. 0:03:27.400 --> 0:03:30.440 Like the architecture is, yes, it's difficult to move off of. 0:03:30.520 --> 0:03:34.880 But also on the consumer side, which is what he's 0:03:34.960 --> 0:03:37.800 talking about with buying a car and then buying a 0:03:37.800 --> 0:03:43.920 different one, you can in fact do exactly that with GPUs. 0:03:44.240 --> 0:03:46.040 So like, yeah, I get what. 0:03:46.000 --> 0:03:50.960 He means more on the business side, but it is uh, 0:03:52.880 --> 0:04:00.480 I don't It's just such a sensitive topic for him lately, 0:04:00.520 --> 0:04:02.880 and I don't understand if that's coming from a place 0:04:02.880 --> 0:04:06.560 of insecurity or I don't know. Just at this point, 0:04:06.880 --> 0:04:09.320 you know who you need to have on It's not me. 0:04:09.560 --> 0:04:15.120 It's like a therapist, Toshy analyze. He's like a therapist 0:04:15.400 --> 0:04:16.880 to break down his mental state. 0:04:17.000 --> 0:04:19.760 A director's commentary style, thinking like well, okay, stop there. 0:04:19.800 --> 0:04:21.400 You see when he says he has a car, this 0:04:21.480 --> 0:04:24.200 is a sign that just like in here, he's he's 0:04:24.200 --> 0:04:27.800 gonna cry. But they paused there. I So my theory 0:04:27.920 --> 0:04:29.640 is and we kin, we kind of talked about this 0:04:29.720 --> 0:04:31.200 and then we're not dipped into it a tom But 0:04:31.600 --> 0:04:33.920 there was a short cell report from Corporate Research that 0:04:33.960 --> 0:04:37.360 came out that suggested that over twenty percent of Nvidia's 0:04:37.920 --> 0:04:41.280 fiscal year twenty twenty six revenue came from China, right 0:04:41.360 --> 0:04:44.640 through a company called Megaspeed. I think what it might 0:04:44.760 --> 0:04:49.000 be is that he really does need China. I think 0:04:49.040 --> 0:04:52.080 it may be as simple as he really needs China. 0:04:52.160 --> 0:04:57.200 He needs China so bad, and Chinese GPUs terrify him 0:04:57.400 --> 0:05:00.920 in a way that that he's like the he's perhaps 0:05:00.960 --> 0:05:04.640 not willing to willing to totally ye talk about. And 0:05:04.720 --> 0:05:08.440 also he's not scared of am D. I don't think them. 0:05:08.480 --> 0:05:12.120 I instincts are gonna get a scare and video at 0:05:12.160 --> 0:05:13.400 all that would require. 0:05:13.440 --> 0:05:16.440 I don't think so. I Also, I think a m 0:05:16.520 --> 0:05:20.359 D is a known quantity at this point for Johnson, Yeah, 0:05:20.400 --> 0:05:25.560 and their strategies are known, their technology is known. It's 0:05:25.600 --> 0:05:31.200 really it's the the unknown technologies and strategy is probably 0:05:31.200 --> 0:05:32.760 what he needs to worry about. And right now, most 0:05:32.760 --> 0:05:36.360 of that's coming from Chinese companies. There's there's a lot 0:05:36.400 --> 0:05:40.839 of them. I we have some coverage coming up on Yeah, 0:05:41.000 --> 0:05:44.080 on Chinese GPUs, the rise of them kind of like 0:05:44.080 --> 0:05:50.080 our rise of Chinese memory video and they get It's interesting. 0:05:50.120 --> 0:05:53.880 It's interesting to watch companies like Jaoshan, which makes CPUs, 0:05:54.360 --> 0:05:57.120 come to market and when they do in the consumer space, 0:05:57.680 --> 0:06:00.720 they get kind of ship on for being a couple 0:06:00.800 --> 0:06:02.880 generations behind, because I remember when some of their first 0:06:02.880 --> 0:06:06.280 cps came out and the early media reports that tested 0:06:06.320 --> 0:06:08.800 it were, oh, they're about the same as a three 0:06:08.880 --> 0:06:14.400 or four generation old Intel CPU, And like I the 0:06:14.480 --> 0:06:20.000 perspective was one of mockery. But I think that should 0:06:20.080 --> 0:06:23.440 really concern some of these companies because like, if you 0:06:23.560 --> 0:06:28.960 come from nowhere, Silicon is not an easy technology. Architectures hard, 0:06:29.400 --> 0:06:30.960 and if you come out of nowhere and you're like 0:06:31.040 --> 0:06:35.000 four generations back instantaneously, that should be pretty scary, I 0:06:35.000 --> 0:06:38.320 think for the companies that are in the space, because 0:06:38.800 --> 0:06:41.920 it's it's the getting started that's the hard. 0:06:41.720 --> 0:06:44.560 Part, right What do you mean by that? What is 0:06:44.600 --> 0:06:46.239 the getting started in this case? 0:06:46.440 --> 0:06:49.440 The getting started in this case, at least in the 0:06:49.480 --> 0:06:52.240 scope of China, is largely going to be things like 0:06:52.279 --> 0:06:55.320 figuring out your fab. So who's going to make the product? 0:06:55.960 --> 0:06:57.800 Do they have the technology to make it? And if 0:06:57.800 --> 0:06:59.279 they don't, how do you kind of work within the 0:06:59.320 --> 0:07:01.000 limitations of what you have access to? 0:07:01.560 --> 0:07:01.680 Right? 0:07:02.200 --> 0:07:02.360 Right? 0:07:02.440 --> 0:07:07.839 Because technically TSMC is not supposed to make products for 0:07:07.920 --> 0:07:11.000 like Chinese GPUs or I mean sometimes it kind of 0:07:11.000 --> 0:07:15.320 happens anyway, but in the past, but there's limitations of 0:07:15.320 --> 0:07:18.600 the fab that's a big challenge for them. And then 0:07:19.280 --> 0:07:25.760 there's also just the actual first coming up with an 0:07:25.920 --> 0:07:30.120 architectural design. Once you have that, you can iterate on 0:07:30.200 --> 0:07:34.240 it and make changes that advance it. But getting anything 0:07:34.320 --> 0:07:37.520 to work at all, that's that's the hard part, right. 0:07:37.560 --> 0:07:40.200 So once they pass that barrier and they have a 0:07:40.240 --> 0:07:43.040 fab that can make it from there, you just iterate. 0:07:43.120 --> 0:07:45.200 And as long as they have some customers, they yeah, 0:07:45.200 --> 0:07:47.960 they can like any other company, they just iterate. And 0:07:48.600 --> 0:07:52.320 if China has a built in pool of customers because 0:07:52.320 --> 0:07:56.880 they have restricted access to external hardware, then whether or 0:07:56.880 --> 0:08:01.920 not they're competitive, they're they're gonna have buy because yeah. 0:08:02.000 --> 0:08:03.640 Well the other thing is as well, and this is 0:08:04.000 --> 0:08:06.720 one of my own personal things is I also have 0:08:06.760 --> 0:08:09.960 to wonder if there's not this whole thing I've been 0:08:10.000 --> 0:08:12.160 said about data center's not getting built. And when I 0:08:12.200 --> 0:08:14.440 say that, because I always get one comment from someone saying, 0:08:14.520 --> 0:08:18.040 actually I just saw one being built near me, and uh, 0:08:18.600 --> 0:08:21.800 people just with I assume no object permanence, which like 0:08:21.920 --> 0:08:24.800 I sought one, even though I'm saying they're not getting finished. 0:08:25.360 --> 0:08:28.560 If that's the case, I imagine that's quite scary to get. 0:08:28.600 --> 0:08:31.040 Maybe this isn't it, but I imagine that's quite scary 0:08:31.040 --> 0:08:34.480 to Jensen as well, because if these things aren't getting 0:08:34.600 --> 0:08:38.040 built and China is saying we don't need these GPUs, 0:08:38.360 --> 0:08:43.280 then that makes vera Rubin a much weirder proposition as well, 0:08:43.320 --> 0:08:45.720 because it's like, Okay, if China doesn't need these and 0:08:46.080 --> 0:08:47.800 we don't have the space for the new ones, where 0:08:47.800 --> 0:08:50.240 would we buy the new ones or what would we 0:08:50.280 --> 0:08:52.040 do with these old ones? Do we just junk them? 0:08:52.920 --> 0:08:54.719 Is the walls kind of closing infamous to one? 0:08:55.480 --> 0:08:59.680 Yeah? I think so, I will say, though on the 0:08:59.679 --> 0:09:01.720 premier of death and are not being built. I'm not 0:09:01.720 --> 0:09:04.720 sure if you saw, but Oracle has a very nice 0:09:04.880 --> 0:09:07.240 beam in the ground, several of them. 0:09:07.520 --> 0:09:10.960 I love that. There's one photo I forget which data 0:09:11.000 --> 0:09:13.680 center is. There's just they put up the first steel 0:09:13.720 --> 0:09:16.320 beam I think in Wisconsin and there's one bloke with 0:09:16.400 --> 0:09:19.160 his hand on it and three others walking up. Yeah, 0:09:19.200 --> 0:09:23.120 And it's just it's so good because it's like, yeah, yeah, 0:09:23.440 --> 0:09:25.680 I've seen this in London a hundred times, like a 0:09:25.679 --> 0:09:28.760 bunch of guys with hard hats going or my yellow 0:09:28.880 --> 0:09:31.560 to it's going to take a minute or is it stable? 0:09:31.559 --> 0:09:34.200 We better lean on it another minute, mate. It's so 0:09:34.679 --> 0:09:37.200 it is funny watching that happen. But it is also 0:09:38.320 --> 0:09:40.880 so very weird that going back to the Stanford thing 0:09:40.920 --> 0:09:43.559 as well, at that moment where it's like, oh, thank 0:09:43.600 --> 0:09:48.320 you for triggering me again. These are children, These are children. 0:09:48.360 --> 0:09:50.560 These are children with questions that you should be ready for. 0:09:50.600 --> 0:09:53.040 But these are children. How do you and your seventeen 0:09:53.120 --> 0:09:56.640 thousand dollars leather jacket not just reflect these unless there 0:09:56.640 --> 0:09:59.560 are like very simple problems. 0:09:59.080 --> 0:10:01.880 Here on that. That one was interesting too because that 0:10:02.360 --> 0:10:04.880 segment he gets into I'm not sure who the guy 0:10:05.000 --> 0:10:08.040 was sitting with him. I assume someone who works for Stanford. 0:10:09.240 --> 0:10:13.880 He's the guy okay, yeah, yeah, yeah right, the guy 0:10:13.920 --> 0:10:17.560 fielding the questions, though he makes a count at one 0:10:17.559 --> 0:10:19.400 point where he says, you know, we at Stanford have 0:10:19.480 --> 0:10:22.600 a forty billion dollar endowment, and he's talking about how 0:10:23.160 --> 0:10:27.920 they're having trouble getting access to compute and GPUs and 0:10:28.720 --> 0:10:32.120 Jensen effectively me paraphrasing it, but not that much. He 0:10:32.160 --> 0:10:36.120 effectively says, you just need money, like you, And he 0:10:36.160 --> 0:10:37.520 talks about how the first thing he would do is 0:10:37.559 --> 0:10:39.280 cut a billion dollars and send it off to a 0:10:39.320 --> 0:10:44.679 cloud data data center service provider for compute, as opposed 0:10:44.720 --> 0:10:48.440 to you know, buying and having their own solution in house, 0:10:49.040 --> 0:10:58.160 which that I think also is it's concerning because it's 0:10:58.360 --> 0:11:00.560 the thing we've been worried about in then ssumer space 0:11:00.559 --> 0:11:02.560 for a while, which is that you'll own nothing. And 0:11:02.679 --> 0:11:03.960 you know, the companies. 0:11:03.800 --> 0:11:06.959 Right right, they want to get away from providing anything 0:11:07.000 --> 0:11:08.760 other than dumb clients, right. 0:11:08.720 --> 0:11:12.160 Except this is we're not talking about a guy who's 0:11:12.200 --> 0:11:14.400 saying he can't get a GPU for video games. We're 0:11:14.400 --> 0:11:20.280 talking about like a university that's considering whatever, hundreds of 0:11:20.280 --> 0:11:23.960 millions or a billion dollars of spend and Jensen's telling 0:11:24.120 --> 0:11:27.959 them to fuck off and use laptops to connect to 0:11:28.000 --> 0:11:32.600 a cloud provider. And to me, it just it kind 0:11:32.600 --> 0:11:36.960 of seems like they're being a lot more open with 0:11:37.400 --> 0:11:44.720 the concept that owning the devices is dead and it's 0:11:44.760 --> 0:11:47.680 all going to cloud, and that is he was talking about. Well, 0:11:47.720 --> 0:11:49.280 you know, you see people today, they all have a 0:11:49.320 --> 0:11:52.640 laptop and they connect and the compute's not local anymore. 0:11:53.000 --> 0:11:56.120 That's what he's talking about, and that should scare a 0:11:56.120 --> 0:11:56.560 lot of people. 0:11:56.559 --> 0:11:59.559 How likely do you think that actually is though, because 0:12:00.320 --> 0:12:06.240 know it just with the speed of internet connections, I 0:12:06.480 --> 0:12:10.720 just don't know what like I don't know how like 0:12:10.840 --> 0:12:13.360 practically it would be. But no, these people don't really 0:12:13.400 --> 0:12:16.280 give a fuck, Like I think that Jensen Wong doesn't 0:12:16.280 --> 0:12:18.439 give a fuck about other countries outside of how many 0:12:18.480 --> 0:12:23.040 GPUs they buy. But it's like I was on an airplane. Yes, 0:12:23.080 --> 0:12:24.760 I was on an airplane. It's one hundred years old, 0:12:24.840 --> 0:12:26.960 and like I could barely connect to the internet if 0:12:27.000 --> 0:12:30.880 at all. That alone makes me feel like this might 0:12:30.920 --> 0:12:33.080 not be or is the suggestion that this would be 0:12:33.120 --> 0:12:35.400 like a very classist thing. This would be like if 0:12:35.400 --> 0:12:37.680 you the only laptop you can afford is one that 0:12:37.720 --> 0:12:41.560 is effectively just a device that connects to the cloud. 0:12:41.880 --> 0:12:45.880 I do think you start to see it cause segregation 0:12:46.320 --> 0:12:50.319 by class in terms of access to hardware, although that's 0:12:50.360 --> 0:12:56.160 maybe not unique to this industry, but the the likelihood 0:12:56.200 --> 0:12:58.880 just to pull a number g fours now, which is 0:12:58.920 --> 0:13:04.280 the low end online gaming service game streaming to basically 0:13:04.320 --> 0:13:07.600 any device. Yeah, they claim to have thirty million users 0:13:07.600 --> 0:13:10.520 for that hu which seems like a lot. 0:13:10.880 --> 0:13:17.959 That also really yeah this is g force now. 0:13:18.240 --> 0:13:20.840 Yeah, g Force now, there's a public document that they 0:13:21.040 --> 0:13:22.560 talk about thirty million users. 0:13:22.640 --> 0:13:27.520 Yeah, thirty million users. I don't know that that just 0:13:27.559 --> 0:13:30.800 feels very I'm just thinking of the PC gaming industry, 0:13:32.480 --> 0:13:35.440 like I for something of that scale, you would, I 0:13:35.600 --> 0:13:38.640 just maybe I'm just terribly ignorant. That just feels very 0:13:38.720 --> 0:13:42.720 high to me, because like every other game streaming service. 0:13:43.600 --> 0:13:46.439 I mean, they might be pumping their numbers because they 0:13:46.480 --> 0:13:50.199 do have an AD supported free tier. 0:13:50.840 --> 0:13:52.760 Well here's the thing. I just looked this up while 0:13:52.800 --> 0:13:55.880 on air. There are thirty five million subscribers to Xbox 0:13:55.920 --> 0:13:56.480 Game Pass. 0:13:57.320 --> 0:13:57.760 Yeah. 0:13:57.800 --> 0:14:01.640 I'll find it very unlikely that video which has failed 0:14:01.679 --> 0:14:06.400 to like they failed every single other like software like 0:14:07.160 --> 0:14:11.280 the omniverse, you know your for you and my favorite subject, 0:14:11.559 --> 0:14:14.640 the omniverse like a terrible face. Have they ever succeeded 0:14:14.679 --> 0:14:20.160 with any kind of external thing other than hardware? Yeah, 0:14:20.200 --> 0:14:22.520 exactly not. I just I don't believe that they have 0:14:23.440 --> 0:14:26.240 nearly as many, even with an AD support thing. 0:14:26.640 --> 0:14:32.600 So they have a PDF called Nvidio dash Story pdf 0:14:32.600 --> 0:14:39.000 our Story twenty twenty five and in here is where 0:14:39.040 --> 0:14:41.440 they I'll have to send it to you, and here's 0:14:41.440 --> 0:14:45.520 where they. Yeah, over thirty million members in more than 0:14:45.560 --> 0:14:48.320 one hundred and ten countries now have access to more 0:14:48.320 --> 0:14:49.400 than nineteen hundred. 0:14:49.440 --> 0:14:53.840 Ah, there you go, there's your thirty million members over 0:14:53.880 --> 0:14:59.760 a nonspecific period now have access could theoretically use so 0:14:59.800 --> 0:15:03.440 that thirty million people have signed up and given an 0:15:03.440 --> 0:15:06.320 email address. Yeah beautiful. 0:15:06.600 --> 0:15:10.320 Yeah, and so like how how likely that is? I'm 0:15:10.320 --> 0:15:13.080 not sure, but that that's an interesting story because I 0:15:13.120 --> 0:15:18.120 just learned thanks to the recent Force Now partner breach, right, Yeah, 0:15:18.480 --> 0:15:23.840 that Nvidia has partners to propagate this service kind of 0:15:23.880 --> 0:15:26.239 like some kind of like a Wi Fi range extender. 0:15:26.520 --> 0:15:28.480 Makes sense, That makes some sense. 0:15:28.760 --> 0:15:32.280 Yeah, And I guess that's to try and solve what 0:15:32.280 --> 0:15:36.920 you're talking about, which is the latency issue. I mean, 0:15:36.960 --> 0:15:40.760 the US is pretty willfully far behind right in a 0:15:40.760 --> 0:15:43.720 lot of places for Internet where it's just non viable. 0:15:55.760 --> 0:15:58.960 It's all just a very strange time. Actually, before I 0:15:59.040 --> 0:16:01.880 go any further though, what is this g Force breach 0:16:01.960 --> 0:16:04.080 you mentioned to me just before we came on the call, 0:16:04.080 --> 0:16:05.760 and I realized we should give it some an time. 0:16:06.600 --> 0:16:12.760 Yeah, it's so the hacking group Shiny Hunters, or someone 0:16:12.960 --> 0:16:16.480 at least posing as them. It's kind of the funny 0:16:16.480 --> 0:16:20.240 thing with the hacking groups. It's difficult to verify. Sometimes 0:16:20.280 --> 0:16:22.200 they impersonate each other and there's no real good way 0:16:22.200 --> 0:16:27.680 to verify it. But a user representing Shiny Hunters posted 0:16:27.880 --> 0:16:33.760 on a breach forum basically dark web stuff with an 0:16:33.760 --> 0:16:35.600 offer to sell I think it was for one hundred 0:16:35.640 --> 0:16:39.920 thousand dollars a data packet where they claimed that they 0:16:40.000 --> 0:16:46.720 had breached and harvested data including two factor authentication information, 0:16:47.160 --> 0:16:52.760 Jesus access flags, usernames, dates of birth, things like that, 0:16:53.800 --> 0:16:56.360 and they're offering it for sale for I think it 0:16:56.360 --> 0:16:58.400 was one hundred grand I'd have to check. And that 0:16:58.440 --> 0:17:01.160 was on a forum they said. They wrote it as 0:17:01.240 --> 0:17:04.359 if they had compromised all of Nvidia G Force now 0:17:04.840 --> 0:17:08.000 and Video responded within a couple of days and said, 0:17:09.040 --> 0:17:12.560 we've looked into this. Nope, what they've compromised is one 0:17:12.560 --> 0:17:14.919 of our partners representing the Armenia region. 0:17:15.800 --> 0:17:16.240 Uh. 0:17:16.560 --> 0:17:18.320 And then Video kind of pointed the finger to just 0:17:18.440 --> 0:17:22.320 Armenia for the breach, that the partner actually does service 0:17:22.400 --> 0:17:25.200 other countries as well and regions, so like, I think 0:17:25.240 --> 0:17:28.320 it's I want to take Kazakhstan and news Backstan. And 0:17:29.160 --> 0:17:31.479 they didn't talk about if any of them were breached, 0:17:31.480 --> 0:17:33.760 so it could actually be wirer than Nvideo is representing. 0:17:34.640 --> 0:17:37.320 But anyway, doesn't I find it hard to believe in 0:17:37.400 --> 0:17:39.440 Video would misrepresents something. 0:17:40.800 --> 0:17:44.800 Yeah, straightforward, Yeah, never any slipper rewording. 0:17:45.320 --> 0:17:51.760 But let's change tag to China, China, China, China, and 0:17:51.840 --> 0:17:55.240 this is so you've got you. We just what matched 0:17:55.520 --> 0:17:58.360 us up for this episode was that GPU smuggling thing, 0:17:58.400 --> 0:18:00.640 and we can talk about the super micro stuff, which 0:18:00.680 --> 0:18:04.040 I think rocks. I think that I've this whole time. 0:18:04.440 --> 0:18:07.000 All of these scumbags have been scummy, but in a 0:18:07.119 --> 0:18:11.960 very boring way. There's not been any goofy science fiction 0:18:12.440 --> 0:18:16.240 share or hair dryers and the fact that this that 0:18:16.359 --> 0:18:19.080 Wali Leao was going one of the blokes was going 0:18:19.160 --> 0:18:22.720 into the warehouses and just kind of hair drying off 0:18:22.800 --> 0:18:24.280 the cereal them. But I know we might have talked 0:18:24.280 --> 0:18:25.879 about this last time, but it was funny enough to 0:18:25.880 --> 0:18:28.720 go over again. I love it. I think it's hilarious. 0:18:29.040 --> 0:18:32.280 And there were a lot of interesting things in the 0:18:32.320 --> 0:18:36.200 timeline where a lot of the stuff happened in August 0:18:36.680 --> 0:18:41.320 and some of the investigation in August, and that was 0:18:41.359 --> 0:18:44.280 the same time we ran our report and we were 0:18:44.280 --> 0:18:46.800 digging into some of that stuff too, and a lot 0:18:46.880 --> 0:18:50.280 of it matched what we found where you know, during 0:18:50.280 --> 0:18:55.840 our Black Market GP report, we did talk to Taiwanese 0:18:55.920 --> 0:19:00.520 company and a company that operates also in Singapore, and 0:19:00.560 --> 0:19:02.880 one of the things we learned was that they'll act 0:19:02.960 --> 0:19:06.080 as intermediaries where this one company I couldn't show them 0:19:06.160 --> 0:19:08.719 or anything on camera back then, but we talked about it. Yeah, 0:19:08.960 --> 0:19:13.960 they work to bring in nvidio. I want to say 0:19:13.960 --> 0:19:17.119 it was GB two hundred, Maybee three hundred, GB two 0:19:17.200 --> 0:19:22.919 hundred systems, so like actual racks of hardware. Yeah, like 0:19:23.040 --> 0:19:26.680 actual you know, full racks, and they'll bring them into 0:19:26.720 --> 0:19:31.800 Taiwan and run testing. So it's a testing house. They 0:19:31.800 --> 0:19:36.160 do pre testing before it gets deployed. They do configuration 0:19:36.320 --> 0:19:39.440 stuff like that on behalf of other companies, not for themselves, 0:19:39.800 --> 0:19:43.480 and their clients were in China. I remember talking to 0:19:43.520 --> 0:19:46.080 them about it because there's just showing us these GPU 0:19:46.160 --> 0:19:50.000 racks and explaining that, oh, we test it and you know, 0:19:50.000 --> 0:19:52.360 it's for a company in China that we work with, 0:19:53.119 --> 0:19:57.359 and they seemed somewhat oblivious to the fact that this 0:19:57.520 --> 0:20:00.359 is not supposed to happen as they were giving us 0:20:00.359 --> 0:20:04.080 all this information. Was it, yeah, go on, sorry, Yeah, 0:20:04.119 --> 0:20:06.639 So I mean they basically test it, forward it and 0:20:07.640 --> 0:20:11.119 that was what we reported, and that's also exactly what 0:20:12.400 --> 0:20:16.520 back last year Jensen was regularly talking about all these things. 0:20:16.520 --> 0:20:19.200 They weighed thousands of pounds. It's not like some small 0:20:19.240 --> 0:20:22.520 potato chip you can put in your pocket. Yeah right, 0:20:22.800 --> 0:20:27.600 And it's a total misrepresentation to your earlier point. And 0:20:27.680 --> 0:20:30.679 as far as we could tell, like this, this is 0:20:30.720 --> 0:20:34.240 not it's not like they're just smuggling in GPUs and 0:20:34.280 --> 0:20:36.720 crates of lobster and shrimp, which has also happened, by 0:20:36.760 --> 0:20:41.840 the way, but actually completed server product. 0:20:41.920 --> 0:20:43.920 But the thing is, I didn't realize. I guess you 0:20:43.920 --> 0:20:45.840 couldn't have set back then. I didn't realize it was 0:20:45.880 --> 0:20:47.000 an MVL seventy two. 0:20:47.600 --> 0:20:49.159 I'm not sure if it was an MVL seventy two, 0:20:49.200 --> 0:20:51.000 which it was a two hundred rack, I don't know 0:20:51.040 --> 0:20:52.000 how many were in there. 0:20:52.080 --> 0:20:55.080 But if it was like a full rank, that's still good. 0:20:55.160 --> 0:20:57.520 That was last year as well. If they had GB 0:20:57.600 --> 0:21:00.159 two hundreds, yeah, I mean I should have really us 0:21:00.280 --> 0:21:01.760 at the time, but like, well, I don't know if 0:21:01.800 --> 0:21:04.639 you could have actually disclosed. But that's the thing, Like, 0:21:05.200 --> 0:21:08.800 because these various reports, the super Micro one, the Culpur report, 0:21:08.960 --> 0:21:10.520 they will come back to a thing which is in 0:21:10.640 --> 0:21:14.000 video frames. It as oh, nothing has got through or 0:21:14.040 --> 0:21:16.480 if anything did. It's just a few things. You couldn't 0:21:16.480 --> 0:21:19.000 get the big stuff through. But every report seems to 0:21:19.000 --> 0:21:22.280 be like, yeah, entire racks are making it over. Yeah, 0:21:22.320 --> 0:21:25.359 we've got and they have dummy servers, which is the 0:21:25.400 --> 0:21:27.520 craziest shit they have just like fake ones. 0:21:27.720 --> 0:21:29.439 That's fascinating to me. 0:21:29.960 --> 0:21:31.080 I wonder what's in them. 0:21:31.920 --> 0:21:36.800 It's yeah, it is, I guess, brilliant and like it's 0:21:36.800 --> 0:21:39.119 a very dumb solution. I don't mean that in a 0:21:39.160 --> 0:21:42.120 mean way, right, Like it's just how do we how 0:21:42.119 --> 0:21:45.080 do we trick the inspectors, because basically inspectors will come 0:21:45.119 --> 0:21:49.960 through to ensure compliance, make sure that people who handle 0:21:50.040 --> 0:21:54.760 large amounts of these restricted GPUs, just like I guess 0:21:54.760 --> 0:21:56.480 handling large amounts of I don't know, if you're in 0:21:56.480 --> 0:21:59.320 the pharmaceuticals industry, there's probably some level of inspection also, 0:22:00.080 --> 0:22:03.560 and they inspect it, make sure it's compliant, and sign 0:22:03.600 --> 0:22:05.560 off on it. And so you work with third parties 0:22:05.560 --> 0:22:11.600 for this process and to pass inspection. They had, like 0:22:11.640 --> 0:22:14.840 you said, used hair dryers to remove serial numbers and labels, 0:22:15.200 --> 0:22:19.119 and then they had staged dummy servers that were compliant 0:22:20.680 --> 0:22:24.480 to kind of disguise the real operation that's the day. 0:22:25.119 --> 0:22:27.400 So I guess it's only a physical inspection. 0:22:28.880 --> 0:22:31.000 Yeah, I'm not sure. I was wondering that too, because 0:22:31.000 --> 0:22:36.679 I did read the entire report, and unfortunately the DJ 0:22:36.920 --> 0:22:40.840 had screenshots of the warehouse, but they didn't. It was 0:22:40.880 --> 0:22:44.520 like screenshots of surveillance camera footage, but no actual surveillance 0:22:44.560 --> 0:22:48.000 camera footage I can find. So from everything we know, 0:22:48.040 --> 0:22:50.560 it looks like a physical inspection. I don't know that 0:22:50.640 --> 0:22:52.679