WEBVTT 00:00:02.180 --> 00:00:04.799 Welcome to the AI Chat Podcast. I'm your host, 00:00:04.860 --> 00:00:07.179 Jaden Schaefer. Every day I cover cutting -edge 00:00:07.179 --> 00:00:09.320 AI news and talk with the leaders behind it, 00:00:09.339 --> 00:00:11.300 breaking down what it means for your life and 00:00:11.300 --> 00:00:18.440 business. Anthropic is localizing Claude's pricing 00:00:18.440 --> 00:00:20.899 in India, which is its second largest market. 00:00:20.960 --> 00:00:22.920 Interesting because people there will be able 00:00:22.920 --> 00:00:25.660 to get it cheaper than people in America or other 00:00:25.660 --> 00:00:27.500 countries that make more money historically. 00:00:28.239 --> 00:00:30.600 Apple's opening I complaint, they're detailing 00:00:30.600 --> 00:00:34.299 a show and tell interviewing over 400 hires claiming 00:00:34.299 --> 00:00:37.320 that opening I is copying trade secrets. World 00:00:37.320 --> 00:00:40.539 models right now are getting billions and Lacan, 00:00:40.539 --> 00:00:43.619 Lee and Runway are all betting against LLMs. 00:00:43.700 --> 00:00:45.340 We're going to get into the disparity there. 00:00:45.899 --> 00:00:49.119 Meta is doubling their Hyperion data center to 00:00:49.119 --> 00:00:52.219 5 gigawatts. It was going to be 2 .5, or I thought 00:00:52.219 --> 00:00:53.840 it was maybe 2 at the beginning, but they're 00:00:53.840 --> 00:00:56.939 pushing the spending past $50 billion, increasing 00:00:56.939 --> 00:00:59.780 it there. Sam Altman and Elon Musk are having 00:00:59.780 --> 00:01:01.920 a ton of beef, which is kind of funny. We'll 00:01:01.920 --> 00:01:04.540 end the show on that. Sam Altman is saying that 00:01:04.540 --> 00:01:07.500 Elon Musk's space data centers are a short -term 00:01:07.500 --> 00:01:10.519 sales pitch, and Elon Musk is calling him Scam 00:01:10.519 --> 00:01:12.760 Altman. So there's lots to get into. ever since 00:01:12.760 --> 00:01:15.840 fable 5 came back to claude anthropic has been 00:01:15.840 --> 00:01:17.879 playing this weird game where they keep saying 00:01:17.879 --> 00:01:20.140 they have a new deadline for when claude fable 00:01:20.140 --> 00:01:22.739 5 is going to get pulled off of the subscription 00:01:22.739 --> 00:01:24.859 tiers and you'll only be able to access it through 00:01:24.859 --> 00:01:28.280 the api initially they extended it till the 12th 00:01:28.280 --> 00:01:30.680 and uh or i think the 7th was like the first 00:01:30.680 --> 00:01:31.959 deadline they're like it's gonna be off after 00:01:31.959 --> 00:01:34.319 the 7th and i remember using up all my tokens 00:01:34.319 --> 00:01:36.390 till the 7th And then they said, look, we're 00:01:36.390 --> 00:01:38.730 resetting it and you actually can use it till 00:01:38.730 --> 00:01:42.569 the 12th, which was yesterday. And so I used 00:01:42.569 --> 00:01:45.329 up all my tokens again on the 12th and my tokens 00:01:45.329 --> 00:01:47.409 don't reset on Claude till like Wednesday. So 00:01:47.409 --> 00:01:49.790 now I literally have no more Claude tokens, can't 00:01:49.790 --> 00:01:53.049 Claude for like three days on two separate Claude 00:01:53.049 --> 00:01:54.790 Max accounts, mind you. And then what happened 00:01:54.790 --> 00:01:57.030 20 hours ago, Claude came out with a tweet and 00:01:57.030 --> 00:01:58.950 said, we're extending Claude Fable 5 access to 00:01:58.950 --> 00:02:01.870 all paid plans, as well as keeping Claude codes 00:02:01.870 --> 00:02:05.040 weekly limit. rate limits to 50 % higher through 00:02:05.040 --> 00:02:08.699 July 19th. This is the second, third time, second 00:02:08.699 --> 00:02:11.780 time that they've extended this. It's super annoying. 00:02:11.960 --> 00:02:14.099 I replied to them on Twitter and said, except 00:02:14.099 --> 00:02:15.939 I already used up all my tokens in anticipation 00:02:15.939 --> 00:02:17.840 of the deadline, so I have no more credits left. 00:02:17.919 --> 00:02:19.810 Thanks for nothing. The sentiment seems to be 00:02:19.810 --> 00:02:22.129 felt by everyone. If you look at Omasar over 00:02:22.129 --> 00:02:24.169 on X, he said, I don't think Anthropic realizes 00:02:24.169 --> 00:02:26.270 how disruptive these changes are to users. I 00:02:26.270 --> 00:02:28.430 appreciate the extension, but please stop playing 00:02:28.430 --> 00:02:30.169 games. Either keep it under the subscription 00:02:30.169 --> 00:02:33.110 or put it under the API already. I 100 % agree. 00:02:33.270 --> 00:02:35.150 This is super annoying because you obviously, 00:02:35.289 --> 00:02:37.629 when someone makes a rule or says that there's 00:02:37.629 --> 00:02:39.229 a deadline, you're going to plan accordingly, 00:02:39.389 --> 00:02:41.750 aka use up your tokens ahead of time. Would I 00:02:41.750 --> 00:02:43.909 have been more frugal? Would I have spent them 00:02:43.909 --> 00:02:46.389 better? You know, if they were been there for 00:02:46.389 --> 00:02:48.789 two weeks? Yes, obviously. Now I'm in a pickle 00:02:48.789 --> 00:02:52.150 and lucky for me and for anyone else that might 00:02:52.150 --> 00:02:54.310 have had an occasion where you ran out, you hit 00:02:54.310 --> 00:02:56.629 some sort of subscription limit, token limit. 00:02:56.789 --> 00:02:58.810 I would love for you to check out my platform, 00:02:58.969 --> 00:03:01.969 AIbox .ai, because when you reach a token limit 00:03:01.969 --> 00:03:06.069 on OpenAI or Claude or any other platform, AIbox 00:03:06.069 --> 00:03:09.750 .ai for $8 .99 a month, $8 .99, you get access 00:03:09.750 --> 00:03:12.449 to over 80 different AI models and it's going 00:03:12.449 --> 00:03:16.229 to hold you over till your next hit of tokens 00:03:16.229 --> 00:03:19.080 comes back in. In addition, to be honest, you 00:03:19.080 --> 00:03:21.939 get access to image, audio, video, music, text, 00:03:22.000 --> 00:03:24.000 AI. So if you just want to have access to every 00:03:24.000 --> 00:03:26.340 single AI model anytime that you need it, go 00:03:26.340 --> 00:03:28.280 check out AIbox .ai. I'll leave a link in the 00:03:28.280 --> 00:03:30.560 description. In other news for Anthropic, they 00:03:30.560 --> 00:03:33.659 are now selling Claude in Indian rupees and they're 00:03:33.659 --> 00:03:36.639 pricing Claude Pro at 2000 rupees. That's about 00:03:36.639 --> 00:03:39.560 $21 a month. And their most powerful model, Claude 00:03:39.560 --> 00:03:44.819 Max at $125 a month American or about 12 ,000 00:03:44.819 --> 00:03:48.199 rupees. India accounts for about 5 .8 % of Claude's 00:03:48.199 --> 00:03:51.000 global usage. So it makes them basically the 00:03:51.000 --> 00:03:54.020 second most popular market after the United States, 00:03:54.159 --> 00:03:57.460 but they can't accept UPI, which is India's dominant 00:03:57.460 --> 00:03:59.500 payment method. So this makes it tricky. But 00:03:59.500 --> 00:04:01.699 what's really interesting is a lot of people 00:04:01.699 --> 00:04:03.400 thought and you know, I kind of mentioned in 00:04:03.400 --> 00:04:05.599 the intro of the podcast that when you localize, 00:04:05.620 --> 00:04:08.319 you typically will make it cheaper in the country 00:04:08.319 --> 00:04:11.319 by based off of what the income is of that country. 00:04:11.400 --> 00:04:13.800 So if India makes less money, you would make 00:04:13.800 --> 00:04:15.280 the product cheaper in India. This is what a 00:04:15.280 --> 00:04:17.740 lot of SaaS subscriptions do. And in this case, 00:04:17.740 --> 00:04:20.360 they did not do that. Anthropic is not heavily 00:04:20.360 --> 00:04:22.540 discounting the model for people in India. And 00:04:22.540 --> 00:04:25.019 in fact, it's actually more expensive because 00:04:25.019 --> 00:04:27.259 in the United States, the pro plan of Clod is 00:04:27.259 --> 00:04:30.199 $20 a month. And it's roughly the conversion 00:04:30.199 --> 00:04:33.459 rate bounces around, but it's roughly $21 if 00:04:33.459 --> 00:04:36.300 you're doing the conversion from Indian rupees 00:04:36.300 --> 00:04:39.519 right now. OpenAI added UPI support when they 00:04:39.519 --> 00:04:42.680 launched rupee pricing in August of last year. 00:04:42.740 --> 00:04:45.490 And Anthropic is still requiring credit cards 00:04:45.490 --> 00:04:49.050 or Apple or Google billing. It's a really big 00:04:49.050 --> 00:04:51.589 friction point in a market where UPI is basically 00:04:51.589 --> 00:04:54.230 the default checkout method in India. Anthropic 00:04:54.230 --> 00:04:57.689 opened a office in February of 2026 in India, 00:04:57.709 --> 00:04:59.709 and they signed some partnerships with Infosys 00:04:59.709 --> 00:05:02.389 and Tata Consulting Services to try to help reach 00:05:02.389 --> 00:05:04.449 Indian enterprises. But they're being criticized 00:05:04.449 --> 00:05:06.970 because they currently cost more than if you 00:05:06.970 --> 00:05:08.750 had tried to get the same service in the United 00:05:08.750 --> 00:05:12.529 States. Apple is filing a 41 -page lawsuit alleging 00:05:12.529 --> 00:05:14.829 that OpenAI has systematically recruited over 00:05:14.829 --> 00:05:17.490 400 Apple employees and extracted confidential 00:05:17.490 --> 00:05:20.709 hardware secrets from them, including by coaching 00:05:20.709 --> 00:05:23.470 candidates to bring actual device parts to job 00:05:23.470 --> 00:05:25.949 interviews and helping one engineer allegedly 00:05:25.949 --> 00:05:29.829 breach Apple's cloud storage after he left. The 00:05:29.829 --> 00:05:31.769 case that they're making right now is targeting 00:05:31.769 --> 00:05:34.550 OpenAI, their hardware subsidiary IO, which Apple 00:05:34.550 --> 00:05:38.149 acquired for $6 .5 billion and three current 00:05:38.149 --> 00:05:40.370 OpenAI executives. And if you guys remember, 00:05:40.649 --> 00:05:43.269 they have this huge deal going on right now. 00:05:43.560 --> 00:05:46.399 with Johnny Ive, who formerly helped design the 00:05:46.399 --> 00:05:48.800 iPhone over at Apple, and OpenAI is working on 00:05:48.800 --> 00:05:51.740 a device with him, so Apple is not happy about 00:05:51.740 --> 00:05:54.319 this. Tang Tan, who's OpenAI's chief hardware 00:05:54.319 --> 00:05:57.339 officer and a 24 -year Apple veteran, allegedly 00:05:57.339 --> 00:05:59.459 instructed job candidates to bring confidential 00:05:59.459 --> 00:06:02.819 parts like batteries and logic boards to show 00:06:02.819 --> 00:06:05.879 -and -tell interview sessions. Former Apple engineer 00:06:05.879 --> 00:06:09.120 Chang Liu downloaded dozens of confidential files 00:06:09.120 --> 00:06:13.680 weeks after leaving in January of 2026. by exploiting 00:06:13.680 --> 00:06:16.639 an authentication vulnerability and then texting 00:06:16.639 --> 00:06:19.120 a colleague, I still have another computer. Apple 00:06:19.120 --> 00:06:21.360 said that it first contacted OpeningEye about 00:06:21.360 --> 00:06:24.139 the alleged misconduct in February and received 00:06:24.139 --> 00:06:27.019 no response before they filed this lawsuit. OpeningEye's 00:06:27.019 --> 00:06:29.519 hardware device is expected to launch next year. 00:06:29.759 --> 00:06:32.220 So, you know, Apple's obviously getting a little 00:06:32.220 --> 00:06:34.300 concerned in the run -up to this. I think this 00:06:34.300 --> 00:06:36.720 is one of the most detailed evidences yet of 00:06:36.720 --> 00:06:39.699 what maybe you'd call corporate espionage and 00:06:39.699 --> 00:06:42.069 for the AI industry, what it looks like. This 00:06:42.069 --> 00:06:43.689 isn't, you know, a spy movie or something, but 00:06:43.689 --> 00:06:46.149 you're basically have a company opening systematically 00:06:46.149 --> 00:06:49.910 hiring specialized engineers from Apple. Now, 00:06:49.930 --> 00:06:51.649 this shouldn't be illegal. You should be able 00:06:51.649 --> 00:06:53.589 to hire people from Apple and even target Apple 00:06:53.589 --> 00:06:55.470 or another competitor if they're building something 00:06:55.470 --> 00:06:57.769 that you want and their employees have the domain 00:06:57.769 --> 00:06:59.649 expertise to do it. I mean, there's nothing wrong 00:06:59.649 --> 00:07:02.829 with that. But if Apple's allegations hold up, 00:07:02.930 --> 00:07:04.509 I think it's going to set a precedent for how 00:07:04.509 --> 00:07:06.490 a lot of companies protect knowledge from rivals 00:07:06.490 --> 00:07:09.370 when it comes to hardware. And, you know, obviously 00:07:09.370 --> 00:07:11.899 there's some there's some red line. to not cross, 00:07:12.019 --> 00:07:13.899 like telling people to bring actual components 00:07:13.899 --> 00:07:16.860 from the company or bringing their computer with 00:07:16.860 --> 00:07:18.639 data on it from the company. So that's all going 00:07:18.639 --> 00:07:21.470 to get litigated in court. Three different AI 00:07:21.470 --> 00:07:24.370 startups, World Labs, Advanced Machine Intelligence, 00:07:24.529 --> 00:07:28.029 and Runway have all raised a combined $2 .3 billion 00:07:28.029 --> 00:07:31.730 earlier this year. And the goal of these companies, 00:07:31.850 --> 00:07:34.170 both all three of them, is to build world models. 00:07:34.269 --> 00:07:35.910 So these are basically AI systems that are going 00:07:35.910 --> 00:07:38.750 to simulate physical environments instead of 00:07:38.750 --> 00:07:40.790 just generating text, right? LLM, ChatGPT generate 00:07:40.790 --> 00:07:42.689 some text for you. These are going to simulate 00:07:42.689 --> 00:07:44.410 real world environments. And there's a lot of 00:07:44.410 --> 00:07:46.149 interesting use cases. It's interesting that 00:07:46.149 --> 00:07:47.829 Runway's working on this because they need video 00:07:47.829 --> 00:07:50.480 and video you need kind of a simulated. 3D environment 00:07:50.480 --> 00:07:52.240 if you want it to be really powerful so you can 00:07:52.240 --> 00:07:53.600 tell the camera to move around from all different 00:07:53.600 --> 00:07:56.579 angles. But more importantly, I think, is the 00:07:56.579 --> 00:07:59.759 robotics era where this kind of falls into. And, 00:07:59.819 --> 00:08:01.959 you know, you need to be able to simulate these 00:08:01.959 --> 00:08:03.939 physical environments to make these simulated 00:08:03.939 --> 00:08:06.449 environments for robots to train them. Right 00:08:06.449 --> 00:08:08.050 now, the bet that all three of those companies 00:08:08.050 --> 00:08:10.670 are making, I think, is showing a huge split 00:08:10.670 --> 00:08:14.310 in the AI field over whether LLMs alone can lead 00:08:14.310 --> 00:08:17.649 smarter, more capable AI. World Labs and AMI 00:08:17.649 --> 00:08:20.110 each raised about a billion dollars. Runway added 00:08:20.110 --> 00:08:23.610 $315 million in February this year. Google, DeepMind, 00:08:23.670 --> 00:08:25.709 World Labs, and Runway. have all released working 00:08:25.709 --> 00:08:28.589 versions. So there's Genie 3, there's Marble, 00:08:28.730 --> 00:08:31.529 there's GWM1, and they're all using different 00:08:31.529 --> 00:08:34.529 technical approaches. Yann LeCun, who led AI 00:08:34.529 --> 00:08:36.629 research at Meta for years, he left and he founded 00:08:36.629 --> 00:08:39.870 AMI, and he is basically betting that the company 00:08:39.870 --> 00:08:42.549 is gonna be able to build something better than 00:08:42.549 --> 00:08:44.690 LLMs as the path forward, and the LLMs are not 00:08:44.690 --> 00:08:46.669 the way to go. The reason why I bring this story 00:08:46.669 --> 00:08:48.830 up at all is because I think it matters a lot 00:08:48.830 --> 00:08:51.470 when it comes to robotics, video creation, and 00:08:51.470 --> 00:08:53.409 scientific simulation. So we're not just talking 00:08:53.409 --> 00:08:55.279 about chatbots. But what is the next step of 00:08:55.279 --> 00:08:57.220 AI? Like, what are the next industries that this 00:08:57.220 --> 00:09:00.799 is going to go into? Okay, so Meta is doubling 00:09:00.799 --> 00:09:04.740 their Hyperion data center in Louisiana to five 00:09:04.740 --> 00:09:07.379 gigawatts and about $50 billion. That's about 00:09:07.379 --> 00:09:09.720 double their $27 billion they announced last 00:09:09.720 --> 00:09:12.210 October, if you remember. This expansion that 00:09:12.210 --> 00:09:14.309 they're doing now, I think, is showing that Meta 00:09:14.309 --> 00:09:16.669 is making a huge bet that they are going to need 00:09:16.669 --> 00:09:19.110 some enormous amounts of AI computing power to 00:09:19.110 --> 00:09:21.970 stay competitive with Microsoft, Google and Amazon. 00:09:22.250 --> 00:09:24.730 And to be fair, like I would have said, I don't 00:09:24.730 --> 00:09:27.009 know, like I would have discounted Meta and Zuckerberg. 00:09:27.029 --> 00:09:29.090 But I mean, last week they came out with an absolutely 00:09:29.090 --> 00:09:31.250 incredible model. It feels like they're trying 00:09:31.250 --> 00:09:32.779 to get it there. kind of getting a little bit 00:09:32.779 --> 00:09:35.120 of their mojo back. The project that they started 00:09:35.120 --> 00:09:38.480 back in December of 2024, they had, I think, 00:09:38.480 --> 00:09:40.820 $10 billion that they were going to put towards 00:09:40.820 --> 00:09:43.779 a two gigawatt facility. And now Meta is looking 00:09:43.779 --> 00:09:47.039 to reach two gigawatts by 2030. And they have 00:09:47.039 --> 00:09:49.419 not set a timeline for the full completion on 00:09:49.419 --> 00:09:51.259 that. Now, I think there's a lot of reasons why 00:09:51.259 --> 00:09:53.080 they might be expanding this project. In particular, 00:09:53.220 --> 00:09:55.039 we've seen we've all seen the headlines of people 00:09:55.039 --> 00:09:57.679 kind of opposing data centers, or people trying 00:09:57.679 --> 00:10:00.679 to slow them down. And there's been like a lot 00:10:00.679 --> 00:10:02.429 of I think, there's like a statistic that something 00:10:02.429 --> 00:10:05.029 like, you know, six out of 10 proposed data centers 00:10:05.029 --> 00:10:07.350 gets shut down due to like community outreach 00:10:07.350 --> 00:10:09.649 that they don't want these data centers. And 00:10:09.649 --> 00:10:11.830 so I think if you have the approval for a data 00:10:11.830 --> 00:10:14.029 center, and you have the space and the land and 00:10:14.029 --> 00:10:15.909 the government connections and approval, I think 00:10:15.909 --> 00:10:17.309 we're going to start seeing some of these companies 00:10:17.309 --> 00:10:20.129 just decide to build one extra large one wherever 00:10:20.129 --> 00:10:22.690 they're at. Elon Musk is doing this with SpaceX 00:10:22.690 --> 00:10:26.090 building the you know, his giga data center things, 00:10:26.250 --> 00:10:28.230 I can't remember what it's called is Colossus 00:10:28.230 --> 00:10:31.070 three or whatever. Meta is now apparently doing 00:10:31.070 --> 00:10:32.389 the same thing. And I'd be curious to see if 00:10:32.389 --> 00:10:34.990 we see others making the same play where if you 00:10:34.990 --> 00:10:37.190 have the land next to you, might as well expand 00:10:37.190 --> 00:10:39.409 the one that you currently have, rather than 00:10:39.409 --> 00:10:41.350 try to get a whole bunch approved all over the 00:10:41.350 --> 00:10:44.289 place. Louisiana granted Meta a 20 year sales 00:10:44.289 --> 00:10:46.610 tax exemption on equipment purchased for data 00:10:46.610 --> 00:10:50.570 centers built before 2029, which I think lowers 00:10:50.570 --> 00:10:53.149 the cost a ton, maybe 10s of billions of dollars 00:10:53.149 --> 00:10:56.129 in hardware. And local businesses have already 00:10:56.129 --> 00:10:59.090 earned $1 .6 billion in contracts since the construction 00:10:59.090 --> 00:11:01.909 began. So overall for the community and the economy 00:11:01.909 --> 00:11:04.330 there, it's making a lot of money and Meta is 00:11:04.330 --> 00:11:06.110 getting this big kickback in sales tax. So they're 00:11:06.110 --> 00:11:08.230 happy to expand. I think they know that they're 00:11:08.230 --> 00:11:10.269 not going to get it cheaper anywhere else. Meta 00:11:10.269 --> 00:11:12.730 committed over a billion more for roads, water 00:11:12.730 --> 00:11:15.149 systems, and other local infrastructure with 00:11:15.149 --> 00:11:17.309 this project. So hopefully this becomes a real 00:11:17.309 --> 00:11:19.629 win -win for just not just Meta, but also the 00:11:19.629 --> 00:11:21.429 community. Okay, we got to get, we got to end 00:11:21.429 --> 00:11:23.830 the podcast today on some beef in the tech world. 00:11:23.850 --> 00:11:26.590 world, which is Sam Altman and Elon Musk. Elon 00:11:26.590 --> 00:11:29.169 Musk was tweeting and he said that Sam Altman 00:11:29.169 --> 00:11:31.850 takes scamming to a whole new level when talking 00:11:31.850 --> 00:11:34.309 about Apple lawsuit that's going on right now. 00:11:34.389 --> 00:11:36.950 Sam Altman replied by saying, homeboy, you're 00:11:36.950 --> 00:11:39.330 the one selling public market investors on short 00:11:39.330 --> 00:11:42.919 term space data centers. Elon Musk replied and 00:11:42.919 --> 00:11:44.899 said, we start flying them next year. Maybe you 00:11:44.899 --> 00:11:47.120 can come see them if your parole officer approves. 00:11:47.120 --> 00:11:50.059 After stealing an open source AI charity, you 00:11:50.059 --> 00:11:52.340 then stole all of Apple's phone technology. Wow, 00:11:52.519 --> 00:11:54.659 what do you plan for an encore? That's tough 00:11:54.659 --> 00:11:56.360 to beat. It's always fun when the billionaires 00:11:56.360 --> 00:11:58.799 are feuding over on X, but what are the facts? 00:11:59.019 --> 00:12:01.379 Now, there was an article published on TechCrunch 00:12:01.379 --> 00:12:03.279 that I thought was interesting breaking down, 00:12:03.399 --> 00:12:06.120 I guess, the case against Elon Musk. And TechCrunch 00:12:06.120 --> 00:12:08.980 is, I find it just tech, it usually has that 00:12:08.980 --> 00:12:11.360 sort of coverage. But I do have - interesting 00:12:11.360 --> 00:12:14.080 points on all of that. So the first thing that 00:12:14.080 --> 00:12:16.899 TechCrunch says is that SpaceX told IPO investors 00:12:16.899 --> 00:12:19.879 that the Starship was not going to be fully reusable 00:12:19.879 --> 00:12:22.659 near term and it's going to need to expend second 00:12:22.659 --> 00:12:25.159 stages on every launch. According to TechCrunch, 00:12:25.240 --> 00:12:27.240 they said this is going to erase the cost math 00:12:27.240 --> 00:12:30.419 for orbital data centers. They say true operational 00:12:30.419 --> 00:12:33.460 reusability requires years of iteration beyond 00:12:33.460 --> 00:12:36.279 successful test flights. Starship's near term 00:12:36.279 --> 00:12:39.320 manifest is spoken for by NASA and Starlink before 00:12:39.320 --> 00:12:42.059 orbital compute. gets a slot. So they're saying, 00:12:42.120 --> 00:12:43.480 look, like they're not even going to have the 00:12:43.480 --> 00:12:46.059 bandwidth to get this compute up that it's going 00:12:46.059 --> 00:12:47.980 to need tons of iterations is probably not going 00:12:47.980 --> 00:12:51.059 to be very feasible. They say that they said 00:12:51.059 --> 00:12:54.159 scale orbital compute needs multi megawatt satellites, 00:12:54.580 --> 00:12:57.580 vacuum thermal management, ground station bandwidth 00:12:57.580 --> 00:13:00.100 and launch costs low enough to be earthbound 00:13:00.100 --> 00:13:03.059 data centers, a combination experts say arrives 00:13:03.059 --> 00:13:06.799 in 2030s. So TechCrunch is agreeing with Sam 00:13:06.799 --> 00:13:08.519 Altman and saying, look, there's no way that 00:13:08.519 --> 00:13:10.509 SpaceX is going to get the cost efficiency or 00:13:10.509 --> 00:13:12.549 have the bandwidth to put these space data centers 00:13:12.549 --> 00:13:14.629 up there. But a huge part of their two trillion 00:13:14.629 --> 00:13:17.429 dollar valuation was relying heavily on orbital, 00:13:17.590 --> 00:13:20.750 you know, this orbital compute thing. So, you 00:13:20.750 --> 00:13:22.529 know, it's just probably not very feasible. Now, 00:13:22.629 --> 00:13:24.629 on the one hand, I think criticism of these companies 00:13:24.629 --> 00:13:26.970 is fantastic and important. But on the other 00:13:26.970 --> 00:13:28.990 hand, for some reason, all of this just reminds 00:13:28.990 --> 00:13:32.730 me of a post I saw online. Well, there's actually 00:13:32.730 --> 00:13:34.690 been a handful of them. One post I saw online 00:13:34.690 --> 00:13:38.269 in regards to SpaceX was you've probably all 00:13:38.269 --> 00:13:40.330 seen the post. where it's like it's like continue 00:13:40.330 --> 00:13:42.210 iterating it's like always a caption on this 00:13:42.210 --> 00:13:45.769 post but it's the spacex rocket boosters when 00:13:45.769 --> 00:13:47.350 they first started and they were like these really 00:13:47.350 --> 00:13:49.549 complex rocket boosters with wires all over the 00:13:49.549 --> 00:13:51.769 place they look like frankenstein and then it 00:13:51.769 --> 00:13:54.009 shows uh whatever the rocket boosters are today 00:13:54.009 --> 00:13:56.070 which is like really sleek it looks like they've 00:13:56.070 --> 00:13:57.730 gotten all the wires and a lot of the internal 00:13:57.730 --> 00:14:00.629 stuff moved together i remember when this this 00:14:00.629 --> 00:14:02.649 post first went out and they posted a picture 00:14:02.649 --> 00:14:04.950 of it uh there was a guy that like commented 00:14:04.950 --> 00:14:07.289 on it and he was like i work in aerospace and 00:14:07.289 --> 00:14:09.470 this is literally physically impossible This 00:14:09.470 --> 00:14:12.090 is an AI generated image. You couldn't, you know, 00:14:12.090 --> 00:14:15.190 the, the, whatever the wire that goes here, XYZ 00:14:15.190 --> 00:14:17.090 and goes to this place. Like you've literally 00:14:17.090 --> 00:14:19.230 removed this part, this component of the rocket 00:14:19.230 --> 00:14:20.710 booster. That's impossible. This isn't real. 00:14:20.970 --> 00:14:22.870 And replying to that guy was someone that said, 00:14:22.950 --> 00:14:25.110 I'm an engineer at SpaceX. And we worked on, 00:14:25.149 --> 00:14:28.389 um, simplifying this. We were able to move that 00:14:28.389 --> 00:14:29.970 thing over to this other component. So you can't 00:14:29.970 --> 00:14:32.529 see it. This is actually the real rocket booster. 00:14:32.629 --> 00:14:35.659 And the guy said, uh, he's like, dang, you know, 00:14:35.659 --> 00:14:37.559 hats off to you. And that was kind of the thing. 00:14:37.639 --> 00:14:40.700 So all I'm going to say is a lot of people find 00:14:40.700 --> 00:14:44.480 it very easy to keyboard warrior, criticize SpaceX 00:14:44.480 --> 00:14:46.580 in particular, because it's rocket science. And 00:14:46.580 --> 00:14:48.159 you know, if someone sort of knows what they're 00:14:48.159 --> 00:14:50.120 talking about, they can say it sounds super complicated. 00:14:50.220 --> 00:14:52.240 When SpaceX originally started, they said it 00:14:52.240 --> 00:14:55.620 would never be viable because the to hit their 00:14:55.620 --> 00:14:57.600 financial goals and their their launch goals 00:14:57.600 --> 00:14:59.720 because they would have to accelerate launches 00:14:59.720 --> 00:15:02.179 by a certain degree that was just too high. I 00:15:02.179 --> 00:15:03.840 remember seeing an interview by someone in the 00:15:03.840 --> 00:15:05.440 space industry just saying that there was like, 00:15:05.500 --> 00:15:08.600 you know, Elon's mask math was not possible math 00:15:08.600 --> 00:15:09.879 and that it was never actually going to happen. 00:15:09.899 --> 00:15:12.700 So all this to say, when it comes to SpaceX, 00:15:12.960 --> 00:15:16.019 they have proved time and time again to, you 00:15:16.019 --> 00:15:18.990 know. prove all the haters wrong. So maybe everyone 00:15:18.990 --> 00:15:22.169 over at TechCrunch does truly know more than 00:15:22.169 --> 00:15:25.809 what SpaceX says they're planning on doing. But 00:15:25.809 --> 00:15:28.269 I would say give them a little bit of a benefit 00:15:28.269 --> 00:15:30.009 of the doubt because they've proved the doubters 00:15:30.009 --> 00:15:32.269 wrong a couple times. But who knows? Only time 00:15:32.269 --> 00:15:34.070 will tell. If you enjoyed this episode, make 00:15:34.070 --> 00:15:36.809 sure to go check out AIbox .ai if you want to 00:15:36.809 --> 00:15:39.269 get access to over 80 different AI models. So 00:15:39.269 --> 00:15:41.710 the next time Claude runs out of tokens for you, 00:15:41.730 --> 00:15:43.830 you have a place to go. Or if you want to just 00:15:43.830 --> 00:15:45.870 be able to access all of the different AI models. 00:15:45.960 --> 00:15:48.080 You can even pay for a higher subscription tier. 00:15:48.120 --> 00:15:50.879 We have 20 and 40 and I think $80 subscription 00:15:50.879 --> 00:15:53.139 tiers. So you can get tons of tokens, tons of 00:15:53.139 --> 00:15:55.539 usage to access all of the different audio image 00:15:55.539 --> 00:15:58.340 and video models in one place. Go check it out. 00:15:58.360 --> 00:15:59.899 It is AI box dot AI.