WEBVTT 00:00:02.239 --> 00:00:04.780 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 --> 00:00:17.660 business. OpenAI is shutting down their Atlas 00:00:17.660 --> 00:00:20.280 browser nine months after it launched, and they're 00:00:20.280 --> 00:00:23.559 folding the features into ChatGPT, into a native 00:00:23.559 --> 00:00:27.059 app, and into a Chrome app. Meta is launching 00:00:27.059 --> 00:00:30.300 MuseSpark 1 .1, and this is directly competing 00:00:30.300 --> 00:00:34.579 with Claude and GPT 5 .6 on coding. Sunrun is 00:00:34.579 --> 00:00:37.039 going to pay customers to host AI compute nodes 00:00:37.039 --> 00:00:39.299 in their homes so you can turn your home into 00:00:39.299 --> 00:00:43.119 an AI data center. And SkyHinix has raised $26 00:00:43.119 --> 00:00:47.520 .5 billion. in the largest ever foreign IPO in 00:00:47.520 --> 00:00:50.719 the US market. Fiji Simone is stepping down as 00:00:50.719 --> 00:00:53.200 OpenAI's application chief because of chronic 00:00:53.200 --> 00:00:55.579 illness. The craziest thing just happened to 00:00:55.579 --> 00:00:57.259 me yesterday. If you listen to the podcast, you 00:00:57.259 --> 00:00:58.320 know that for the last couple of months, I've 00:00:58.320 --> 00:01:00.960 been talking trash about a bunch of Chinese AI 00:01:00.960 --> 00:01:03.600 companies on my podcast, but they have finally 00:01:03.600 --> 00:01:06.780 proved me wrong. So yesterday I discovered the 00:01:06.780 --> 00:01:11.120 incredibly innovative AIbox365 .com. So essentially 00:01:11.120 --> 00:01:13.780 I was getting Claude to do SEO for me, asking 00:01:13.780 --> 00:01:15.879 - you know, a bunch of a bunch of different tips 00:01:15.879 --> 00:01:18.760 and strategies for AI box .ai, my own startups, 00:01:18.840 --> 00:01:21.340 SEO, and it identified for me, it said, Look, 00:01:21.500 --> 00:01:23.439 there's a red flag, because there's another startup 00:01:23.439 --> 00:01:26.959 called AI box 365 .com. It does exactly what 00:01:26.959 --> 00:01:29.000 you do. So it's a company, they went and they 00:01:29.000 --> 00:01:32.439 registered our exact name, they added 365, they 00:01:32.439 --> 00:01:34.959 cloned our landing page, and our pricing, it's 00:01:34.959 --> 00:01:37.280 like the same pricing. But the only difference 00:01:37.280 --> 00:01:39.620 is if you click the sign up button on their website, 00:01:39.840 --> 00:01:41.620 it doesn't sign you up right there, it actually 00:01:41.620 --> 00:01:43.829 redirects you to a complete completely different 00:01:43.829 --> 00:01:48.310 site called AI mirror hub .com. And that one 00:01:48.310 --> 00:01:50.909 has a bunch of Chinese characters on it. I think 00:01:50.909 --> 00:01:53.870 the site is actually like translated from Chinese 00:01:53.870 --> 00:01:56.829 into English, just automatically on Google, how 00:01:56.829 --> 00:01:58.549 you know how Google does that. If you've seen 00:01:58.549 --> 00:02:00.390 Silicon Valley, you know, they basically just 00:02:00.390 --> 00:02:03.489 got Jin Yang, where I mean, he pitches, you know, 00:02:03.489 --> 00:02:05.670 Uber, but it's Chinese Airbnb, but it's Chinese, 00:02:05.750 --> 00:02:09.050 we now have AI box, but it's Chinese. And everything 00:02:09.050 --> 00:02:11.129 else is the same, except that it now points to 00:02:11.129 --> 00:02:12.870 their website, just basically. basically poaching 00:02:12.870 --> 00:02:15.129 off of our SEO. So how do we stack up against 00:02:15.129 --> 00:02:17.110 this incredibly innovative Chinese competitor? 00:02:17.310 --> 00:02:19.389 Well, I'm proud to say that even with the ability 00:02:19.389 --> 00:02:21.949 to go and clone our site, we still beat them 00:02:21.949 --> 00:02:26.349 on pricing. They are $9. We're $8 .99. So we're 00:02:26.349 --> 00:02:28.650 one cent cheaper. They have 10 models on their 00:02:28.650 --> 00:02:31.050 platform. We have over 80 different models on 00:02:31.050 --> 00:02:33.050 our platform. And if you go and get an annual 00:02:33.050 --> 00:02:36.590 subscription from them, it's $109 where our cheapest 00:02:36.590 --> 00:02:41.090 annual subscription is $86 .99. So we're beating 00:02:41.090 --> 00:03:09.099 them on annual. I'll leave a link in the show 00:03:09.099 --> 00:03:11.840 notes. OpenAI is... Shutting down Atlas. This 00:03:11.840 --> 00:03:14.240 is a browser that they launched nine months ago, 00:03:14.319 --> 00:03:15.960 and they're moving all of the features from that 00:03:15.960 --> 00:03:18.520 into a Chrome extension. And they're also upgrading 00:03:18.520 --> 00:03:21.639 their chat GPT desktop app instead. I mean, this 00:03:21.639 --> 00:03:23.419 is basically what Claude already has with the 00:03:23.419 --> 00:03:25.560 Claude app that has like all of their different 00:03:25.560 --> 00:03:27.780 Claude code and Claude co -work and even the 00:03:27.780 --> 00:03:29.900 chat all inside of one app. It looks like OpenAI 00:03:29.900 --> 00:03:31.560 is trying to consolidate and do that same thing. 00:03:31.659 --> 00:03:34.319 And I even saw Claude today launch the fact that 00:03:34.319 --> 00:03:36.639 they have a browser built into that. So I wouldn't 00:03:36.639 --> 00:03:38.860 be surprised if chat GPT did something similar 00:03:38.860 --> 00:03:40.819 and had some sort of browser. built in eventually 00:03:40.819 --> 00:03:43.180 to the ChatGPT app. They want to consolidate. 00:03:43.240 --> 00:03:45.419 They don't want multiple apps. And Claude also 00:03:45.419 --> 00:03:48.259 has a Chrome extension. And so it looks like 00:03:48.259 --> 00:03:51.039 OpenAI is basically following Claude's playbook. 00:03:51.159 --> 00:03:53.460 It works really well. I use both the Chrome extension 00:03:53.460 --> 00:03:56.780 and the app on my desktop. And sometimes they 00:03:56.780 --> 00:03:58.659 chat back and forth and they use each other. 00:03:58.740 --> 00:04:00.400 But anyways, this is, I think, the way to go. 00:04:00.479 --> 00:04:03.419 And OpenAI is copying that. The new ChatGPT Chrome 00:04:03.419 --> 00:04:05.919 extension is going to compete directly with Google's 00:04:05.919 --> 00:04:08.280 Gemini side panel. And I think it's going to 00:04:08.280 --> 00:04:09.650 basically answer any questions. that you have 00:04:09.650 --> 00:04:12.469 about a web, any specific web page. So I can 00:04:12.469 --> 00:04:14.210 look at the content on that page. It can handle 00:04:14.210 --> 00:04:16.529 multi -step tasks. You're not leaving the browser. 00:04:16.709 --> 00:04:18.290 The other cool thing that they do is they're 00:04:18.290 --> 00:04:20.490 building a cloud browser. So it's going to run 00:04:20.490 --> 00:04:23.069 on their own servers. And basically it lets ChatGPT 00:04:23.069 --> 00:04:26.009 agents complete tasks for you while your desktop 00:04:26.009 --> 00:04:29.250 app is staying responsive. There's a couple of 00:04:29.250 --> 00:04:31.250 different cool things here. One is like, I love 00:04:31.250 --> 00:04:33.310 that Cloud Cowork, for example, uses the files 00:04:33.310 --> 00:04:37.550 on my computer or, you know, let's say the Cloud 00:04:37.550 --> 00:04:40.350 Chrome extension. use and take control of my 00:04:40.350 --> 00:04:42.449 screen. But a lot of times I just want something 00:04:42.449 --> 00:04:44.290 done and I don't want it to mess up with my screen 00:04:44.290 --> 00:04:46.670 or what's happening on my computer. I just want 00:04:46.670 --> 00:04:47.990 it to go happen in the backend. So that looks 00:04:47.990 --> 00:04:50.589 like what OpenAI is building here. But this is 00:04:50.589 --> 00:04:53.089 now leaving room for the browser company and 00:04:53.089 --> 00:04:56.009 for Perplexity that both have their own AI native 00:04:56.009 --> 00:04:58.730 browsers to potentially take more control of 00:04:58.730 --> 00:05:02.310 the industry. Meta has just released MuseSpark 00:05:02.310 --> 00:05:04.930 1 .1. This is a coding model and it's designed 00:05:04.930 --> 00:05:07.389 to build features and manage workflows inside 00:05:07.389 --> 00:05:10.240 of enterprise software without any human intervention. 00:05:10.540 --> 00:05:14.040 At 1 .25 per million input tokens, it's basically 00:05:14.040 --> 00:05:16.379 priced directly to compete with Claude Haiku 00:05:16.379 --> 00:05:20.100 4 .5 and GPT 5 .6 Luna. So these are kind of 00:05:20.100 --> 00:05:22.240 the dominant cheap coding models. So it's not, 00:05:22.279 --> 00:05:23.839 you know, the best. It's basically trying to 00:05:23.839 --> 00:05:25.560 compete with the cheap coding models. The thing 00:05:25.560 --> 00:05:27.220 that I love about this is that Mark Zuckerberg 00:05:27.220 --> 00:05:30.720 has broken his three year silence on X to promote 00:05:30.720 --> 00:05:33.319 this. This is his first post since July 2023, 00:05:33.579 --> 00:05:35.319 back when I think threads dropped. And when threads 00:05:35.319 --> 00:05:37.180 dropped, it's kind of funny because he was making 00:05:37.180 --> 00:05:39.220 a post about threads and it was, you know, cloning 00:05:39.220 --> 00:05:41.800 X. Specifically what he said about this, he said, 00:05:41.839 --> 00:05:44.000 today we're releasing MuseSpark 1 .1, a strong 00:05:44.000 --> 00:05:46.600 agentic encoding model at a very low price. It's 00:05:46.600 --> 00:05:49.360 available through our new meta model API and 00:05:49.360 --> 00:05:52.660 meta API. MuseSpark 1 .1 is the strongest at 00:05:52.660 --> 00:05:55.180 agentic performance, tool use and computer use. 00:05:55.300 --> 00:05:57.600 It does well on long running tasks with a million 00:05:57.600 --> 00:06:00.319 token context window and can delegate execution 00:06:00.319 --> 00:06:02.759 to sub agents running in parallel and is trained 00:06:02.759 --> 00:06:05.240 to use computer interfaces on desktop, mobile 00:06:05.240 --> 00:06:08.199 and browser. browser. The meta model API allows 00:06:08.199 --> 00:06:10.839 developers to build using Muse spark. For the 00:06:10.839 --> 00:06:12.800 first time, our focus is on delivering strong 00:06:12.800 --> 00:06:15.740 agentic and multimodal models at a very low cost 00:06:15.740 --> 00:06:18.279 more to come soon. Honestly, I'm pretty stoked 00:06:18.279 --> 00:06:20.420 that we're getting Mark Zuckerberg tweeting again. 00:06:20.519 --> 00:06:22.660 But also I think this is a good model. I'm excited 00:06:22.660 --> 00:06:24.620 to see what they do with this. And a lot of people 00:06:24.620 --> 00:06:26.879 have been pointing out that like just from internet 00:06:26.879 --> 00:06:29.600 traffic, meta is scraping like crazy. They're 00:06:29.600 --> 00:06:31.519 obviously building some big things. We don't 00:06:31.519 --> 00:06:33.519 know exactly what but I'm excited to have another 00:06:33.519 --> 00:06:36.329 competitor in the ring. Accompanying called Sunrun 00:06:36.329 --> 00:06:38.970 is paying customers to host AI compute hardware 00:06:38.970 --> 00:06:41.509 in their homes, aka little mini data centers 00:06:41.509 --> 00:06:43.810 at your house. They're trying to basically spread 00:06:43.810 --> 00:06:46.790 the AI workload across 1 .1 million residential 00:06:46.790 --> 00:06:49.889 solar and battery systems instead of building 00:06:49.889 --> 00:06:51.569 massive data centers. There's a couple interesting 00:06:51.569 --> 00:06:54.209 things right now. One of them, I think it basically 00:06:54.209 --> 00:06:56.069 gets them around a huge problem, which is that 00:06:56.069 --> 00:06:59.129 70 % of Americans oppose new data center construction. 00:06:59.529 --> 00:07:01.250 You know, they're worried about like pollution 00:07:01.250 --> 00:07:03.490 and noise and power drain. There's also a ton 00:07:03.490 --> 00:07:05.990 of like political campaigns going up. Some people 00:07:05.990 --> 00:07:08.110 have accused China of funding a lot of these 00:07:08.110 --> 00:07:09.810 different campaigns. Kevin O 'Leary said that's 00:07:09.810 --> 00:07:12.569 the case in many instances. So being able to 00:07:12.569 --> 00:07:14.470 just, you know, not have to worry about getting 00:07:14.470 --> 00:07:16.350 one of these things approved, because I saw a 00:07:16.350 --> 00:07:18.230 stat that it was like 40 to 60 % of all data 00:07:18.230 --> 00:07:21.569 center projects don't get approved on the local 00:07:21.569 --> 00:07:23.449 level because of, you know, people pushing back 00:07:23.449 --> 00:07:25.970 or whatever. So if you could just go directly 00:07:25.970 --> 00:07:27.629 to consumers, and if everyone had a little mini 00:07:27.629 --> 00:07:29.389 data center on the side of their house, I've 00:07:29.389 --> 00:07:31.569 seen people, you know, float like, hey, you get 00:07:31.569 --> 00:07:34.040 free electricity and other different benefits 00:07:34.040 --> 00:07:37.160 to it. So if that was the case, I guarantee tons 00:07:37.160 --> 00:07:39.079 of Americans would sign up for free electricity 00:07:39.079 --> 00:07:41.560 to host a data center on the side of their house. 00:07:41.819 --> 00:07:43.899 And then you wouldn't have any of this kind of 00:07:43.899 --> 00:07:47.480 regulatory burden as well. Sunrun has completed 00:07:47.480 --> 00:07:49.500 a proof of concept pilot and they're planning 00:07:49.500 --> 00:07:52.839 to resell the compute capacity to AI companies 00:07:52.839 --> 00:07:54.860 like Anthropic and OpenAI within the coming months. 00:07:55.060 --> 00:07:57.500 Home -based nodes like these little mini quote 00:07:57.500 --> 00:08:00.120 unquote data centers, will essentially run partly 00:08:00.120 --> 00:08:03.240 on rooftop solar and also batteries. And then 00:08:03.240 --> 00:08:05.379 they're going to basically reduce grid draw compared 00:08:05.379 --> 00:08:08.160 to a single warehouse scale facility, which is 00:08:08.160 --> 00:08:09.920 drawing hundreds of megawatts. And I think this 00:08:09.920 --> 00:08:11.459 is kind of the concept that a lot of people have 00:08:11.459 --> 00:08:12.819 talked about with building these data centers. 00:08:12.899 --> 00:08:14.379 Trump has talked about it a lot saying, hey, 00:08:14.399 --> 00:08:15.899 look, if you're building the data centers, you 00:08:15.899 --> 00:08:17.800 should go and build like your own energy generation 00:08:17.800 --> 00:08:19.920 plant next to it, basically telling people they 00:08:19.920 --> 00:08:22.560 should build their own gas, electric, solar, 00:08:22.800 --> 00:08:25.839 wind, whatever you want, like energy generation 00:08:25.839 --> 00:08:27.279 should be built in. to these so that it's not 00:08:27.279 --> 00:08:29.620 pulling off of the grid. And if you do this the 00:08:29.620 --> 00:08:31.019 way that they're doing this now, they're going 00:08:31.019 --> 00:08:33.580 to bring their own batteries, solar, and it would 00:08:33.580 --> 00:08:36.860 just basically increase what is on the grid and 00:08:36.860 --> 00:08:39.659 not take from it. A lot of the top AI labs are 00:08:39.659 --> 00:08:42.019 trying to get these data center deals. It's basically 00:08:42.019 --> 00:08:44.580 a land grab right now. And Anthropic just signed 00:08:44.580 --> 00:08:47.620 a 20 year Kentucky facility lease, but the permits 00:08:47.620 --> 00:08:50.179 are a huge bottleneck and it hasn't actually 00:08:50.179 --> 00:08:53.450 solidified yet. SK Hynix. I've been talking about 00:08:53.450 --> 00:08:55.529 them a ton on the podcast lately. I've just raised 00:08:55.529 --> 00:08:59.110 $26 .5 billion. This is the largest ever foreign 00:08:59.110 --> 00:09:01.629 IPO that's happened in the United States. I've 00:09:01.629 --> 00:09:03.210 been talking about this as it's been coming up. 00:09:03.350 --> 00:09:05.090 But the thing that's interesting is that they 00:09:05.090 --> 00:09:09.009 actually beat Alibaba's 2014 record. And Wall 00:09:09.009 --> 00:09:12.269 Street is bidding really aggressively for exposure 00:09:12.269 --> 00:09:14.970 to this one particular chip maker because they 00:09:14.970 --> 00:09:17.330 make high bandwidth memory, which basically powers 00:09:17.330 --> 00:09:21.360 NVIDIA's AI accelerators and AMD's as well. interesting 00:09:21.360 --> 00:09:23.440 is there's only three players that make this 00:09:23.440 --> 00:09:25.960 particular high bandwidth memory. Samsung is 00:09:25.960 --> 00:09:29.200 one of them. SK Hynix is another. And there's 00:09:29.200 --> 00:09:31.299 one called Micron. I believe they're based in 00:09:31.299 --> 00:09:33.120 Idaho, but there's only three of these companies 00:09:33.120 --> 00:09:34.980 in the world. Their stock prices have absolutely 00:09:34.980 --> 00:09:37.759 been soaring. The companies are raising and making 00:09:37.759 --> 00:09:40.200 tons of money. It's absolutely eye -watering. 00:09:40.240 --> 00:09:42.120 And so right now they've decided that they're 00:09:42.120 --> 00:09:43.799 going to go public and capitalize on this. They 00:09:43.799 --> 00:09:50.049 sold 1 .779 So 177, basically, or 178 million 00:09:50.049 --> 00:09:54.830 shares at $149 each. It was priced 2 .7 % above 00:09:54.830 --> 00:09:59.190 their South Korean average. And there is demand 00:09:59.190 --> 00:10:01.830 for more than seven times the available supply. 00:10:02.009 --> 00:10:05.590 The stock opened 14 % above the IPO price. Friday, 00:10:05.590 --> 00:10:09.549 the ticker is SKHYV. It then started doing regular 00:10:09.549 --> 00:10:14.250 trading on Monday and under SKHY. U .S. Commerce 00:10:14.250 --> 00:10:16.230 Secretary Howard Lutnick said that he's already 00:10:16.299 --> 00:10:18.840 negotiating with SK Hynix and Samsung to both 00:10:18.840 --> 00:10:20.740 build new chip factors in the United States. 00:10:20.820 --> 00:10:22.820 We already have one in Idaho, but you know, or 00:10:22.820 --> 00:10:25.179 one company that's based in Idaho, but we want 00:10:25.179 --> 00:10:27.700 to bring everyone else in as well. I think basically 00:10:27.700 --> 00:10:30.019 this just shows that in the United States, there's 00:10:30.019 --> 00:10:32.580 a ton of appetite for AI infrastructure, even 00:10:32.580 --> 00:10:34.740 if it's not in the US, even if it's not based 00:10:34.740 --> 00:10:36.279 in the United States. Like, I mean, essentially 00:10:36.279 --> 00:10:38.340 by investing in this IPO, you're investing in 00:10:38.340 --> 00:10:40.740 hardware companies based in Korea, which isn't 00:10:40.740 --> 00:10:43.320 a bad thing. Washington right now wants to rebuild 00:10:43.320 --> 00:10:45.960 a lot of the domestic memory manufacturing. which 00:10:45.960 --> 00:10:49.360 all of it has kind of been abandoned in the last 00:10:49.360 --> 00:10:52.840 10 or 20 years. Micron fired back with a $250 00:10:52.840 --> 00:10:55.659 billion US investment pledge. A big part of this, 00:10:55.740 --> 00:10:59.639 I think, is federal subsidies and talent. Fiji 00:10:59.639 --> 00:11:01.700 Simone is stepping down as opening eyes number 00:11:01.700 --> 00:11:05.139 two executive and she is moving to part time 00:11:05.139 --> 00:11:07.940 advisor after a three month medical leave. So 00:11:07.940 --> 00:11:10.299 she had consolidated products and a lot of business 00:11:10.299 --> 00:11:13.539 operations under her since May of last year, 00:11:13.580 --> 00:11:15.960 but she had a really severe relapse of a chronic 00:11:15.960 --> 00:11:18.399 illness and that led her to prioritize recovery 00:11:18.399 --> 00:11:21.879 ahead of. OpenAI's IPO that's happening next 00:11:21.879 --> 00:11:25.519 year. She was diagnosed with POTS and Greg Brockman, 00:11:25.559 --> 00:11:27.039 I think is going to take over a lot of her responsibilities. 00:11:27.200 --> 00:11:28.460 The reason I wanted to bring this up, because 00:11:28.460 --> 00:11:29.860 I know this is sort of just like a random thing 00:11:29.860 --> 00:11:31.940 to throw on the end of a podcast. I mean, it 00:11:31.940 --> 00:11:34.100 is the number two person at OpenAI, but my point 00:11:34.100 --> 00:11:37.419 being life is short. All of this and AI, everything 00:11:37.419 --> 00:11:39.240 happening in the markets, all of this, you know, 00:11:39.240 --> 00:11:40.779 technology, it's super exciting and interesting. 00:11:40.919 --> 00:11:42.990 But at the end of the day. Nothing is more important 00:11:42.990 --> 00:11:44.789 than taking care of yourself, your family, your 00:11:44.789 --> 00:11:48.129 help. There are much bigger things than AI and 00:11:48.129 --> 00:11:49.929 technology, even though this is what we talk 00:11:49.929 --> 00:11:52.389 about every day. I think she obviously realized 00:11:52.389 --> 00:11:54.789 that life is short. I'm really, you know, praying 00:11:54.789 --> 00:11:57.309 for a speedy recovery for her and for anyone 00:11:57.309 --> 00:11:59.889 that is suffering from illness. All right, guys, 00:11:59.929 --> 00:12:01.309 thank you so much for tuning into the podcast. 00:12:01.490 --> 00:12:03.730 If you enjoyed this episode, make sure to go 00:12:03.730 --> 00:12:08.250 check out therealaibox .ai, not the fake one 00:12:08.250 --> 00:12:10.990 that is based in China. And also make sure to 00:12:10.990 --> 00:12:13.049 leave a review if you haven't left a review already 00:12:13.049 --> 00:12:15.769 on the podcast it helps the show out so much 00:12:15.769 --> 00:12:18.409 i am eternally grateful and i hope you guys all 00:12:18.409 --> 00:12:19.470 have a great rest of your day