WEBVTT 00:00:02.239 --> 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 --> 00:00:11.320 breaking down what it means for your life and 00:00:11.320 --> 00:00:16.920 business. Welcome to the podcast. Today on the 00:00:16.920 --> 00:00:20.620 show, OpenAI is limiting the GPT -5 .6 rollout 00:00:20.620 --> 00:00:22.760 because the government has requested that they 00:00:22.760 --> 00:00:25.019 push it back while they test it out. Because 00:00:25.019 --> 00:00:28.420 of that, Europe is pushing AI sovereignty because 00:00:28.420 --> 00:00:31.219 Trump is doing export controls on Fable 5. And 00:00:31.219 --> 00:00:34.340 now we have OpenAI GPT 5 .6 that they say they 00:00:34.340 --> 00:00:36.100 need to test before it can go out to the world. 00:00:36.240 --> 00:00:39.240 And also we have a company called Patronus AI 00:00:39.240 --> 00:00:42.659 that just raised $50 million to stress test AI 00:00:42.659 --> 00:00:45.780 agents in simulated worlds. So again, we're having 00:00:45.780 --> 00:00:48.039 this thing where these models are super powerful 00:00:48.039 --> 00:00:49.560 that are coming out and the government says, 00:00:49.679 --> 00:00:51.280 hey, let us look over them. And we have these 00:00:51.280 --> 00:00:53.179 new companies that are raising money to stress 00:00:53.179 --> 00:00:55.700 test them. OpenAI is going to be shipping GPT 00:00:55.700 --> 00:00:57.859 5 .6 in three different tiers, and they're going 00:00:57.859 --> 00:01:00.320 to be undercutting Claude on price, which I think 00:01:00.320 --> 00:01:03.640 is important given parts of their mission. And 00:01:03.640 --> 00:01:06.280 Micron, a new company, is overtaking Meta and 00:01:06.280 --> 00:01:09.359 Tesla in market valuation because AI memory demand 00:01:09.359 --> 00:01:11.540 is just exploding. And when I say new company, 00:01:11.640 --> 00:01:13.599 this is a new company, but it's a new company 00:01:13.599 --> 00:01:15.599 for many people in the zeitgeist where we're 00:01:15.599 --> 00:01:17.319 starting to talk about it because of how important 00:01:17.319 --> 00:01:19.379 it is to the supply chain. Let's get into what's 00:01:19.379 --> 00:01:21.840 going on with OpenAI. And speaking of OpenAI, 00:01:21.939 --> 00:01:24.420 if you've ever wanted to use... chat gpt but 00:01:24.420 --> 00:01:27.040 you're like man i wish it could do audio generation 00:01:27.040 --> 00:01:29.680 right inside of it or you're like shoot i'm you 00:01:29.680 --> 00:01:31.420 know really bummed that chat gpt is getting rid 00:01:31.420 --> 00:01:33.400 of sore their video model and you want to use 00:01:33.400 --> 00:01:36.640 google vo3 inside of it or perhaps you use claude 00:01:36.640 --> 00:01:38.760 and you're like man why doesn't claude do video 00:01:38.760 --> 00:01:41.219 image or audio i would love for you to try out 00:01:41.219 --> 00:01:44.719 my own startup aibox .ai we have a new mcp connection 00:01:44.719 --> 00:01:47.540 that you can connect into into your claude into 00:01:47.540 --> 00:01:49.840 your chat gpt or into your gemini and give it 00:01:49.840 --> 00:01:53.000 access to over 80 different ai models right inside 00:01:53.000 --> 00:01:56.040 of whatever the workstation you use and whatever 00:01:56.040 --> 00:01:58.219 AI model you're the most familiar with. I love 00:01:58.219 --> 00:01:59.739 it because when I'm using Claude, it's super 00:01:59.739 --> 00:02:01.579 annoying that it doesn't generate images. And 00:02:01.579 --> 00:02:04.120 I frequently have to go over to chat GPT and 00:02:04.120 --> 00:02:06.480 generate the images over there. So I've actually 00:02:06.480 --> 00:02:09.240 been able to create some tools on AI box, some 00:02:09.240 --> 00:02:10.919 image generation boxes for different projects 00:02:10.919 --> 00:02:12.960 I have. And inside of cloud, I just tell it to 00:02:12.960 --> 00:02:15.780 go use AI box and call the tool that I've created 00:02:15.780 --> 00:02:18.060 and it can generate the images and it gives me 00:02:18.060 --> 00:02:20.500 a link right inside of cloud. So if you want 00:02:20.500 --> 00:02:22.840 to give cloud access to generating images, audio, 00:02:22.960 --> 00:02:26.120 and video, go check out the AI box MCP. It's 00:02:26.120 --> 00:02:29.620 $8 .99 a month. So it's less than 10 bucks, less 00:02:29.620 --> 00:02:31.240 than ChatGPT or any of the other subscriptions. 00:02:31.500 --> 00:02:33.840 And by the way, you can basically use it to replace 00:02:33.840 --> 00:02:36.319 ChatGPT or Gemini or any. thing. It has 80 plus 00:02:36.319 --> 00:02:39.039 different models. It has Grok, it has Mistral, 00:02:39.060 --> 00:02:41.180 it has tons of cool open source models for audio 00:02:41.180 --> 00:02:45.080 image and music. Go check it out, AIbox .ai slash 00:02:45.080 --> 00:02:48.240 MCP. Okay, let's get into what's going on with 00:02:48.240 --> 00:02:50.719 OpenAI. The biggest news here, of course, everyone's 00:02:50.719 --> 00:02:52.340 talking about the fact that the government is 00:02:52.340 --> 00:02:55.219 forcing them to not roll out the latest model, 00:02:55.319 --> 00:02:58.599 GBT 5 .6. They're arguing that this should be 00:02:58.599 --> 00:03:01.139 temporary. This is coming because Washington 00:03:01.139 --> 00:03:03.300 right now is really tightening their oversight 00:03:03.300 --> 00:03:05.129 on a lot of these different AI models i think 00:03:05.129 --> 00:03:07.590 anthropic had the biggest pullback because it 00:03:07.590 --> 00:03:11.189 released fable 5 before um you know and the government 00:03:11.189 --> 00:03:12.889 was like hey look you can't let anyone with foreign 00:03:12.889 --> 00:03:15.310 access get that and anthropic was like hey look 00:03:15.310 --> 00:03:17.210 we think this is a big misunderstanding you guys 00:03:17.210 --> 00:03:18.550 shouldn't be regulating us what's interesting 00:03:18.550 --> 00:03:21.189 is we also have anthropic that just last week 00:03:21.189 --> 00:03:23.449 uh but put out a report just a few days ago put 00:03:23.449 --> 00:03:25.330 out a report to you know some government agency 00:03:25.330 --> 00:03:28.030 where they're like hey look alibaba over in china 00:03:28.030 --> 00:03:30.449 did a model distillation attack on us with like 00:03:30.449 --> 00:03:33.449 20 million messages to our model they made 25 00:03:33.710 --> 00:03:36.069 thousand fake accounts and they did 20 million 00:03:36.069 --> 00:03:38.110 messages and they're going to use those messages 00:03:38.110 --> 00:03:40.270 to train their model for way cheaper and this 00:03:40.270 --> 00:03:42.370 is dangerous yada yada and then they're like 00:03:42.370 --> 00:03:44.050 they're like well i don't know why you know the 00:03:44.050 --> 00:03:46.389 government is trying to not let us roll out fable 00:03:46.389 --> 00:03:48.770 5 well obviously the government's worried that 00:03:48.770 --> 00:03:51.009 when fable 5 gets out alibaba is going to go 00:03:51.009 --> 00:03:54.090 and uh you know or and not just alibaba they're 00:03:54.090 --> 00:03:56.990 probably a i mean semi good actor in this i don't 00:03:56.990 --> 00:03:58.849 know but there's probably a lot of other bad 00:03:58.849 --> 00:04:00.750 actors that could do the same model distillation 00:04:00.750 --> 00:04:03.759 attack they could do fable 5 or anything So Anthropica 00:04:03.759 --> 00:04:05.020 has two sides of the messaging. They're like, 00:04:05.060 --> 00:04:06.379 look, our models are so good. They're going to. 00:04:06.699 --> 00:04:08.479 destroy the world. And then they're like, whoa, 00:04:08.659 --> 00:04:10.819 why are we being regulated? Anyways, so now we 00:04:10.819 --> 00:04:12.580 have this environment where the government thinks 00:04:12.580 --> 00:04:16.339 it can decide what models are good and bad. And 00:04:16.339 --> 00:04:18.199 maybe they can, maybe they can't. I don't know. 00:04:18.220 --> 00:04:19.600 A lot of people have different opinions on that. 00:04:19.819 --> 00:04:24.519 GPT 5 .6 costs about $5 per million input tokens 00:04:24.519 --> 00:04:27.860 and $30 per million output tokens. Terra, which 00:04:27.860 --> 00:04:29.939 is their next one. So they have three, sorry, 00:04:30.060 --> 00:04:32.160 they have three tiers. They got Sol, Terra, and 00:04:32.160 --> 00:04:37.819 Luna. Sol is the $5. Terra is $2 .50, and it 00:04:37.819 --> 00:04:41.959 is then $15, and then Luna is $1 and $6. So Luna 00:04:41.959 --> 00:04:44.040 is the cheapest, Sol is the best. OpeningEye 00:04:44.040 --> 00:04:46.680 says that Sol basically matches Anthropic Clawed 00:04:46.680 --> 00:04:49.259 Mythos 5 on coding, and they say that they're 00:04:49.259 --> 00:04:52.920 using a third of the tokens, of the output tokens, 00:04:53.040 --> 00:04:55.819 to build things. They have safety guardrails, 00:04:55.839 --> 00:04:57.779 and they say that's all inside of the core model 00:04:57.779 --> 00:04:59.759 behavior rather than adding this as a separate 00:04:59.759 --> 00:05:02.000 filter. So with Anthropic, they're like, look, 00:05:02.160 --> 00:05:04.339 we have Mythos 5. It's super dangerous, but we're 00:05:04.339 --> 00:05:06.500 going to add all these guardrails on top of it. 00:05:06.560 --> 00:05:08.399 It'll be fine. And OpenAI says, look, we already 00:05:08.399 --> 00:05:10.199 built the guardrails inside the model, so you 00:05:10.199 --> 00:05:13.079 don't need to do that. I mean, is it true? Well, 00:05:13.220 --> 00:05:15.079 time will tell, right? Is anyone able to hack 00:05:15.079 --> 00:05:16.839 it? We'll find out. But they're saying that it's 00:05:16.839 --> 00:05:19.980 just as good as the Mythos 5 model. And they 00:05:19.980 --> 00:05:22.879 say that it is one third fewer output tokens, 00:05:23.019 --> 00:05:25.439 which would be incredible if it's actually. cheaper 00:05:25.439 --> 00:05:27.600 and more efficient. A recent Trump executive 00:05:27.600 --> 00:05:30.019 order required all furniture companies to submit 00:05:30.019 --> 00:05:31.980 models for government review up to 30 days before 00:05:31.980 --> 00:05:34.660 they release. There was no defined safety standards 00:05:34.660 --> 00:05:37.259 specified there. So anyways, again, just a lot 00:05:37.259 --> 00:05:39.920 of drama on that. And because of all that drama, 00:05:40.060 --> 00:05:43.829 Europe is now pushing for AI sovereignty. Trump 00:05:43.829 --> 00:05:45.810 put in these expert controls and France just 00:05:45.810 --> 00:05:48.670 secured a hundred billion euros in AI infrastructure 00:05:48.670 --> 00:05:52.089 pledges. This is anchored by SoftBank's $75 billion 00:05:52.089 --> 00:05:53.870 data center bet that they're making right now 00:05:53.870 --> 00:05:56.709 inside of Europe. Europe is, I think, getting, 00:05:56.810 --> 00:05:59.750 honestly, people make this like a big, huge, 00:05:59.750 --> 00:06:01.930 dramatic thing where they're like, oh, you know, 00:06:01.949 --> 00:06:04.449 the United States has betrayed Europe or whatever. 00:06:04.629 --> 00:06:06.610 And the US is just going to do what the US is 00:06:06.610 --> 00:06:08.930 going to do. I mean, we got all sorts of crazy 00:06:08.930 --> 00:06:11.189 stuff going on over here. And we also have all 00:06:11.189 --> 00:06:13.509 of the technology. At the end of the day, I think 00:06:13.509 --> 00:06:16.089 that sovereign AI is a fantastic thing. I think 00:06:16.089 --> 00:06:17.970 every country should focus on having their own 00:06:17.970 --> 00:06:19.769 data centers, their own power supply, their own 00:06:19.769 --> 00:06:22.290 AI models, and shouldn't be, you know, reliant 00:06:22.290 --> 00:06:24.870 on anyone else. So I think at the end of the 00:06:24.870 --> 00:06:26.990 day, this isn't a bad thing. This isn't like 00:06:26.990 --> 00:06:29.250 some big drama needs to happen just because Anthropos 00:06:29.250 --> 00:06:32.779 Claude. you know, Fable model got pulled because 00:06:32.779 --> 00:06:35.199 I think a lot of people are like, you know, the 00:06:35.199 --> 00:06:37.139 United States like betrayed Europe by pulling 00:06:37.139 --> 00:06:39.240 the Fable model from European users. I mean, 00:06:39.259 --> 00:06:40.660 they just pulled it from everyone in the whole 00:06:40.660 --> 00:06:42.360 world. And it seems like there's kind of a big 00:06:42.360 --> 00:06:44.000 policy debate that we're having over here. But 00:06:44.000 --> 00:06:47.339 in any case. I think that Sovereign AI models 00:06:47.339 --> 00:06:49.759 are fantastic. I'm excited that SoftBank is backing 00:06:49.759 --> 00:06:52.800 a bunch of those programs. Macron has this thing 00:06:52.800 --> 00:06:55.120 called Choose France. It's an initiative that 00:06:55.120 --> 00:06:56.480 they announced where they're going to be building 00:06:56.480 --> 00:06:58.160 a bunch of their own data centers. Fantastic. 00:06:59.019 --> 00:07:02.560 Anthropix single $65 billion fundraise exceeded 00:07:02.560 --> 00:07:05.500 the entire sum invested in all European and UK 00:07:05.500 --> 00:07:08.560 AI startups last year combined. So at the end 00:07:08.560 --> 00:07:11.480 of the day, there's all of these national programs 00:07:11.480 --> 00:07:13.680 that are going on, but there's still a drop in 00:07:13.680 --> 00:07:16.949 the bucket compared. to what just one company 00:07:16.949 --> 00:07:18.910 in the United States was able to raise. And then 00:07:18.910 --> 00:07:21.110 we have a lot of different companies. So I think 00:07:21.110 --> 00:07:23.610 at the end of the day, Europe should build out 00:07:23.610 --> 00:07:26.829 what they're focusing on. And I think that there's 00:07:26.829 --> 00:07:30.529 a lot of opportunity for new innovation to come 00:07:30.529 --> 00:07:32.790 out of these countries because they'll, you know, 00:07:32.790 --> 00:07:35.550 people will finally be funding national like 00:07:35.550 --> 00:07:38.970 AI, like I guess smaller AI companies for geo. 00:07:39.900 --> 00:07:41.860 geographic reasons rather than everyone just 00:07:41.860 --> 00:07:44.660 dumping more money into the mega caps, the open 00:07:44.660 --> 00:07:46.879 AI and the anthropic. So I'm excited to see more 00:07:46.879 --> 00:07:49.160 competition in the marketplace. One, so with 00:07:49.160 --> 00:07:51.560 all of this drama around these AI models, you 00:07:51.560 --> 00:07:53.420 know, coming out and how dangerous are they and 00:07:53.420 --> 00:07:54.860 how good are they? There's a new company called 00:07:54.860 --> 00:07:56.800 Patronus AI, and they just raised $50 million. 00:07:57.040 --> 00:07:59.439 It's their series B. And the way that what they're 00:07:59.439 --> 00:08:02.079 doing is they're trying to simulate digital worlds. 00:08:02.240 --> 00:08:04.379 So they're basically testing the new AI agents 00:08:04.379 --> 00:08:06.420 before they get deployed. You can imagine, right, 00:08:06.459 --> 00:08:08.480 the government was like, look, you got to give 00:08:08.480 --> 00:08:10.720 us 30 days. before you put it out. And in my 00:08:10.720 --> 00:08:12.420 mind, I'm like, what are you guys doing in 30 00:08:12.420 --> 00:08:13.899 days that you could possibly test these things? 00:08:14.199 --> 00:08:16.420 Apparently, Patronus AI could be an answer to 00:08:16.420 --> 00:08:18.939 that, where they're simulating the real world 00:08:18.939 --> 00:08:20.620 and they're taking these AI models and they're 00:08:20.620 --> 00:08:22.439 just telling them to wreak havoc and seeing what 00:08:22.439 --> 00:08:25.120 they could actually do. I mean, not really just 00:08:25.120 --> 00:08:26.500 wreak havoc, but they're trying to stress test 00:08:26.500 --> 00:08:28.019 it, right? So they're building a simulated world 00:08:28.019 --> 00:08:29.980 and letting it run there. Their revenue grew 00:08:29.980 --> 00:08:32.960 15x in the last year. There's a bunch of different 00:08:32.960 --> 00:08:35.299 AI labs that are trying to get these kind of... 00:08:35.759 --> 00:08:39.700 um testing's done and i think this is uh this 00:08:39.700 --> 00:08:41.360 is a great company that's working on it patronus 00:08:41.360 --> 00:08:44.059 has digital replicas of websites and internal 00:08:44.059 --> 00:08:46.460 systems where agents can train under reinforcement 00:08:46.460 --> 00:08:48.879 learning it's going to reward task completion 00:08:48.879 --> 00:08:51.440 and it's going to penalize errors so it's interesting 00:08:51.440 --> 00:08:53.440 right they have like you know, like when they 00:08:53.440 --> 00:08:55.820 say they've made a digital world, it's not like 00:08:55.820 --> 00:08:57.639 they've like a literal, it's not like, you know, 00:08:57.659 --> 00:08:59.720 Mark Zuckerberg's metaverse. They're making a 00:08:59.720 --> 00:09:02.159 world of computers and systems and code. And 00:09:02.159 --> 00:09:03.460 they've probably grabbed a whole bunch of open 00:09:03.460 --> 00:09:05.759 source projects, if I'm being honest, and spun 00:09:05.759 --> 00:09:08.080 up a lot of. You know, a lot of stuff where these 00:09:08.080 --> 00:09:10.360 models have to try to interact, get tasks done, 00:09:10.440 --> 00:09:12.220 to probably do penetration testing, that kind 00:09:12.220 --> 00:09:15.379 of stuff. So Notable Capital, Lightspeed, Datadog, 00:09:15.480 --> 00:09:17.879 Samsung, all of them joined Greenfield Partners 00:09:17.879 --> 00:09:20.500 in this round. And the total funding for this 00:09:20.500 --> 00:09:22.799 company is about $70 million. It was just founded 00:09:22.799 --> 00:09:25.759 in 2023. I'm excited to see where they go with 00:09:25.759 --> 00:09:28.980 this. A company called Micron has just overtaken 00:09:28.980 --> 00:09:31.679 Meta and Tesla in market value because of the 00:09:31.679 --> 00:09:36.799 crazy demand for AI memory. This is a problem 00:09:36.799 --> 00:09:38.919 basically for everyone. We just had Apple yesterday 00:09:38.919 --> 00:09:41.720 say that they're raising the prices on all MacBooks. 00:09:42.190 --> 00:09:45.049 by $100 MacBook Airs and then $300 on the MacBook 00:09:45.049 --> 00:09:46.690 Pros. And like basically everything's going up 00:09:46.690 --> 00:09:49.070 in price because there's this huge memory shortage 00:09:49.070 --> 00:09:51.789 everywhere, which is caused by AI. It's just 00:09:51.789 --> 00:09:54.289 like really explosive demand for high bandwidth 00:09:54.289 --> 00:09:57.490 memory or HBM. And this is of course used for 00:09:57.490 --> 00:09:59.549 AI training clusters. So something that a lot 00:09:59.549 --> 00:10:01.830 of people are bringing up is that this is basically 00:10:01.830 --> 00:10:05.139 showing. that memory and not just GPUs alone 00:10:05.139 --> 00:10:08.220 are now kind of gating or kind of like a constraint 00:10:08.220 --> 00:10:10.340 for the AI infrastructure build out. I think 00:10:10.340 --> 00:10:11.799 in the past, everyone was like, oh, man, how 00:10:11.799 --> 00:10:13.679 are we ever going to get enough GPUs from NVIDIA? 00:10:13.700 --> 00:10:15.820 And NVIDIA has like, you know, limits to how 00:10:15.820 --> 00:10:17.639 many they'll sell to a specific company. And 00:10:17.639 --> 00:10:19.019 that was kind of, I guess, the last couple of 00:10:19.019 --> 00:10:20.779 years, what everyone's been worried about. No 00:10:20.779 --> 00:10:23.019 longer is it GPUs. Now we're just literally running 00:10:23.019 --> 00:10:26.320 out of memory. And that's going to kind of hold 00:10:26.320 --> 00:10:29.309 back the AI infrastructure build out. So I think 00:10:29.309 --> 00:10:31.830 these suppliers that are really kind of have 00:10:31.830 --> 00:10:35.049 a bottleneck on the memory infrastructure are 00:10:35.049 --> 00:10:36.730 going to be making a ton of money. And Micron 00:10:36.730 --> 00:10:39.850 is one of them. And by the way, HBM is basically 00:10:39.850 --> 00:10:41.929 a custom, something that is custom engineered 00:10:41.929 --> 00:10:44.210 for NVIDIA accelerators, and it's sold under 00:10:44.210 --> 00:10:47.009 multi -year supply agreements. They sell this 00:10:47.009 --> 00:10:49.610 at a premium pricing that basically traditional 00:10:49.610 --> 00:10:53.549 DRAM can never get. So only three companies can 00:10:53.549 --> 00:10:56.029 actually manufacture HBM at scale. That's Micron. 00:10:56.610 --> 00:10:59.529 SK Hynix, which is, I believe, in Korea, and 00:10:59.529 --> 00:11:01.470 then also Samsung. And I believe SK Hynix is 00:11:01.470 --> 00:11:03.730 planning an IPO in the United States soon, or 00:11:03.730 --> 00:11:06.509 they're at least looking at this. And then Samsung. 00:11:06.629 --> 00:11:09.950 And so basically, there is some enormous price 00:11:09.950 --> 00:11:12.970 power for suppliers because there's this kind 00:11:12.970 --> 00:11:15.070 of two -year demand surge that's been coming. 00:11:15.129 --> 00:11:17.110 There's not a lot of suppliers that are able 00:11:17.110 --> 00:11:19.830 to make this. Meta and Tesla both, which are 00:11:19.830 --> 00:11:23.950 huge hyperscalers, now rank below the component 00:11:23.950 --> 00:11:26.840 supplier, Micron. which both of them are using, 00:11:26.919 --> 00:11:28.740 right? They're both dependent on Micron. But 00:11:28.740 --> 00:11:30.779 I think that just goes to show how much projected 00:11:30.779 --> 00:11:33.799 AI growth is tied to infrastructure. You know, 00:11:33.799 --> 00:11:35.039 they actually have to go buy this externally. 00:11:35.240 --> 00:11:37.940 And so it's incredible to see the run -up that 00:11:37.940 --> 00:11:39.620 Micron has been doing. Guys, thank you so much 00:11:39.620 --> 00:11:41.440 for joining the podcast today, for tuning in. 00:11:41.500 --> 00:11:43.500 If you learned anything new, if you enjoyed the 00:11:43.500 --> 00:11:45.700 episode, make sure to leave a rating and review 00:11:45.700 --> 00:11:48.360 wherever you get your podcasts. On Apple, leave 00:11:48.360 --> 00:11:50.799 a comment. I read them all. Drop some stars if 00:11:50.799 --> 00:11:52.559 there's any topics or companies you want covered 00:11:52.559 --> 00:11:55.000 or things you'd like me to talk about. about 00:11:55.000 --> 00:11:57.100 more and make sure to go check out AI box dot 00:11:57.100 --> 00:11:59.299 AI slash MCP. I'll leave a link in the description 00:11:59.299 --> 00:12:02.440 if you want to get images, audio and video inside 00:12:02.440 --> 00:12:05.480 of Claude or any of your other models. You can 00:12:05.480 --> 00:12:08.299 get access to over 80 different AI models inside 00:12:08.299 --> 00:12:11.360 whatever AI tool or whatever AI model you're 00:12:11.360 --> 00:12:13.100 using. All right. Thanks so much for tuning in. 00:12:13.159 --> 00:12:14.539 I'll catch you all in the next episode.