0:02 Welcome to the AI Chat Podcast. I'm your host, Jaden Schaefer. Every day I cover cutting edge AI news and talk with the leaders behind it, breaking down what it means for your life and business. Welcome to the podcast. We've just passed the US midterm election. So today on the 0:19 podcast, I want to talk about some of the implications of the blue wave or the Democrat sweep that we've seen in the midterm elections, both the House and the Senate. We have implications for a bunch of AI data center stocks. Elon Musk has been 0:32 tweaked. about it. We also have a bunch of news out of Anthropix, some new deals they've been signing. We have some huge milestones that Chatshapti has hit as far as usage goes. And also Google has a new strategy for data centers in Malaysia. 0:49 We have a couple of new startups, including Isomorphic Labs that are in talks to raise $5 billion. If you don't know, Isomorphic Labs is... tied to the current CEO of Google AI. We'll get into all of that and what China is currently doing 1:02 as far as blocking mergers and acquisitions of AI tech out of China. So before we get into all of that, I wanted to mention if you have a company that buys data for training AI models. or if you are a creator that generates things like 1:16 podcasts, video, audio, images, and you would like to license that to AI models, I'd love for you to check out my new marketplace, which is called Fiand .com. It's a marketplace that allows you to license your data to AI buyers and allows 1:30 AI buyers to purchase it. This is something I've been working on kind of in stealth for the last few months, and we're super excited to see some of the growth here. So if you're interested, I'll leave a link in the description. It is Fiand 1:42 .com. And yeah, I would love to have you check it out. Okay, let's get started with the first story here, which is that after the midterm elections, AI data center stocks, and I think the overall stock market right now, it kind of did a big 1:54 spike and then it's... kind of crash. So I think it's not, you know, incredibly different from when that happened. But AI data center stocks are starting to fall with Democrats winning both chambers of Congress. We have CoreWeave, Applied 2:06 Digital, and Nebius that are all falling. They all fell pre -market. We had Microsoft and Alphabet that were rallying a little bit. I mean, I think they've kind of all fallen back down. So it's a little bit of an up and down right now. I think 2:16 we can expect that regulators are going to have tighter data center rules. Texas and Georgia that have now went blue, I think are going to face that a lot more. And there was a recent post, you know, I've for anyone that knows I've been living in Arizona for the last five years. 2:30 So our representative out of Arizona was recently tweeting Congresswoman Yasemin Ansari, who was just elected. And she said top of the Democrat agenda on day one needs to be a pause on advanced development regulation for this out of control 2:45 industry and a ban on artificial super intelligence. 2:48 This is coming as Bernie Sanders was recently calling for, you know, 20 year prison sentences. 2:53 for developers working on artificial superintelligence. 2:56 So I just have to do a formal PSA. If you're listening to this podcast, you know, be warned. 3:02 You may serve a 20 -year prison sentence for it. Just kidding. I know this is probably going to trigger people. It's a joke, guys. Elon Musk... 3:09 responded to that tweet and said it's a brave new world you know a lot of people in the comments are saying that China is funding her and they're you know licking their lips or you know China won whatever so there's definitely a major debate 3:20 and I think it's probably you probably won't be surprised my opinion on this is not to slow down AI development. If that's shocking to you, then you might not be listening to the right podcast. Okay, let's talk about what's going on with Anthropic. Actually, I was encouraged 3:34 by this. They signed a licensing deal with the Associated Press right before the midterms. And this is actually the very first deal that they've signed with a digital news publisher, period. 3:44 I mean, obviously, AP is a good one to get started with. It's interesting. This came right before the midterms. And that basically meant that election coverage was happening live inside of Claude. 3:52 I'm curious if this was more like a move where they're like, look, there's this big event that a lot of people People are interested in talking about, let's make sure that we have the data for that particular event in there. Or if this is a long -term thing where they're going to 4:02 now start getting AP news inside of Claude going forward long -term, which I think overall will make the product a lot better. What's interesting to me is that we actually have Opening Eye that has signed more than a dozen publisher licensing 4:12 deals. They have Axel Springer News Corp in the Atlantic and a whole bunch of others. So Opening Eye has been doing this for a while. It's interesting to me that Anthropic is just getting into this. 4:22 It's pretty impressive. I mean, just as a company that Anthropic is, you know, it feels like in features, right? They don't have an image generator. 4:27 They don't have a video generator. They don't have an audio generator. Like it feels like in features, Claude is far behind maybe some of the other frontier labs, like opening eye and Google, but at the same time. Or even like Meta, I'm pretty sure has like some sort of probably 4:40 ghetto image generators. But what's interesting to me is because they excelled at one thing that had a really good use case, which was code, Claude just jumped, leaped ahead of everyone as far as enterprise spending goes. So you can't say 4:51 it was a bad strategy, but I think at some point the company matures, they're going to have to. 4:55 compete on all of those fronts, in my opinion, if they want to say a relevant tool, I know I find myself like I have a subscription to chat GPT is, you know, top tier instead of anthropics anymore, just because it can also generate images 5:07 and do a lot of other things for me. And so for the same price, I'm just going to go in that direction. Okay, Claude code product lead cat woo has just left anthropic and is joining open AI right now. The only reason I bring this up is I think that It's interesting because Wu led 5:23 Claude Coden, also co -work, and she's going to now run product management for OpenAI's core products group. She's going to be under Thiebaud Socio. I think this is interesting just because I feel like OpenAI and Anthropic are kind of 5:34 this revolving door. Anthropic spun out of OpenAI when Dario had some beef with Sam Altman. And it just feels like the employees go back and forth. This is no shocker. Those are the two biggest labs. But it's just interesting to see, you know, the loyalty. Always seem like it's 5:49 really with the company. They move back and forth between them, depending on, you know, who's got kind of the best offer on the table. And for, you know, competition and making these tools better, I think it's probably a good thing. So 6:01 overall, net positive. Okay, as far as Gideon to OpenAI News goes, they just launched a brand new model. It's called ChatGPT Images 2 .6. I was testing it out today. The big things that it can do is that it can do animated GIFs. And it is faster. So it's actually 50 % faster at 6:19 generating images. And it can do SVG exports. 6:22 It can do, I mean, they like, they kind of demo. 6:26 They have like a whole site where they go and demo a bunch of the different tools and features that it can do. The images are apparently, they look better. So it's like more professional of images. They can do GIFs, which is cool. They 6:38 showed like... you know draw a picture and turn it into a gif i mean they already had the feature where you could draw a picture and turn it into an image so i mean i think they're just kind of showing some things you could do but that's 6:47 not necessarily revolutionary compared to what they were doing in the past GIFs are cool, though. 6:50 And I actually was playing around with it. I got it to just animate a logo for me for fun with with for for find just an animated GIF thing. 6:58 The reason why I think this is impressive or important or interesting is because I actually do a lot of animated GIFs and icons on websites. 7:05 But usually they have to go from scratch. I'm like, recreate this GIF or recreate this icon or make a new icon and make an animation for. 7:13 And I'm like sitting there like talking with Claude about how to make the animation. It's like doing it from scratch or doing it from scratch. 7:19 So I think this is cool that you can, if you're just in a hurry, you can one -shot a GIF and probably use this for a lot of different use cases. In their demo website, they were showing if you had a chatbot and the chatbot had a special 7:31 loading animation that you could just kind of throw something on there. GIFs typically are good if it's like a smaller thing because the resolution isn't incredibly high. So it's not like something big that you want on your screen. 7:42 And in fact, when I was recently doing... Yeah, I was actually recently doing some cool animations on my website, Fyand. If you go look at the homepage, there's like a translucent glass circle that like spins. And if you move your cursor over 7:56 it, it kind of can wobble in whatever direction your cursor's at. Now, the cool thing about it is that it's translucent and there's a video in the background that's also moving. So anyways, I mean, not to just sit here and describe it, 8:08 but because it's translucent, there's stuff moving in the background. You can kind of see it. It has like a very a liquid glass effect. It's very cool. Obviously, none of this, you know, none of that feature. Maybe maybe this is not super 8:18 necessary, but it's fun to create. Maybe none of this could be done with a GIF. So if you're trying to do something quick, a quick little animation, the GIF works. But if you want something a little bit more robust like this, I mean, I 8:28 had to go find GitHub libraries and a whole bunch of stuff and work on. building this particular website hero section animation thing. We also have some other interesting news out of OpenAI, though. Greg Brockman said that Enterprise is 8:42 now driving most of their revenue. This is interesting because they were in a kind of a, I don't want to say a tight spot, but with the IPOs coming up between Anthropic and OpenAI, Anthropic was... 8:51 you know, had way more revenue than OpenAI, which was a shocker because OpenAI had way more users. 8:55 And that's because it was all just enterprise revenue. And they were saying, you know, a lot of their customers were paying a million dollars. 9:01 They had some that were paying $10 million. I believe they had some that were paying $100 million a year. So when you had these, you know, a customer that would pay $100 million a year, sure, OpenAI has way more customers, but they were all paying 9:13 20 bucks a month. I think OpenAI has done something. 9:15 good in really pushing for the $100 and $200 tiers. Once they once they kind of rolled out their tools, and they they launched their chat GPT work, that was something that sold me on it. I'd tried the $100 $200 tiers in the past get access some of the like video creation agents, 9:32 it wasn't good. But today, it's it's very good. 9:34 And so I think that they're they're pushing up the revenue and going from $20 to 200. I mean, they basically 10x their revenue. So If they get the percentage of their users that pay to see the value in that and move up, I think that does a lot. And most of those users are going 9:48 to be enterprise. You know, probably your average person doesn't want to pay $200 a month, but employees will ask their company if they can pay that for them. I think it's good. Brockman called the first half of 2026 pretty rough. He credited GPT 5 .6 sole and chat GPT work for 10:03 their turnaround in Q3. And he says that OpenAI could be IPO. IPO ready in 2027. I love this for them. I mean, you would hate to see them kind of stumble and fall out of the races with Anthropic just taking it away. Like we want a lot of competition in the race. We have SpaceX 10:17 that are IPO. I think we're going to have Anthropic go out next. We already have Google and Meta that are working in public, but. I think we want both of these companies to IPO. So hopefully next year will be the year we see that for OpenAI. 10:28 It seems like the revenue, they've got that under control. Basically, they pioneered the tech, but Anthropic, I think, was the winner at finding who the best use cases and who would pay for it. So now they're kind of copying that playbook 10:39 and it seems like they're going to get their revenue figured out again. OpenAI also released some new metrics. They said for usage that ChatGPT search volume is about 20 of Google's. And they've also said they just hit 1 .2 billion users. They 10:52 said they added about 200 million users in 60 days. And they said that they're currently generating 1 .5 billion images a week. They're going to be merging their chat and work. modes i think claude is also doing that like they kind of had 11:05 these different toggles for like talking to chat gpt versus having using work where like control your computer and get stuff done i think you can now just tell the chat to control your computer and get stuff done and it will you know, initiate 11:16 work mode or whatever. I think that's way cleaner. 11:18 I hated the toggles. In fact, Claude was like the worst for a moment where it had like, it had Claude and then it had Claude co -work and then it had Claude code and co -work and code were like slowly merging their features. So it basically became the same thing. There's way 11:31 too many toggles and each toggle had its own conversations in there and it was just a mess up, nice nightmare. So I'm stoked that we're consolidating it, but these metrics are incredible. 11:40 200 million new weekly users in 60 days. OpenAI is on a tear. And how they're measuring this, they said right now they're at 20 % of Google's search. I remember maybe two years ago, everyone was like, oh my gosh, ChatGPT is going to replace 11:53 Google search. And Google was like, no way. And Google rolled out their own chat search features. 11:58 And we're saying, look, our searches are actually up once we've rolled this out. And then, you know, the metrics ended up coming out. And it was like OpenAI was like 2 % of Google search volume. And everyone's like, oh, don't worry 12:09 about it. It's, you know, negligible. 20 % number is probably, is probably pretty accurate. They're, they're counting this. They say basically every time that someone does a chat on chat GPT or like a, you know, they ask it a question and 12:23 it has to search the internet for answers and put it into the question. They're counting that as a search. So if I was just to say like, you know, what's the capital of France and it just knows that because that's in its model and it's 12:33 not going to go do a Google search or, you know, any sort of like internet search, it would just. 12:37 tell me the answer that's not counting which someone you know that technically is like a search thing so it's probably it's possibly even more than 20 of google search volume if you're if you're accounting for just like the the amount 12:50 of questions that are getting asked to chat GPT. 12:53 So yeah, I think this is actually probably conservative how they measured this. And it's, you know, very directly competing with Google on that. And I think at the same time, Google probably has a lot of questions that now are just embedded into 13:03 their model. A lot of people have complained that site traffic is down because people are asking a question to Google. And they're just showing it at the top of the top of the search results with the Google AI snippet. And they're like, we don't need to send you to a website. 13:15 In the past, if you ask what's the capital of France? Google would send you to a website that would tell you all about Paris and you don't really need. I don't think that is super necessary for some of this stuff that feels more like you're 13:26 using Google like an encyclopedia kind of. So anyways, there's there's just two different use cases that we're showing. But yeah, opening eye is seeing some big. Okay, on to Google News. 13:36 We have, I think, especially interesting with the US midterms and the political climate right now. Google is planning to invest $15 billion in Malaysia data centers in 2027 and 2028. They have new campuses in Selangor and Negeri Sembilan. 13:51 Both of those are going to come with a one gigawatt of new clean power. They have a reservoir and a goal to train 800 ,000 Malaysians. So Google, I think, is seeing that we... have to look broader than Google is a global company. But I think 14:06 they're really realizing that the United States is politically not a place that really wants data centers right now. It's a huge fight. And so they're just looking at other places. And I think this is probably if this is probably the goal of some countries. A lot of people say 14:18 that this is China's goal is to drive data centers out of the United States because the vast majority of data centers right now are in the United States by a large margin. So if we can make that really hard to permit and to build. Those will go outside. 14:30 And I mean, it's phenomenal for Malaysia. 8000 Malaysians are going to be trained. So that sounds like, you know, that sounds like a big boost to Malaysia. But at the same time, you know, this Google is a company in the United States. 14:42 And so you would imagine their first bet would have been to build the data centers here if it wasn't so hard and regulatory environment. So for America, bad for Malaysia. Good, in my opinion. 14:53 All right. Let's talk about. companies raising money and sort of on the on the topic of google isomorphic labs which if you don't know isomorphic labs is backed by alphabet they're a drug discovery company they raised 2 .1 billion dollars in may 15:08 but i will say that none of their drugs have you know went to went to human trials yet but they're looking at raising another 5 billion dollars at a 50 billion dollar valuation this company is run by the ceo of of of google ai And they have raised, I think, $600 million in 15:27 2025, $2 .1 billion in 2026. And now they're looking at this new $6 billion round. That would bring their total funding to about $7 .7 billion in 18 months. So Denis Hassibis is the CEO of Google AI. And he says that they're going to target trials by the end of 2026. Previously, 15:48 he said that they were going to do by the end of 2025, and he kind of pushed it to 2026. So that's Demis that is working on that. And he did win a Nobel Peace Prize in chemistry for AlphaFold in 2024. So he's been doing some really incredible things in this space. I'm impressed 16:02 that he's running this company and also being the CEO of Google, you know, Google DeepMind. 16:07 But an impressive guy. They have signed a bunch of different partnerships that I think make them very reputable. player in the space with a bunch of drug makers. I think in 2024, they agreed to a deal with Eli Lilly that was worth about 16:19 $45 million up front and then $1 .7 billion if they hit certain milestones. They also did one with Novartis, which is worth $37 million up front and then $1 .2 billion in milestones. That deal initially converted three targets and it 16:33 was expanded in 2025, so it seems to be on track. 16:36 Johnson & Johnson is also a partner. They haven't disclosed any of their terms yet, but they're working with all of these major companies. One thing that I will say, though, not to throw any shade at them, but the CEO of Eli Lilly, which 16:48 is David or Dave Ricks, was talking about it. 16:52 And he was actually I think he was on a podcast recently on the Derek Thompson's podcast. And he said that large language models are, quote, not particularly good at biology and chemistry. 17:02 So I'm wondering if that has. anything to do with this deal and maybe it not producing as fast because they're supposed to do trials at the end of last year and now it got pushed to the end of this year. So I'm not sure if, you know, he's kind of coming from that angle. The 17:14 other thing that I will say, and we've seen this across every industry is, and these things are getting way better at basically everything and the models are getting, they're improving rapidly. 17:22 So if it's not great at doing some of the AI discovery a year ago, I think it's going to be. 17:28 you know, it's a lot better today. And in a year, it's going to be 100 times better. So I have a lot of hope in this. And we've seen a lot of, you know, really positive impacts that AI and drug discovery has done, even if it's kind of 17:39 some of these smaller things, obviously, anthropic investing in the wet lab in San Francisco is, you know, proof that this is an area that a lot of people are interested and pursuing. So I'll be excited to see if they can come up with anything. 17:50 There is one of the areas I'm the most excited about, but I'll definitely keep you guys updated. 17:54 All right, finishing up the podcast, let's talk about what's going on with China. They've just blocked SoftBank's $5 .4 billion purchase of ABB's robotics unit. I think China is about 30 % of the unit's business. What was interesting 18:07 to me, though, is that this is a deal that China blocked, but SoftBank agreed to this deal in October of last year. The European Commission cleared it. Their robotics unit had about... 18:19 $2 .3 billion in 2020. 24 revenue had about 7 ,000 employees. China is their largest market. 18:25 But there is one report that cited concerns that China's they're basically worried that this robotics data could train overseas AI models, especially open AI. So China's worried that they are the biggest customer, they're worried that the data 18:37 is going to go over to the United States. South Bank shares fell about 5 % in Tokyo on Thursday after this news broke. And I will also say that neither the rejection nor the review details have been confirmed in an official announcement. 18:49 So this isn't something that's been officially announced. And I believe, I'm trying to remember exactly what it was, but I think China's antitrust review in February did clear this deal, according to JKN, which is out of Korea. It's a news outlet 19:01 out of Korea. But apparently it kind of ran into some trouble with the NDRC's National Security and Supply Chain Assessment. So inside of China, there's some people that approved it. There's some people that disapproved it. I mean, it's 19:12 going to be like the United States. It's hard to get data and AI companies in and out. and doing acquisitions because of the national security interest that these countries are holding. Having a competitive edge in the data side of things 19:25 and also in the model side of things makes a really big difference. So it seems like China is blocking that. They don't want to lose any data that... you know, that could be used by the United States. But maybe it doesn't matter 19:37 because maybe the United States is banning superintelligence altogether and they're going to run away with it. Who knows? All right, I'm not salty. You're salty. Okay, guys, thank you so much for tuning into the podcast. I hope you enjoyed this episode. 19:47 If you did, please leave a rating and review on the show. It helps the show a ton, helps me know what you guys appreciate on the show. And otherwise, if you're interested in buying or selling data for AI licensing, go check out my startup. fiund .com. I'd love to help you make 20:05 money or help you acquire some high quality data. 20:08 All right. Hope you all have a fantastic rest of your day and I'll catch you in the next episode.
Community discussion