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. Today on the 0:17 show, we're talking about OpenAI, which has just launched ChatGPT Wallet. It's an agent that, well, essentially allows agents to purchase using Visa, MasterCard, and American Express. We also have Google, which is opening pre -orders for 0:31 Gemini -powered... glasses. We have Cerebrus, who has just posted a record Q3 profit, and they are guiding Q4 to as much as $330 million. I think the entire sector there is going to be seeing some incredible growth. OpenAI is backing a bipartisan bill to tighten Chinese AI chip 0:51 export controls. We have a whole bunch of nonprofits that are suing OpenAI, Google, Anthropic, and Meta, and they want them to halt all frontier AI development. Pretty wild story right there. 1:03 SpaceX is planning more than $74 billion in AI supercomputing campuses in Texas and Arizona. 1:10 We'll cover some of those deals. And we have a big raise, $1 billion Series A from Merendil. 1:17 And they have been led by Kleiner Parkins and Clearlake. So top tier VCs putting a billion dollars in for a new AI research lab. So exciting to see some new stuff. coming down the pipe. 1:28 I'll get into all of that. Before we do, I wanted to mention, if you want to grow or scale your business, if you have one or a side hustle with AI, I'd love for you to go check out the AI Hustle School Community. Every single week, I post in 1:41 there breaking down different side hustles I have, different businesses I have, all of the AI strategies, tools, and processes I set up to grow and scale my business with AI. We have over 1 ,300 members, and it is an incredible place to learn from other people working in the 1:55 industry. what's working right now, how to grow and scale my goal. It's $19 a month, but my goal is that you'll be making $10 ,000 a month in additional income on your business or on a new side hustle with a lot of these strategies. So if you're interested in it, go check out the 2:11 AI Hustle School community. I'll leave a link in the description to it. Let's get into the episode. First thing I wanted to cover today was just the new ChatGPT wallet that has dropped today. This is a cool feature. And for those that don't know, there's a ton of new tools that 2:25 have been Coming out recently, we have GrokBots, we have MetaDot, or we have OpenAI Dots, we have Meta's Muse, a lot of these different agents that can go around and do things. And one of the things that people are excited about is their 2:39 ability to go and actually make purchases. So if you are a US Pro subscriber, you're going to get basically it's like a quote unquote tokenized card. It needs biometric approval. So it's going to use like, you know, the Apple Face ID or the 2:52 fingerprint for every single... purchase, but you can go tell your dot to, you know, go buy airline tickets, go buy. purchase tickets. And it's going to actually work with American Express MasterCard or Visa. So it's going to have actual 3:03 cards that can go make those purchases. Now, those actual cards, none of the details of those cards are going to get passed along to the actual company that's processing the payments. And so with that, it's less risk for fraud, like an 3:15 agent, or I guess less risk for an agent accidentally leaking your credit card details. That's what they're trying to get around here. And it's interesting because I've heard a lot of startups working on this. particular problem. But it feels like 3:26 OpenAI just decided this was something that they had to do themselves. You obviously have to be over 18 years old in order to do this. And OpenAI is not taking any cut of wallet transactions at launch, which is different than Meta's Muse. 3:38 This is literally what I've heard Mark Zuckerberg say his monetization method for MetaMuse is they'll take a percentage of all transactions that go through the Muse platform, although they do have a $20 subscription tier that's rolling out for 3:49 some of the more advanced features. So Perhaps Zuckerberg and Meta will switch gears and focus a little bit on that. Partners right now, they have a bunch of payment providers. They have Aiden, Checkout .com, Stripe, and then a couple 4:01 of fraud firms. They have Forwarder, Risky Feed, Sardine, and Signified. All of them are working on this. So overall, I think this is an exciting step. It's kind of funny because we also have United Airlines and we have eBay and we have 4:14 a bunch of other, you know, Amazon that all famously block bots from purchasing. software services on their platform, which again, funny because I'm just going to probably use other services if I can't, you know, use AI to help me get my 4:27 job done faster. But overall, this is going to be an interesting moment. Google has just opened pre -orders for Gemini powered glasses. They're doing this with Warby Parker and Gentle Monsters. 4:38 So they have Samsung working on this. They have these frames. They started about $400 the way about. 41 grams and they go on sale in stores and online on December 10th. So just in time for the Christmas rush. If you go and look at their website, it's incredibly well made. They 4:55 have shaded glasses. They have just clear glasses. 4:58 They have a camera. So it's basically just like what Meta has done, but they have a camera. And I remember being at Google I .O. a couple of years ago and seeing them demo some of this. 5:06 But basically, if you're driving with. or you're, you know, going down the street with your glasses, you can talk to it. You can say, hey, Gemini, take a video. Hey, Gemini, you know, what am I looking at? Hey, Gemini, give me Google, give 5:18 me Google Maps to where I need to go. Or if you're, you know, at a restaurant and the menu's in another language, maybe you're traveling, you say, hey, translate this for me and it can translate. You can ask it any questions about what you're doing. 5:29 So, I mean, overall, this is a lot of what Meta has been doing. They look like incredible glasses. 5:35 They do have a bit bulkier of a side, but... 5:38 Honestly, all in all, it's like basically like Meta Ray bands and all the other meta collaborations for glasses. They have speakers built in. They have cameras built in. It's Google and Samsung that have built these things out. They look great. 5:49 And I think they have a lot of I think Google has a leg up in this compared to Meta, just in that they have Google Maps and a bunch of other Google specific. franchises that will help sell more of their products. So I'm really excited 6:01 for this. You can pre -order them now, but these are going to be ready in time to be sold for Christmas. So I think these will be a big seller. 6:07 I'm excited to see how much market share these start pulling away from meta or maybe growing the overall pie as people start doing. facial computing. Very wild times. All right, as far as how are data centers, how's computing going right now in revenue regards? Cerebrus has just 6:25 posted a record Q3 revenue. Q4, they're putting that up to $330 million. They've raised that. 6:31 Their core revenue rose 70 % to $230 million, which is higher than estimates. And I think they said that they just signed a deal with a new top three hyperscaler. I think it was kind of leaked that it was Microsoft that signed a deal 6:45 with them. They already have a deal with Oracle. 6:50 And I believe they have another deal with one of the other big labs as well. So they're doing quite well. And it's exciting to see. We saw a lot of the, we saw Forweave and other people get hammered in the stock market with NVIDIA getting hammered in the stock market. Even before 7:03 the whole Jensen thing where he got kicked out as CEO, NVIDIA was struggling. If you look at the last month, they're down about 25%. It wasn't, it didn't just happen with Jensen. It was kind of happening before that as well that they saw 7:16 their stock slide. People were worried that the market was cooling. But it's impressive to see some of these companies are literally doing better than expectations. So I think we can confidently say that if you're in the compute industry, yes, 7:28 it's hard to build data centers. Yes, it's hard to get a regulatory approval. There's a lot of obstacles. But if you get those data centers up, there are definitely going to be buyers for them. OpenAI right now is backing a bipartisan 7:38 bill to tighten Chinese AI chip export controls. 7:42 This push is targeting the year end defense bill. 7:46 And it is the same week, of course, that Jensen Wang just got kicked out of, or I guess he technically resigned, but probably was strongly requested to resign by the board because he had an affair with an alleged Chinese spy in Singapore. And, 8:02 you know, that spy was executive at a company who ended up buying $7 .5 billion worth of NVIDIA GPU chips. And because of that, You know, it's obviously potentially a national security issue. 8:13 Potentially there's some legal ramifications. 8:15 Jensen's going to have to go testify in front of Congress. I think NVIDIA didn't want the negative press related to those decisions. And so Jensen has resigned and we'll see what happens there. 8:26 I would predict that if NVIDIA, if the stock and the company tumbles over the next couple of years, they might bring him back. You know, but it also all depends on what was, if anything, national security. anything of national security 8:40 importance was leaked, what kind of issues there are, what kind of legal ramifications there are to all of this. So I think that's what we'll be focusing on. But with all that being said, opening, I told lawmakers this week that they're 8:51 supporting the new bipartisan bill that's now tightening China's access to all of these chips, because as it turned out, the AI company that was buying those $7 .5 billion of chips from Jensen, they were most likely just being sold to companies inside of China. There's a lot of 9:05 rumors and leaks to that. essentially to that there's a lot of nvidia chips that are in china that no one knows how they got there there is an ai oversight or overwatch act from elizabeth warren and jim banks that is expanding congress's 9:19 role when chip export rules change the chips security act would require tracking of advanced processors and alerts on suspected diversions so if those chips are going somewhere they're not supposed to they're basically tracked there's like tracking devices built into this and geo 9:34 you know, geo shutdown. So I'm sure people will find ways to hack that. But it's pretty wild where that, you know, kind of where that goes. 9:41 The people that support this right now have are trying to attach the measure to this year's national defense authorization bill. So trying to get added to the national defense budget. Anthropic left a tech trade group in September over It's 9:55 pushed to block chip control legislation. It makes a lot of sense why OpenAI, Anthropic, and all the others would want to block those sales. 10:02 They don't want more competition out of China. 10:04 It makes sense why NVIDIA is financially incentivized to make those sales because they want more money. 10:09 And the US government obviously wants to remain in control and in the lead on AI over China. 10:15 So it makes sense why those budgets are all passed. 10:17 We'll see what happens if this bill is successfully attached to the end of year spending. There's a whole bunch of nonprofits right now that are suing OpenAI, Google Anthropic, and Meta. I mean, this honestly almost wasn't a story because it 10:30 was so ludicrous, but it is so ludicrous I have to cover it. And I just wanted to talk about this. Basically, they're trying to sue them to halt frontier AI development. It's called Public Citizen, and there's three other groups. And 10:39 they're all citing California's unfair competition law to seek a court order, stopping work on models, you know, Chachapi, Gemini, and Claude. It sounds pretty ludicrous, but I think essentially this comes down to... publicity these ai i mean like 10:53 maybe they're ideologically believe that ai could become dangerous or uncontrollable whatever right like all this stuff we've we've seen kind of peddled by random people that end up getting million dollar jobs at safety labs afterwards 11:05 but at the same time you have to also imagine this is probably a great publicity stunt for these for these non -profits and they probably get a bunch of donations from them and you know who knows who funds them but it's obviously a lawsuit that is not going to produce much, but 11:21 maybe they're trying to have some sort of favorable ruling that could have some sort of legal precedent, could give courts power to restrict future AI development. And I think if nothing else, this lawsuit is going to create a bit of pressure 11:32 for some stricter AI regulation. So that also is potentially the direction that they're... 11:36 hoping to push this into. SpaceX right now is planning more than $74 billion in AI supercomputing campuses. They're going to do two states, Texas and Arizona. They have a roughly $50 billion that they're going to be spending in Cochise County. That campus is going to scale to six 11:53 gigawatts, while a $24 billion Starbase site is going to run in its own gas fired power. So both Texas and Arizona are going to be spinning up some. massive data centers. And I remember that Elon recently talked about how he was able to raise funding for these big projects. And 12:12 he said, look, the only reason we could raise funding for this is because we have Starship's launch revenue and we have Starlink and the Starlink revenue that we could say, look, we actually are making money from all from our company, SpaceX. 12:24 So they're able to get it, he said, for companies that are not. don't have a lot of revenue, it's very hard to get these deals. And so that's what I was alluding to when I was talking about Cerebrus earlier. Yes, they have some incredible revenue, 12:35 but getting more loans approved right now is pretty tricky. It's hard to get some of those loans. The SpaceX project right now that they have going on is about $5 ,500. construction jobs and 1000 permanent jobs that will be in Arizona for that project, more than 500 permanent 12:52 jobs at Starbase. So probably 1500 jobs for the Arizona project, and then 5500 construction jobs to help build it out. It's a 747 acre Arizona campus, they're gonna have four different phases, which is usually how these big compute campuses 13:08 grow. It's like 1 .5 gigawatts. And it just keeps growing until it gets to six gigawatts by 24 gigawatt hours of battery storage that it's going to all be backed by. And the Cochise County's Planning and Zoning Commission and the Starbase 13:25 City Council both held hearing or both are going to be holding hearings on December 9th about these different projects. SpaceX ended Q3 with 1 .7 gigawatts of name plate compute and are targeting at least 10 gigawatts more in 2027. 13:40 Just think about that for a second. They have 1 .7 gigawatts. They want 10 gigawatts by the end of next year. These projects are obviously big parts of that, but it is, I mean, it's going to scale like crazy. And I think we are going to use all of that compute on the projects and 13:57 tools that currently exist. I wanted to wrap up the episode today talking about a new AI model because I feel like we talk about the same Google, Meta, Amazon, not even really Amazon that much, but Google, Meta, OpenAI, Anthropic. And SpaceX, 14:09 sure, as far as the big leading AI companies, but there is a new one called Mirandil. They just raised a billion dollars. This is a Series A. It was led by Kleiner Parkins and others. 14:18 And this is a frontier lab. They're building models and agents that do AI research. They have some coding model. They have a researcher pre. 14:25 like for science and stuff. This was pre -trained for under $10 million, which is phenomenal. So obviously they have some top tier talent and I think it's pretty impressive. You still can, apparently, with some great talent, go and create 14:38 a model for something very specific. Their first model is called Marendal Base. It's a coding model, according to their company. A single researcher worked with agents. and pre -trained it from scratch in 3 .5 months using less than $10 million 14:52 in compute. Raising a billion dollars, they probably were able to say, look, if we could get this done for $10 million in compute, imagine what we could do for a billion. And so this is how they were able to raise money. I'm excited to 15:03 see more competitors in this space and more people working on all of this. So I'll definitely keep you guys up to date. Again, if you haven't already, I would love it if you would join the AI Hustle School community. If you're looking to grow, 15:14 scale your business using AI tools. And as always, if you would be so kind as to leave a rating and review on the show, it helps it out a ton to be found by more incredible people like yourself. 15:25 Thank you so much for tuning in. I'll catch you all in the next episode.
Community discussion