Limitless Podcast - THIS WEEK IN AI: Google's AI Laptop, Everyone Goes to China, Thinking Machines

Episode Date: May 14, 2026

This week, we saw Google’s new AI-focused laptop, Gemini’s growing role across Google services, and Google’s reported work with SpaceX on space-based data centers. We also cover Isomorphic Labs’ $2.1 billion raise, a Chinese mecha robot, and new developments from Thinking Machines Labs, Anthropic, OpenAI, and the Sam Altman-Elon Musk disputes.------🌌 LIMITLESS HQ ⬇️NEWSLETTER:    https://limitlessft.substack.com/FOLLOW ON X:   https://x.com/LimitlessFTSPOTIFY:             https://open.spotify.com/show/5oV29YUL8AzzwXkxEXlRMQAPPLE:                 https://podcasts.apple.com/us/podcast/limitless-podcast/id1813210890RSS FEED:           https://limitlessft.substack.com/------TIMESTAMPS0:00 Googlebook6:48 Google’s SpaceX Deal10:51 Isomorphic Labs15:19 Robot Giants17:16 Thinking Machines21:45 Anthropic’s Voided Deals25:19 China Summit29:56 Leopold Next------RESOURCESJosh: https://x.com/JoshKaleEjaaz: https://x.com/cryptopunk7213------Not financial or tax advice. See our investment disclosures here:https://www.bankless.com/disclosures⁠

Google just launched the first laptop built for AI, and it's everything Apple said Siri would be, and a lot more. It's pretty awesome. The signature feature is this thing called a magic pointer, where you can wiggle your cursor over a date in an email, and Gem and I will schedule a meeting. You hover it over a living room and a new couch, and Gemini will composite the two together and render what that would look like. You can ask it to plan a family reunion, and it builds these live dashboards with the flights, hotels, and countdowns all built in natively to this new laptop, they're calling the Google Book. If you'll remember from 2011, there was a little thing called the Chromebook that revolutionized how we use laptops forever. Today, we have some new innovations from Google that are going to do that once again for the AI era. This laptop is pretty amazing. And it's one of many things that was announced at Google's I.O. conference yesterday. This was surprising to me for many reasons. Number one, we are a week away from Google's flagship I.O. event where we were expecting all these major AI announcements. And it seems like a week before, they're giving us like three to four new updates. This is just the warm-up. Yeah, it's just the warm-up, and I'm actually quite impressed.

So there was a sentence that was repeated twice across all of these announcements yesterday by Google, which was, we are moving from an operating system to an intelligence system. Now, what that means is they're upgrading a traditional computer operating system to something that is more compatible with AI agents or AI models in general. It's a new way to interact in this new AI-powered world. Now, if that sounds familiar, that's because OpenAI announced something similar with the rumors of their upcoming phone, and we did an episode on that.

But anyway, back to Google. What did they release? So on the hardware side of things, there's a brand new laptop, which is engineered from the ground up to be suited towards Gemini, their flagship AI model. They also released a bunch of new software,

which includes something called Gemini Intelligence, which is basically their Gemini model, but actionable across all the different apps, tools and products, software products, specifically that Google has. So if you understand the Google model, they're vertically integrated, They have the model layer, the GPU layer,

and they have all this amazing distribution through Google Maps, G Suite, and a bunch of other, like Gmail as well. So the point is they created Gemini intelligence to work across all of these things. So if you wanted to order on DoorDash or if you wanted an AI agent to go scow your email and figure out what books you need to order

for your syllabus at college, it can all do that seamlessly without you needing to prompt it on its own. And then on the ecosystem side, one thing that I found interesting is they've now made it incredibly easy to port over all your Apple products and software data into the Google ecosystem, which is something that is very Android, very open source,

which is what Google is, I guess, kind of known for. But I'm excited about this. Now, the flagship product, I know you said it was the pointer, Josh. For me, it has to be the laptop, this new Google book. It's pretty impressive. The Google book is the national extension of the Chromebook. Now, if you'll remember way back in 2011, 15 years ago, Google released the Chromebook. This was a $200 laptop that was available basically to anyone.

It was the first time that a device that powerful was accessible to the rest of the world. And what happened was is that the Chromebook kind of ate the software stack. And they tried to make everything exist in the browser. So the Chromebook, if you remember, you can't actually download applications. Everything ran within Google Chrome. Now, the natural extension of that is happening again with the Google book, where instead of the browser cannibalizing software, the AI is now cannibalizing the browser. And what we have as a result of this is this thing that Google is calling Gemini Intelligence.

And now Gemini Intelligence is basically that baked in operating system into this new device. The Chromebook itself is fairly beautiful. It looks somewhat similar to a MacBook. If you're familiar with the MacBook Air, MacBook Pro, it's kind of a hybrid between those two. They're pricing it at about $200 to $500. So it's very competitive with the MacBook Neo, which is sitting just above at $600. And I have a feeling that a lot of people are going to be interested in this because of the native AI features built. in. Now, what you'll notice throughout this video is a lot of these features are actually baked into

being an extension of your phone. This currently works with Android phones. So if you are a Android user, this is probably an incredibly compelling product because in a way, they're building the Apple ecosystem. They're building everything that Apple said they were going to do, but failed to deliver on in terms of software. And it serves as this really good companion to Android phones. Now, if you're an iOS user, this is probably more of a fun experimental laptop, but the MacBook Neo is still looking like it's a little bit more compelling than a laptop like this. The other exciting thing is the cursor thing that you mentioned. What I like about this is there's a lot more of a human feel about this feature.

So instead of using a cursor to click on things and then typing something to action it or clicking on various different buttons, this is more of a point and shoot type of feel where you can circle a particular image or point at the crab as you're seeing on this screen today, and then you can voice what you want the Gemini AI assistant to do things for you. Now, this is part of a broader theme that I'm noticing over the last couple of weeks where certain AI labs, this week it was actually thinking machine labs, started by Miramorati. They've started producing these new types of AI models that can ingest and produce voice that understand what you see.

And it's not just LLM-based. It's not just word-based. And I like that Google, who has been a leader in Omni models for a while, is able to kind of put this into their products or embed this into their products in a very seamless way. Like, this looks fun and something that I would use. To your earlier point, though, a lot of people are just like mass users of Apple's ecosystem and products. It seems like Google is aware of that, and they're making a concerted effort to focus on this. So I've got like a list here that I wrote up, which was like the Apple Target list. So for the MacBook, you now have this Google book or this

new AI laptop. For IMessage and iPhone, there's like a massive lock in here. but now you can do like a bunch of wireless transfer of this data. AirDrop is something that Google kind of hacked and prototyped about a month ago. And so I feel like whilst Apple is kind of dwindling and twiddling their thumbs, Google is slowly eating into their market share. I mean, Apple signed this billion dollar a year partnership with Google to license their Gemini model, and there's no signs of them creating their own base model. So I feel like Google is now taking the moment a week ahead of their major announcements

to eat into Apple's market share. Yeah, I mean, it's pretty clear that Gemini is pivoting. from an app to a full system layer. I mean, Gemini is now becoming the operating system of this new device layer. And these are the first series of devices that we're really seeing that are AI first. And a common conversation that we have on the show is what the new AI hardware stack will look like. What does that AIOS look like? And this is one of the first real physical manifestations of that being shipped out as a teaser, again, to Google I.O., which is happening properly May 19th and May 20th. So we will be here to cover all of the news as it relates to that as soon as it drops. But this was

not the only huge news coming out of Google this week. The second involves our pals over at SpaceX, who seemed to be working together now for AI data centers in space. The trend that everyone has come to know in love, we know Anthropic last week. They just signed with XAI, or SpaceX AI, I should say, to fulfill Colossus 1 and start using it for their own training data, with the option to send data centers in space. Google is now hopping aboard. This is a pretty big deal in the world of SpaceX AI and the move to outer orbit with data centers. Yeah, Elon Elon has been busy, basically. He's realized that the demand for GROC, XAI's product specifically, has been lower than he thought.

So he's pivoted a lot of the business lines. You mentioned that he signed that deal with Anthropic last week, where he's basically released out his entire Colossus 1 data center to them for inference. And so he's making about $5 billion from that. And now he's, the rumor is, or the breaking news here is that he's signing a new deal with Google, whereby he is going to launch a bunch of their TPUs, which is their own version of their GPUs, out into space. It seems like Google and Anthropic are realizing that SpaceXAI has some unique advantages when it comes to scale that they can't achieve. Google specifically in this deal is a cheaper highway to space. Now, it does say in the announcements that they're looking at other providers to do

the same, but I think that that's basically a very weak hedge. SpaceX is obviously the cheapest way to do this. And this isn't birthed out of novelty, by the way. It's not just a trend thing. Sundar Pichai about, I don't know, six months ago at this point. point announced that they're prototyping and already working on radiation-resistant TPUs to put into outer space, but they need a way to get there. So it's interesting that Elon is willing to allow them to pay the toll. We mentioned this on previous episodes. He'll be the Tollmaster. And then there'll be competitive out in space itself. Now, what I noticed, Josh, is over the last two weeks recently, there seems to be a kind of AI alliance forming amongst

a bunch of these different companies. It's between SpaceX AI Anthropic, Tesla, Google and cursor, and it's incredibly fulfilling for all parties involved. So I wrote a list here, which is like Google gets cheap access to space, right, plus the infinite energy that they can harness from the sun through solar power. Anthropic gets 300 megawatts of inference compute from Colossus 1. SpaceX gets around $5 to $10 billion between the deals that they signed with Anthropic and Cursor. And Cursor gets a flagship coding model because they get access to all the computer that they couldn't afford necessarily themselves. So it's a very simple, biotic relationship between these companies, and there's one noticeable AI lab that's left out of

this, which is, of course, Open AI. Yikes, it's our rough week to be Open AI, although it seems like in the court cases that we're going to get into a little bit later, they're doing all right. The polymarket has been going up for Open AI chances of winning the case, but some key facts that I think might be noteworthy and important. EJS, if you'll remember, Google already owns 6.1% of SpaceX, the company. So Google is a very large shareholder in this company. They have a vested interest in making it work. If you'll also remember, Google has Project Suncatcher, which was announced in November of last year that we actually covered on an episode, and that is Google's existing space-based machine learning initiative. So they've been working on this technology for a while.

They're also talking with these other rocket launch companies and working with Planet Labs, which is another satellite designing company to actually design the satellites that they're going to go in space with. And then as SpaceX goes public, one of the clearly stated missions that they have is to be the picks and shovels provider for this next space race. So there's a lot of converging energy that points to the fact that Google has been working on this for a long time. They just kind of officially penned a deal with SpaceX here to make it so that SpaceX is more inclined to want to send them and make sure that they reserve that space on these ships that I'm sure is going to be very precious given the amount of tokens that were generating. And overall, really big news for Google, really exciting for SpaceX AI. I mean, all of these things are pointing to an absolutely gigantic IPO. There's no large tech lab outside of OpenAI that doesn't stand to benefit from SpaceX. and I think that's going to be huge when they do go public sometime in a few months. We're getting

pretty close. Now, there's more news on the Google side. Google is just having themselves an unbelievable week, this time coming in the form of a company that most people may not have heard before called isomorphic labs. It raised $2.1 billion, which is a huge number, and the reason they were able to get that is because Demis Sasabas, the CEO of Google DeepMind, he is at the helm of this. So they spun out, they raised a huge amount of money, And from what I understand, this company is pretty badass. This is like the biology copy of Google DeepMind. Is that right?

Yeah. So one of the major breakthroughs in AI that Demisisizizab is something known as protein sequencing. So a big problem in science is we can understand and identify proteins, but we need to understand the genetic makeup of these proteins because they regulate how our bodies work. And that can be attributed and applied towards cures to, specific diseases. So what he thought of was maybe if I apply this AI model to protein sequencing, it'll be able to guess what the protein will eventually do. And he nailed it. It's called Alpha

Fold and AlphaGo. These are two separate AI models that basically come together to produce that exact result. And what they've ended up doing is it's being used by, I think, 300,000 frontier scientific researchers all over the world globally for free, by the way. And they've discovered a host of new molecules which can potentially be formed into applicable cures. And we're talking about cures are like major diseases like Alzheimer's and cancer. So the next step then is how do we actually create the drugs and distribute the drugs to the mass audience, you know, the mass amount of patients that require these types of cures. That is isomorphic labs. And up until this day, it's kind of been internal to Google. In fact, they acquired the company a while back and it's just

kind of been incubated for a while until this week where they announced that they are spinning it out and raising $2.1 billion. It's being led by, I believe, Thrive Capital and their flagship model, which has a very weird name, which I'm not going to try and remember. It is ISODDE, very, you know, very catchy term there, is basically able to identify a host of new molecules. They've already proven that already. And the idea with this money is they're going to scale this out into human trials and start releasing hopefully the first bunch of cures over the next couple of years. I like the name ISODDE, the isomorphic AI drug design engine. It's a mouthful, but it gets the job done.

And tell me if this is wrong, I was trying to figure out what this company does in summary for people who are novices in the world of biology. And my understanding is that proteins are the machines that run pretty much every process in your body, and then drugs work by binding to specific proteins, neither turning them on and off. And that's the entire thing. They're either on, off, or they're blocked. So what isomorphic labs does is it uses AI to predict exactly how a protein folds into its 3D shape. And you mentioned protein folding.

that actually won Demis Sissabas, the Nobel Peace Prize in 2024, or the Nobel Prize, not the Peace Prize, Nobel Prize, Nobel Prize in Chemistry for Alpha-Fold II. So these are people who have really established themselves in this industry and now are using AI to apply it to the frontier. So using these AI models, they then design a molecule that's custom built to slot into every one of these proteins, kind of like a key into a lock, and that unlocks the ability to turn them on, turn them off, or block them in ways that can basically manipulate the human body into doing a lot of pretty sci-fi-esque things. It starts with blocking diseases but turns into,

I mean, eventually allowing us to do things like see infrared light with our eyes and kind of turn us into superhumans. So this is a really compelling technology that is being led by the guy. Like if there is a guy to do this, it is Demis and Co. And this $2 billion investment is hopefully going to really accelerate the progress towards this. Yeah, I mean, he's been obsessed with this for decades down. So it doesn't surprise me that he's, the guy to birth this product.

To give an even simplified version of what you just described, Josh, it's like, when you look at a disease, it's usually bad because it can plug into like a cell, a hole. Imagine like a keyhole in your body, right? And it causes havoc. What this model does is it identifies the keyhole. It can create a blueprint for a protein that can fill that keyhole and block the bad guy from connecting to that keyhole. That's like a super, super dumb down version.

And it sounds simple when I put it like that. It is incredibly hard to do because of all the molecules that need to be put together in the exact same formation and then to mass produce that as a cure. So what they're doing here is pretty remarkable. But moving on, Josh, in the land of sci-fi, that story kind of like hits pretty high up. But what if I told you that there's a real-life mecca that you can buy for $50,000, a mecha robot that you can jump into and walk about New York City in? How do you feel about that? I used to watch Power Rangers. I feel like this is something you'd see in like Power Rangers where

they're just jumping into the suit and walking around as robots. This is an interesting launch for people who are listening. We're looking at literally a Mecca robot and there's this man who is climbing up into the cockpit strapping himself in and is now moving around in what I assume is what like 20 feet tall, 25 feet tall. It weighs 500 kilograms and it is a really impressive looking robot. What I find most interesting about this is the concluding sentence of this post, which says, please, everyone, be sure to use the robot in a friendly and safe matter. As the robot leans back on his hind legs and now walks on all fours. This is pretty crazy, sci-fi-esque. I think this is a testament to China being, you know, pretty cracked when it comes to making robots, but also a testament

to kind of like what the devices of the future could look like. This is a pretty weird one. Yeah. So this thing weighs well over a thousand pounds. And I was thinking why on earth would someone like use this thing? And I think the answer is simply because just because, right? So China has for a long time led the robotics industry, not because they have like an amazing lead in robotics models specifically, but the hardware. They're always nail it and they can scale it out pretty massively. And they've been experimenting with this stuff for a while. So where we sit right now is this robot costs you about $50,000. I saw a hilarious tweet yesterday, which was like, what do you want more? Do you want to put a deposit down on a Ferrari or do you want to

buy this robot? A lot of people replied, I would rather buy this robot. The friendly thing is weird because in this demo, they literally show this robot knocking down a brick wall. So I don't know what they're trying to get at here. But moving on, we have some news from thinking machines labs. Now, if that's not a name that you've heard for a while, that's because this is a company that was hyped a lot when it was announced, and it was Miramirati, which was a former co-founder and pretty high. She was the CTO at OpenAI who left to found this startup. She raised the biggest ever CERRAN. I believe it was like $15 billion or something crazy like that. And then it was radio silence. We haven't heard anything from them for about a year and a half to almost two years.

And finally, this week, they released their new model. And I'm pleased to say that it's not like every other model that we speak about typically on the show. It's not an LLM. It's something pioneering an new called an interaction model. Now, this model is unique in one specific way, which is it is, what's the term? I guess it's like uni model. So the way it works is it can hear what you hear, see what you see, and it can speak to you all at the same time. Now, typically, when you speak to an AI model, when you interact with an AI model, there's a harness around it. It kind of acts like a walkie-talkie. So if you notice, if you speak to Open AI, chat GPT, if you interrupt it, it kind of stops and it forgets what it was saying.

And it kind of, it's like, the visual analogy is like as if it has like earmuffs on. So once you start speaking, it kind of like, once it starts speaking, sorry, it doesn't hear what you say. But this particular model, as they demonstrate here in this demo and we're going to show you a few more, basically is very universally interactive as a human. And the point that they're making here is it's a new type of model that can work with you. It's meant to be an AI agent that feels and thinks more human. and that you can actually work with in your daily workflow or whether you're using software

or whether you're doing things in real physical life. So it's a pretty cool model. Yeah, I think one of the constraints with a lot of models is that it feels like that walkie-talkie where you can talk to it or you can listen but you can't do both at once. This fixes that. And it also solves a huge question mark that a lot of the industry had, which is what is Mira and Ilya, two of the co-founders of Open AI that left to build their own companies. We have thinking machines. We have safe superintelligence. We now have the answer to what one of them building, and I have to be honest, it felt a little underwhelming. It is a,

they call it an omnibodel, and it's very impressive by nature. It processes audio, video, and text simultaneously, and it takes these, they pride themselves on 200 millisecond micro turns, meaning it can listen while it talks very, very quickly, and respond very quickly, having been able to interrupt more naturally. Meta,

a few hours after, ship something somewhat similar to this, after the team from thinking machine said, this was kind of impossible. Yeah, the meta-AI voice conversation power, so literally within hours of Mira's company producing this and saying we created this unbelievable breakthrough, meta was like, no, no, no, not so fast. We actually figured out the same exact thing you have. And to me, this is a little discouraging because Mira and Ilya were supposedly

working on these moon shots, these things that would really push the frontier forward. What we got is just a, sure, it's a piece of frontier technology, but at a very narrow context of a use case that I'm sure other companies will commoditize. So it's interesting, nonetheless, it is best for today, but I know a lot of companies specifically Open AI are really working hard on that voice interface, and they really want to perfect it because they have those hardware devices coming, and the primary interface is likely going to be voice, and that's going to be a huge focus. So for thinking machines to continue progress down this route, they're going to have their work it out for them. I think it's just hard, man. Like, as you mentioned, like the Omni model specifically

isn't unique to thinking machines. Google's pioneered it, Open AI's working on it, Anthropics doing the same thing. Also, the model size of this. this thinking machines model is 12 billion parameters. Now, if you're familiar with like some of the more bleeding edge models, they're around 1.5 trillion parameters, if you want to believe the rumours around Claude Mithos, right? So the point is, parameters or size does matter, and you need to embed intelligence as much as you can into the actual foundational model before it actually ends up becoming useful. That being said, these demos do look encouraging. And if this is something that they can achieve with such a small model, I look forward to seeing what a model two to,

maybe five times bigger than this actually operates out. So, you know, I'm still going to cross my fingers and hope that they produce something cool. But the point is, like, it's hard for smaller labs, even if you have a famous co-founder and you raise a ton of money to actually do something impactful. The bigger AI model labs are like completely running away with it. And it's hard to catch up at this point. And speaking of this hype, obviously there's a lot of investor demand to try and get access to some of these top model labs, namely the ones being anthropic and open AI. And there was a specific story this week concerning Anthropic where it all started off with this one tweet from this lady called Ash Aurora, where she goes, simply brokering an Anthropic secondary

deal made me more money than my entire net worth from working in my 20s. This is insane. Now, what she's referencing is Anthropic has raised a series of rounds equating to like hundreds of billions of dollars. And that is the equivalent of investing in Anthropic before it goes public. Now, of course, these rounds are reserved exclusively for specific partners. It's not applicable or accessible to you and me or the general retail public. That's why it's private versus public. Now, what can happen is there's an entire secondaries market where the people who are contributing or investing in Anthropic rounds who have been blessed with the crown can basically syndicate their allocation out to regular people who Anthropic hasn't necessarily improved.

and this creates a structure known as the SPV, which is quite common in the investing realm, where you know, you or I could get access to invest in Anthropic, but it's not officially on the cap table, but it's through someone who actually does have access or allocation to it. And if you are able to broker such a deal, you can usually get a percentage upfront fee based on that. And because Anthropic is the biggest, hottest AI lab right now and they're raising billions of dollars, that could equate to like a couple million if you're like playing your cards right. She announced this, and almost immediately after there was an update to Anthropic support page where they go, any sale or transfer of Anthropic stock or any interest in Anthropic stock that has not been approved by the board of directors is void and will not be recognized on our books and records. And this led to an absolute crash out on X because obviously, I'm assuming quite a bunch of people have invested in these dodgy SBVs and they haven't been basically able to confirm that they have investment contracts in Anthropic. If Anthropic actually goes through with this and voids all their contracts, their money may end up in the ethosphere and they may not have actual ownership of Anthropic at all.

Yeah, it's a little concerning. And the same also happened with Open AI, where Open AI went ahead and said, hey, actually, we're doing the same thing. If you didn't acquire these legitimately, these are not yours to own, and we are going to take action to prevent you from owning them. Now, this is a big problem because a lot of money has changed hands, and a lot of these platforms are actually fairly popular platforms. I mean, Forge Global was one of them. Hive is one of them. These are commonly places where retail investors will go to in order to source these funds. And now Anthropic and Open Eye went ahead and said, these are actually invalid. You can't do this. We will negate the ability to redeem these four actual shares in the future when they go public.

So it's an interesting conversation that leaves a little bit of a bad taste because, like, man, go public then. You guys have been building so much wealth and you want it to be aligned. And yet that alignment isn't offered to the public. That sucks. but then there's also the issue that I see where these tokens are being, or the shares are being tokenized on chains like Solana and being traded at a $1.5 trillion valuation, and they're speculating on a market cap that doesn't even exist and it's not public. I mean, Anthropic is just now rumored to raise $30 billion at a $900 million valuation. So when they published this, it at least recalibrated the market where a lot of the speculation that was happening on secondaries kind of cooled off a little bit. And maybe that's mission

accomplished. We'll see. We'll continue to follow this along with all the I PO progress. Now, there is some more big news this week that doesn't relate to any of these companies, but instead, the CEO of SpaceX AI, who is on a plane currently that just touched down in Beijing. This is like The Avengers. Donald Trump has assembled the top people in the country to go visit Xi Jinping to talk business. And there's a few goals that are here on the timeline. So they're looking to rebalance trade, prioritizing reciprocity. and fairness to restore American economic independence. A lot of this is going to be kind of legal jargon, but basically the idea is that they're going to kind of put some pressure on G to make

changes that improve the quality of life of America and hopefully make some deals. Now, one of the big things that's obviously going to be discussed is Iran, getting China to push Tehran to open up that straight of Hormuz and initiate a peace deal. This would be great for markets. This would be great for energy prices because this current blockade is spiking energy prices around the globe. A big conversation is also going to be held around rare earth supply because China has a lot of this. And then generally setting up some framework for these bilateral AI discussions, particularly what I would assume around AI risk and safety, which Elon cares deeply about. And also Jensen Huang does because he's working on the open source version that is leading the American race. They're both there,

along with Larry Fink and a lot of other profound business leaders in America. And it's funny. It feels like watching this video or watching the vendors touchdown in a foreign land ready to go to business and to sign some hopefully big deals that can really impact the U.S. positively and can perhaps find some peace in this all-out war that we are facing with China on the AI progress. I can't imagine it's a coincidence that he chose Elon and Jensen specifically to accompany him on the flight and touch down and make that visual presence confirmed. Both of these guys have been pretty pro-China versus all the other kind of tech moguls like Elon sold Tesla in China. for the longest amount of time,

like he had a big market out there. And of course, Jensen has tried his hardest to sell Nvidia GPUs out there. And the reason that he's stated for selling GPU specifically is it's important to know how advanced China is,

and the best way to do that in AI specifically is to know what the goods are that they're training their AI models are. Specifically, if they train on American hardware, we know how good technically their AI models are. Now, the big issue of score that he ran into recently, is Trump said, no, we're going to ban this completely. It is an AI war against China, and we need to leave

them in the dust, and we need to onshore all manufacturing processes, GPU onshoreing, all into America. Now, of course, that's going to take a bunch of time, and China has like a massive lead for this. So they're in this conundrum where, like, China's seeing this, they want to buy the GPUs, but now they're realizing that America basically wants to play hardball, and they're like, you know what? We actually don't need you. And so for the last month, China has actually put it, the Chinese government specifically has put a mandate on all their top AI labs to use Chinese hardware and GPUs to basically train their models. And the latest version of Deepseek and Kimi K2 from Moonshot Labs are all trained on Huawei GPUs or largely served through Chinese GPUs. And guess what?

The models are pretty damn good. DeepSeek V4 and 4.1, which is coming out pretty soon, is as good as Claude Opus 4.7, but a lot, lot cheaper and quite quick as well. So the point being is, I think Trump's a little nervous. I think he wants to kind of like make amends. And of course, Jensen's there because he wants to tap into one of the biggest markets outside of the West who are going to buy his GPUs and, you know, push his company well above $6 trillion valuation. I believe they hit $5.5 trillion on this news today. In addition to all the serious conversations, there are some pretty good memes that are happening around this as well. The boys getting dinner is one of them.

I mean, the AI photos are just great. Tim Cook is there next to Elon. Trump and Xi Jinping all wearing some tank top. Jensen's wardrobe. Jensen's wardrobe, leather jackets only, that's all he wears. And then we also have this really fun polymarket in, it's so ridiculous. Everything is a polymarket nowadays, I swear, which is, will Trump and G kiss at the summit? And then the actual parameters for what makes this possible are really funny.

A qualifying kiss is defined as an in-person greeting or gesture involving the lips of one individual, touching another individual. It's so ridiculous. This is trading at 1%. So if you believe that there will be some touching, there's a 99-x. upside on this market, which it's just ridiculous. But it's a testament to how fun polymarket is and how you can really use it as a source of truth as you can navigate this crazy whirlwind. Thank you so much to Polymarket for sponsoring this part of the episode.

Now, we also have one last update from the court trials between Sam Altman and Elon Musk. We have some leaked texts from Satya Nadella. Basically, after Sam got fired, he came back to open AI, but the contingency was that they needed to decide who the new board of directors was. There are some updates, none super noteworthy this week. We're going to continue to keep our fingers on the pulse. Wait, wait, wait, wait, Josh, didn't you see the other text? Surely you didn't miss this one. From Sam Olman, let's see.

Just finished the latest limitless episode again. Still the best podcast in the world. Wow. We're being dropped in the core filing to which Satya Nadella replies. You say that every week. And Sam says, because it's true. Flattery will get you a guest spot, not more compute. That's so funny.

You know how I know AGI is achieved because they were able to quote Limitless back at a time where Limelis didn't, you know, even exist. So this is pretty impressive. AGI has arrived. But with that little meme of the day, thank you all so much for watching. You'll notice that this episode is a day early. And that is intentional because we are monitoring the situation. There is a Mr. who goes by the name of Leopold, Ashen Brenner, who is publishing a 13F filing.

Now, Leopold, the guy who turned 1.5 billion and 6 billion, probably one of the most profound AI investors today, is releasing the filing of the new holdings that he has. and we are waiting at the edges of our seats because it's supposed to be coming by Friday. That way we can be ready to record and publish an episode about all of the holdings and all of the new meta that people are going to be investing in over the next quarter.

So keep your eyes peeled for that. It's coming. It's going to be exciting and it is sure to move markets when it does release. Yeah, I'm super excited about this one. It was one of our most watched episode and Josh and I, if you've been watching the show

for any length of period of time, are obsessed with which stocks to own, mainly because we're such bad traders ourselves and so we need to know which ones to buy. So if you don't want to get FOMO, definitely tune into that episode. We will be the first ones to release that. But aside from that, that is all we have for you this week. We'll be bringing you more news next week.

But until then, we will see you on the next episode. Thanks so much for watching. See you guys.