Limitless Podcast - THIS WEEK IN AI: Fable 5 Returns, OpenAI Might Nationalize, Meta Sells Compute

Episode Date: July 3, 2026

We are so back. Anthropic announced the return of Fable 5, its safety updates, and related releases like Sonnet 5 and Claude Science. In other news this week, we look into Meta’s brain-to-text research, OpenAI reports, new inference hardware, and recent robotics demos.------🌌 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 Fable 5 Returns3:12 Banned, Safeguarded, Back7:58 Sonnet 5 And Science11:33 Meta Reads Your Mind14:11 Meta's Compute Pivot17:17 OpenAI's Power Play23:26 Etched's Chip Breakthrough27:37 Memory Market Shock31:02 Robots In The Home36:54 Industry, Not Housebots39:01 Episode 200 Milestone------RESOURCESJosh: https://x.com/JoshKaleEjaaz: https://x.com/cryptopunk7213------Not financial or tax advice. See our investment disclosures here:https://www.bankless.com/disclosures⁠Josh works with Anthropic as a contractor. All views expressed are his own and do not represent Anthropic, its leadership, or its affiliates. Nothing in this episode is investment advice.

The King is officially back. Anthropics flagship model Fable 5 is finally available for everyone after being banned for almost two weeks by the US government. It is still the world's best model. I've been using it in the 12 to 24 hours since it's been launched. I'm super excited about it. There's a bunch of demos that we want to get into almost immediately. Starting off with my favorite kinship as a child, Hogwarts and Harry Potter. What you're seeing in front of you is a completely simulated fabrication of the entire Harry Potter unit. starting off with Hogwarts and you'll notice that I can click into this universe immediately and start looking around. It's important to mention that this was built from a single

prompt. I'm not entirely sure how long it took, but Fable 5 did this in one complete shot. And if I press B, you'll notice something pretty spectacular. I'm on a broom and I'm flying around in probably my favorite childhood universe. Josh, I don't know if you're a Harry Potter fan, but this is pretty freaking awesome. It's unbelievable. And I'm not sure. if it was one shot. This is pretty impressive for one shot, but it was absolutely generated with Fable 5, and it's a testament to how impressive the model is and how, like, happy I am that it's back. Thank God, holy shit, it's been a long couple of weeks without it. I mean, what's most amazing about this, I find, is that not only is it able to generate the map of Hogwarts, but it's able

to generate the unique pieces. Like, you just went into the dining room. This is the Weeping Willow tree. It has sound design built into it. It understands the physics and the scaling of the way things should be built and look and like now you're going over to play quidditch. It's this like unbelievable awareness of the world and this unbelievable ability to generate these visual elements. Like this is this is so amazing the fact that you're playing this in your browser just by someone who prompted the fable a couple of times and then deployed this. And you could see it's like pretty accurate. You have hoggworth. You have the dining hall. Everyone has fable until July 7th, I believe the day is in which the limits get. Unpaid subscriptions. And then the limits change a little bit.

And this is subject to change as always. But currently right now you have. have about a four to five day window to really play around with this as freely as it's likely ever going to be. The intention of showing this demo is to showcase that the possibilities of what you can make with this model really are pretty much endless. And I think the challenge to everyone now is now that there is this model that's out that's truly unbelievable in capability relative to anything we've seen. It's to go out and actually try unique and fun things. Like I would have never thought to go off and make a Hogwarts game that I can go and play. I mean, traditionally you buy this game. I have a Hogwarts game that I bought for $75, $80 on Xbox.

to play around with. So what types of difficult challenges do you have that you'd like to try and that you never thought were possible? And like those are the things that I would encourage trying out with Pable 5. Really amazing model. Very cool. You could see here the second demo, this is all about physics and the physics understanding and it shows it on a relative basis compared to the other models. And it's just so far superior in its understanding of the worlds of physics. So a lot of these are visual examples. It does really well on basically all other types of work. So that's the thing. It's just try it out. See what you can do. I love this intro post that they shared Fable is back and it's now showing and the whole sign lights up and it's like, man, we are so back.

So a really exciting day. If you were listening to this, go track Fable, go play around with it. It is an incredible model. I want to talk about what has changed because obviously we can't ignore the elephant in the room, which is this thing was banned for about two weeks. The U.S. government specifically banned it because it presented itself as a potential cyber security threat across all national defense systems, which of course is very important. The TLDR of their blog post is essentially they have added further safeguards and more monitoring as to how people use the specific tool. They'll be looking at every single prompt in terms of whether it could potentially be an attack vector on the US or anyone in general. And they've been working closely with partners from Glasswing their collective, which they set up once they created Mithos, to make sure that their defense systems are hardened appropriately against this model.

That being said, there was a lot of fearmongering around the re-release. of this model. There was talks of potential KYC verification or potentially only allowing people in the US or American-born people to get access to this. I'm happy to say that all of that has been dismissed. The government and Anthropic, as well as Open Air and a bunch of other frontier AI labs, have been working pretty closely together to form some form of a framework for any future model release, such that not everyone can be banned against it, but it's released in a more secure way that people can still benefit from the access of it and not be used in a malicious way. So I'm happy about this. This is Fable 5 back in all of its glory. I'm using it across

all of my chats right now. Opus 4.8, you have been a pleasure to work with, but sorry, I need to work with your older brother at this point. But I'm excited about it. You have until July 7th, as Josh mentioned. So use it as much as you can. It switches to usage-based pricing depending on what task you have. But it's great to have it back. Josh, what have you been using it for personally? I'm curious. I've been using it for all of my hardest challenges and all of the things that I do mostly day to day. A lot of just like research, a lot of production, pretty much any challenge that I have, I have switched over to Fable and I noticed that it does it much better. So in the case that we're researching a topic for a show, for example, it will go much deeper. It's more comprehensive. You can kind of read through its reasoning and understand how it comes to conclusions. And it's a really good thought partner, at least for me in a lot of cases. A big problem that I have, again, is figuring out where the, edges of this thing are and what I actually can use it for to be a little bit more creative. It's like, okay, I shouldn't really be using this as an extension of Opus. What new use cases can I try? And I think that's what I'm going to be spending a lot of time this weekend trying to figure out

is what are the Harry Potter-like demos that I could use that I'm actually really interested in building myself. Yeah, so one thing I've noticed is I've never had a personal website before. And I've thought, like, hey, why don't I just give it a go and try and figure it out? Now, the one bit of critique that I've always had with any AI model up until Fable 5 is the taste is just terrible. I'm sure you've tried this before, but you're like, hey, like, build me a website that looks something like the Airbnb website. You can share screenshot demos, and it just kind of does a terrible job or a janky job. Fabal 5 is the first model that I've worked with that has this visual intelligence, which is super weird because I can't currently use Claude right now to generate me

an image. I can use ChatGBT to do that. I can use Google Gemini to do that, Nano Banana, but I can't do that with Claude specifically right now as an LLM, but yet it understands what I see so well. So what I've realized is if I'm using it to code up some form of a visual artifact and we use visual artifacts a lot on our show as props or materials, it does an amazing job and it understands taste. It doesn't look like generic AI slop anymore. So I definitely recommend folks go and check that out. Now, one thing that we have to mention on the show that we noticed pop up was there was a discussion once Fabel 5 was re-released because people started coding with this thing and they updated their recent version and they noticed that there was some form of like they described it as a spyware like code in Claude Code

that would basically track certain users kind of activity on what they were using it and obviously there's a big reason for this because a lot of foreign state-based labs such as in China specifically were distilling Claude models to build their own models which is strictly against Anthropic policy. And what we got was a very quick dismissal from Tharic himself at Anthropics saying that this was an experiment that we were launched to basically prevent distillation attacks from foreign adversaries. Now, if you're Anthropic and if you've been following the news,

especially on Limitless, you'll know that a bunch of the Chinese open source models have effectively attained their status because they've distilled frontier models from American labs such as OpenAI and Anthropic. There was recent reports, I think, this week, of Alibaba's Quinn doing the exact same thing. There was an official announcement from Anthropic. So it's good to see us taking these steps and measures to prevent such attacks. This is part of the

whole safeguard approach, which is new with Table 5. Yes, big week for Anthropic. In addition to that, they also released Sonnet 5 and Claude Science, two major releases. Sonnet 5, as everyone knows, is the kind of smaller model. It's the follow-up to Sonnet 4.6. In fact, it's priced a little bit lower than 4.6, at least for the intermediary up until August 31st. And it really is just like the entry level model and the model you'll find that Fable probably best wraps to. So when you're using Fable, sometimes it gets expensive. You're using it for very long-running tasks. Sometimes not everything needs frontier intelligence.

Sonnet 5 is a pretty good substitute for it to defer and start to offload tokens and request to. And it seems like it's just like, you know, a solid upgrade. There's nothing exceptionally incredible about it. But it is there. It is cheaply price and it is very strong in the parts that it needs to be. I actually really like the look of Sonop 5 for one particular reason. Over the last week, people have been praising GLM 5.2, which is a Chinese model from the lab, Jipu.

And we did an episode on this, and it's a very impressive model. But my take on this is, I think American Frontier Labs aren't just going to sit around and let Chinese AI labs outcompete them on costs specifically. So they have the best cards to play, which is if you have the frontier air model such as Claude Faber 5, or ChatchipT 5.6, Sol, you can just distill your flagship model into a much cheaper model. And that's what we're seeing with Sonop 5. It does Opus 4.8's capability, but at a fraction of the cost. And so I think we're going to see this trend continue until we find that perfect medium in between

how cheap GLM 5 is and how expensive Fable 5 is, and people will just stay within the Clought ecosystem. And I think it's a smart approach, and I'm happy to see it. But you mentioned Clude Science as well, Josh. Super cool, yeah. As a former bioneert that did a FOIA biology degree and like spent a lot of time studying genetics, I wished I had this tool. This is basically the co-pilot for any kind of scientific research that you do. It's a model that basically is fine-tuned and built to evaluate scientific primary research, help you do all your data analysis. And it's been used across a ton of different things. I mean, if you look at this guy called Andre, who has been a...

playing around with Claude Science, he installed it with ligand AI, which is basically a MCP tool, which basically allows him to analyze different molecules. One of the big things when you're kind of figuring out whether an antibody or a molecule is good enough to fight a disease is you need to know whether it can attack a specific active site, which is like basically the vulnerable site of a bacterial virus or whatever that might be. And this system with Claude Science basically helps you identify it de novo, which is like super cool. Yeah, I think this is probably the most interesting and exciting release of the week. Granted, Fable 5 already released. We're just doing it again. But Cloud science, you could think of it, kind of like the way that Claude Code is intended to

support engineering. Cloud science is intended to support science. And that's all of what this is about. It integrates a very specific set of tools. It allows you to audit the outputs. And Anthropic is actually using it. They announced that they're using it to pursue their own drug discovery programs for some rare neglected diseases. And people have been using it. The Allen Institute now completes a hundred plus page literature reviews that would previously take up to two years. And then some analysis now run in a tenth of the time in what they previously did thanks to this new software harness that they've introduced for science-based purposes. So when you combine a model like Fable into a harness like Cloud Science, you get some pretty amazing technological innovation, or at least

the opportunity to do so if you are feeling so inclined to through people who just like are involved in research in general. So I think this is super exciting for the science world. I think we're going to see a lot of really interesting things coming from this. And that moves us on to our next topic, which is science related from meta, believe it or not, the metaverse company that was supposed to make metaverse. They were all supposed to make AI. Well, it turns out the thing they're releasing this week is biological based and it is a brain-to-text decoder. What is a brain-to-text decoder? Well, basically topologically they insert some or they put some electrodes on your head and correct me from wrong you guys but they have actually managed to hit 61% word accuracy versus 8% for

previous non-invasive methods so they're basically able to go into your brain detect which neurons are firing and then derive a series of words that you are meaning to say without actually saying those words it's essentially reading your mind is that right the best part about this is they only did this with nine volunteers nine people no not even double digits and they trained this model on basically 22,000 sentences. So comparably, if you look at like an LLM, which is honestly, the frontier LLM's like Fable Fiver are upwards of like 15 trillion parameters and they use a ton of compute, this is a small model in comparison, but it's equally so, so powerful. They basically are able to interpret different brain signals and convert that into text or words. The reason why this is

so important is, well, firstly, there's the medical side of things. We've spoken about NeurLink quite a bit. This is kind of Meta's answer to that, at least in the initial types of research that they're doing. Now, application-wise, I am curious what Meta is going to use this model for. Obviously, there's the medical side of things. But we know, and we've spoken about this previously, they are working on a series of different brain models, as well as ECG-related devices that you can wear on your risk and on your heart, that can basically interpret everything that you're thinking

and figure out what the best type of content is to produce for you. So it's an interesting thing where Facebook is a social media site. And one thing that they've spent a lot of time figuring out is how to get all the wonderful imaginative thoughts in your head out into the open world. They've done that with Instagram through you picking up a phone and taking pictures of stuff. They've done that with Facebook, translating your thoughts into cool ideas that you can share with the world. This seems to be a step in a direction where for those folks who aren't really good at, you know,

using those skills to translate your thoughts into some kind of a vivid pitch or whatever that might be, you can now use this potential tool in the future for something like that. And listen, I have a more optimistic take for all these kinds of tools, which is, you know, they're going to use it for the advancement of humanity and good. But I don't know what met a specific purpose for that might be. But it's interesting to see nonetheless. And I like that this kind of research is being focused on, despite LLN is being focused on being chatbots or generating images.

This is like advancing real science. And, you know, between open AI's gene benchmark, between clothe science and between this, this is great to see. It's a great week for science in AI, for sure. Yeah, it's been really exciting. It's cool to see progress. And then I'm also just like very confused at what meta is doing because we're talking about science now. But then there was another big pivot that just happened where meta is shifting to cloud infra. They're like pivoting to the SpaceX plan. So it seems like meta over the last decade has just been the series of unfortunate

pivots. And every time we pivot somewhere new and we try this new hard thing, it doesn't quite work out and they pivot again. And it feels like a startup kind of lost trying to find some traction. This is the newest iteration of that. In addition to the science breakthrough that they're sharing in terms of like the topical neuron receiver stuff, they're also just going to start selling compute. So I don't know if this is bullish or bearish for their internal AI program. I have to imagine it's not that exciting because if it was going so well, they wouldn't have extra compute to sell. But the idea here is at least that they will begin to sell their compute to third-party companies, similar to what SpaceX has been doing with their deals with Anthropic and who else did they sign a deal with Google?

And I think it's just another instance of them trying things, throwing stuff against the wall and seeing what sticks. I don't know. You just, do you have any takes when you read this news? Yeah, I have an alternative take, which is, I think it's a smart move for Zuck. It is... Market loved it. Yeah, it's a potential massive revenue earner. The stock went up, I believe, like 6 to 8% on the day. And the reason is pretty simple.

Meta has aggregated one of the biggest fleets of the most important metals needed for AI right now. That is the GPUs. We said the same thing about SpaceX. So SpaceX has Colossus 1, 2, and Data... and three data centers, and that is roughly over, I think it's a million of Nvidia's top GPUs. And so that fleet can be very attractive for a competing lab, such as Anthropic and Kursa, who they are selling their GPUs too. Meta saw this and thought, hmm, that did pretty well for the stock price. Let me do the same over here and bring some money

in to justify additional AI CAPEX spend for future years. So I think this is a strategic move from Zuck to do this. And I have another bit of evidence, which proves that this isn't necessarily meta running for the hills, which is Meta signed a compute deal with Google. And Google this week announced that they have to restrict Meta's use of their Google Cloud because they're using too much of it. So my suspicion, and we'll find out in the Q2 earnings report from Meta in a few weeks' time, is Meta's allowing all their old GPUs to be sold for inference. And they're going to make a ton of money in the same way that SpaceX has charged a huge premium to Anthropic and Cursor for selling inference through their GPUs.

Meta's going to do the same, make a bunch of money from that, and save all the Vero Rubens and the newer GPUs to train future meta models. And we reported, I think, two weeks ago that Mehta's working on a potential mythos-level model that's going to be available nine months from now. I know that's kind of late, but I still think that in this race, they're just very, very, very far behind. Well, hey, all the power to them, I hope they figure it out. Now, we should probably pivot over to someone who has figured out mostly, which is open AI. And they, we've recorded an entire episode on this.

They released 5.6 this week. If you haven't listened, I would highly advise going to check out that episode you recorded. The update now is that it's still not live. And by the time of reading this, that might have changed. By the time you're listening to this, hopefully that's changed. I think everyone's really excited to get their hands on it. But they have not met the same fate that Fable has just yet. So they have Sol, Tara, and Luna, all associated with GPT 5.6.

They are probably being used internally. You have to imagine they are being used in an internal research project that is with select companies, but they are not broadly available. So we're still at a standstill in that point. But the real news that we got this week was this 5% number that's been floating around. And none of this is confirmed by any means. But there's a lot of reporting going on in that OpenAI is proposing to hand the United States government 5% of the company just for free, right? They're just going to give it away. Yeah. Yeah. I was going to bed last night. I thought we had our agenda all prepped and ready, Josh, for today's roundup. And then I saw this breaking news from

Financial Times, which basically says that Sam Altman is reportedly shopping 5% of OpenAI to the US government to take ownership. And the craziest part about this is, I don't think he's even requesting an investment. He's just kind of like handing this over for the good of the world. And if you want to know some context as to why he might be doing this, I think there's two perspectives of this. Perspective number one is Sam believes deeply that governance and national government should be involved in how these AI models are shaped. And we know this because in an earlier blog post from Sam, I think about a year and a half ago, he said this. He said it might actually make a lot of sense for the government to take stakes in major AI labs. The reason for this is, as we've seen with

Claude Meathos and Fable 5 being banned, it is incredibly important that the government gets access to these models pre-release so that they make sure that it's not a cataclysmic threat to government security systems and whatnot. So it seems to be a good person. perspective from that side of things. But on the other side of things, this could be Sam using a chess piece to basically get a political advantage when it comes to writing policy frameworks. As we know, Sam's been heavily involved in lobbying certain governments and laws in terms of making frameworks more favourable for open AI's types of models. We see other AI labs doing the similar types of things. If you were a critical person of Sam, you might think that this is the move that he's making.

It's kind of weird. I don't know how I feel about the whole nationalization of of the AI movement, but it certainly seems to be the trend that we're moving towards currently. Well, I like the fact that it was at least voluntary. This was presumably not forced. They're offering this up. You have to assume, I mean, again, these are all reports. Nobody knows what's actually going on. But the idea is that they would just make a generous donation to the United States government. And this makes sense when you look at it in terms of alignment. It's like, what is the most pressing issue that a lot of people are very upset about right now? It's AI. And the downstream of facts that they think it's going to cause on their life. They're,

scared they don't understand. They have no vested interest in this. And I think giving everyone a, even if it's a small amount of vested interest in the success of these companies, it's probably net good if it could even change the sentiment a little bit, if it can get people aligned, if they could feel good about these companies winning because it means not only are they winning, but the country is winning and everyone is kind of aligned directionally on the success of these companies, I think it creates a really compelling case for handing some over to the government and giving it back to the people that are allowing you to do this. They might just be feeling exceptionally confident this week because we have another report, another rumor. Again,

unconfirmed. All this is unconfirmed. We are waving our hands around. We are putting our tinfoil hats on. But allegedly, there has been some sort of optimization achieved internally that allows them to cut inference costs in half for models that was applied to. That's crazy. Because half of the costs on hundreds of billions to trillions of dollars is not marginal. That is a pretty serious innovation. So what is going on here? EJOTS, is this even possible? Yeah. So if this report is true, the analogy looks like the following.

Right now, if you're a free user, chat GPT, there's roughly around, I'm going to say 100 to 200,000 GPUs that are out there that are processing your prompts, your inference, your request, giving you answers. Those GPUs, you know, I hate to mention it, are super, super expensive. Now, if this report is true, it's cut down the. that inference cost, that cost, to 50%, which means that this is now a couple hundred of GPUs. And this is not a number that I'm making up. This is a number that's in the information's report specifically, a couple hundred GPUs, which means that if you can cut down cost that massively,

you now have free capital to spend on compute in other ways. In open AI's case, it's probably to train more models or to use inference for other types of usage, such as inference at test time scaling or whatever that might be. So this is a major jump for a few different reasons. Infference roughly makes up, I think, 60 to 70% of revenue earnings across top frontier labs right now. So if you're a frontier lab, a lot of your cost typically goes into training. A lot of the money that you make comes in from subscriptions. Now it comes in from inference specifically for people using it to service agents that they're running on Autoloup for hours and hours on end. Agents become basically the new trend. Enterprises are using it 24-7 at this

point. So it's become its own type of economy and inference is basically the bedrock that underpins all of that. Anthropic this quarter is reportedly meant to be the first AI lab that goes profitable in Q2. A large part of this is achieved through the cost or the revenue margins that they earn off of inference specifically. It's roughly around something crazy like 80%. I remember Christian Rao mentioned it on a podcast previously. So the point is inference is very important and any way that you can cut costs down on that and extend your profit margin. is a huge deal. And if Open Air has found some kind of a breakthrough, and I don't know how they've done this, I'm looking forward to actually an official announcement, this would be a huge leap for them in terms of

getting more capital to acquire more compute to build better models. It's pretty awesome to see. We do have a confirmed breakthrough this week. There is actually something that has been publicly announced, and it comes in the form of this company that just exited stealth named Etch. And Etched, I saw the post and I was like, oh, all right, cool. It's like another one of those things. It's like, they think they're going to change the world and they're going to solve the GPU problem and tokens and price per watt and they're going to solve all the efficiency gains. And I was like, absolutely not. These guys are like in their early to mid-20s. They have no experience doing this. Surely this is just another one of those things that is a headline and then it

devolves into nothing. But I started to read about what they were doing and I started to listen to a few of the podcast episodes that these founders were on. And I was shocked at how competent and incredible the product is and how talented the density of this team is. And everything about this story turned into this like unbelievable breakthrough. I was like, wait, holy shit, this is, this is like pretty amazing and pretty magical. And I think so much so that we're probably going to record an entire episode about this company and this general state of inference as it really sees breakthroughs probably next week. But just a TLDR on Etch is they've basically figured out how to create a new chip set from the ground up. First principles, vertically integrated,

meaning they're building all these things from scratch. They're integrating them all into a single unit, their product is the way in which they manufacture this server rack, and the server rack is the delivered good. But within that, it requires cooling and chip architecture and all of the plumbing as it connects to everything. And they've created this series of breakthroughs using low voltage inference, which is one. So they're able to squeeze out a lot more performance per watt. Generally, on a GPU like a Blackwell chip, you are getting maybe 50% of its theoretical peak, because if you go to 100%, the chip melts. So being able to squeeze more tokens per watt is a really impressive thing. They figured that out. They figured out a memory problem where normally you're

constrained to the memory on a chip. Well, they've constrained it to the memory of the cluster. So the cluster becomes one single coherent piece of memory. And there's a series of these breakthroughs that they've kind of discovered that were really impressive. And then you look at the team and you're like, oh, wow, everyone on this team is like shockingly competent and have been working in the industry for many, many decades. In fact, I think the youngest people are the founder. So etch seems like a grand slam. Stay tuned. We're going to have a lot more to say about this company. It was shocking. It was really call to see this week. Well, just to kind of like sing their praises a bit more. These guys have raised, I think, $800 million, but they already have a billion dollars worth of orders for a chip that

currently doesn't exist yet and is currently going to go live by the end of the year. The backers for this team are pretty insane. You've got Peter Thiel, you've got Jane Street, but then you've got TSMC themselves. As far as I'm concerned, TSMC is the company that is required for NVIDIA to stay alive. At the moment, they do all the manufacturing processes around all their GPU. and TSM taking an equity stake in a competitor to NVIDIA in the first instance, as far as I'm aware, is a huge deal. Now, you mentioned what the breakthrough is specifically. I think when we talk about other companies

that have made inference-specific chips, because I'm sure you're thinking about this, we've spoken about Cerebrus that went public in a huge IPO about a month ago. They created an inference chick, and it was the size of like a couple of plates or something crazy like that. Their breakthrough was effectively designing a chip

that would allow your AI models to work incredibly faster, but it would use a lot of power. It was quite expensive to use. The difference with this chip that Etch supposedly is going to create and release at the end of this year is it uses low voltage, which means that it uses 75% less power for the same amount of inference output. That's the breakthrough that they've made. And if this company can pull it off, and like you said, Josh,

it's like a bunch of Harvard dropouts, this would be an incredibly competitive threat to Nvidia, because as far as I'm concerned, Nvidia's focused on general purpose GPUs, which are used for training specifically, but the inference market is pretty open to anyone and everyone. That's what you see, Cerebra, Scrock, and now etched come to light. So it's pretty cool to see it.

It's so exciting. It's rare that a company comes around that's actually new and novel and making real breakthrough. So I'm stoked about this company. I could not be more excited. I would love to participate. I wish they had the ability to buy some shares, because my God, what a monster of a company that's going to be.

For the people who do own shares in public markets, chances are they might not have done too hot this week. The memory stocks have been getting hit. People haven't really liked this. And you just, we have this post on screen that's had 2 million views. And there's no way that this post moved memory markets, right? I think it did. That's insane.

I was racking my head to try and figure out why memory stocks were down 10% yesterday. It was one of the most aggressive dumps we've seen of recent, especially after a meteoric rise in memory stocks over the loss. like yeah, pretty much is up, like, on average, between 3 to 500 percent, depending on which memory provider you pick. Andrew Curran is one of my favorite accounts on X. If you don't follow him, definitely give him a follow. He posts this very cryptic tweet.

He goes, I'm posting this prediction now so I can quote it later. There has been a significant breakthrough in architecture, specifically around memory efficiency, not by one of the big labs, but a team that spun out of Open AI. This will probably be announced soon. And this led on everyone on a spiral in social media. Now, I have a bit of tea because I have some sources which reported a few things to me. Now, apparently the company that he was referencing is this company called Core Automation.

What this breakthrough is, he has no idea, we have no idea. So right now, it's just word of mouth. And if Core Automation wants to make an announcement, they can. But if this announcement is true, it would mean that AI models largely require much less memory. Now, why is that a big deal? The most expensive material that is required to build. GPUs and trained models and inference models right now is memory. It roughly makes up around 50 to 60% of the cost.

And the price per unit of memory has skyrocketed over the last year specifically. Margins on Micron's recent earnings report, for example, has hit 80%. Traditionally, this is around 30%. So the fact that it is over double this is absolutely insane and the margins keep going up. The prices keep going up. In fact, Apple announced recently that they are lobbying the Trump administration to buy memory from Chinese companies, which is a big no-no in terms of import-export controls that Trump and previous U.S. governments have placed. We don't know if this is going to get passed, but the point

is companies like Apple need to now increase the price of their products because they can't get their hands-on enough memory, so they need to end up paying more, which means they need to charge you as a consumer more. So memory is in very scarce supply. My take on all of this is it's being blown out of proportion. Memory is an incredibly constrained commodity, and there's no way to scale it. You need a bunch of very complex machinery and chip fabs to be able to do this. And we are nowhere near. I don't see the supply scarcity being alleviated until at least 20, 28 by the end of that year. Well, yeah, we know this. The markets are very short-sighted always. When there's headline news, they're going to overreact. There's going to be this knee-jerk reaction. That's what we've seen.

In fact, there isn't even really headline news. This is just kind of all speculation. It's funny. I'm looking at core automation and their URL is core auto.com. It looks like a freaking car manufacturing company. But it's like, okay, sure, like they have these breakthroughs. But how even in the case that they do have a breakthrough, open AI or anyone, what is the lag time between breakthrough to actual implementation where it is felt in the market, probably fairly long? There's probably, I mean, certainly a tremendous amount of demand for memory in everything

that's being built, not only on the software side, but also on the hardware side. If you're building robots, if you're building automated hardware manufacturing, all of this requires a tremendous amount of memory. So is there a memory? bubble popping, like probably not, but again, not financial advice. Who the hell knows? Talking about robots, though, we do have a robot topic because we're getting some consumer robots. Like, robots are becoming a thing you can buy in this. Look at this kiddie pie, Josh. You want this to your apartment? You want this to be a little bit adorable. Like, do I want that in my apartment? Honestly, no. My apartment isn't big enough

for the two of us. Like, I feel like we'd probably bump into each other and it probably wouldn't be this thing is massive. It's massive and you know, it's still moving a little slow. It's like not really doing everything that I need for my liking. Maybe if you live in a big house, it's fun to have this guy running around. But I do think it's cool and exciting that more companies are trying this. In fact, this one's actually, dare I say, kind of cute. It looks like this fun little robot that you can have walking around the house. You can get it in different colors and you can pre-order it for $8,000, which, hey, for a humanoid robot that can roll around your house. Or $500 a month. Or $500 a month. Or $500.00. That's true. An expensive New York gym membership, you can have

this robot walking around and cleaning your apartment with these freaky-looking pincers. Did you see the demos? Did it look compelling? Was this an interesting product? Honestly, no. It was too slow for me. And I hate the fact that I'm pointing this with my cursor right now, that it's effectively one of those old...

In high school, Josh, did you have those like projectors where, like, sometimes your teachers would like... Rolls around on the cart. This is exactly like that. My number one question is like, when you come across like a carpet or something, like, how are you getting over there?

Like, I don't want to have to like be the topple monitor. where I have to pick you up if you fall around. Like, how do you know where to place my clothes? I don't know. There's just a number of different questions that doesn't justify $8,000 in my mind or even 500 bucks. But the good news is this isn't the only company that's releasing robots like this. We had Norie L2 demoed there.

Less attractive robot, I have to say, more bare bones, but this is like a demo. It seems to move a little bit more agile and quicker than the previous weave product or robot. But the fact is, a lot of American robotic startups are releasing their products now. And if you compare that to six months ago, it was all theory. It was all hearsay. It was all demos from the likes of figure. You had their robots in the BMW manufacturing point, if you remember this, Josh. I mean, was supposedly like operating all the controls and building cars.

Now we see these robots, one, coming at a more affordable rate and two, being applied to your home. Now, the big question is, do you want to allow any of these robots in your home right now? And can this be scaled, mass produce? Can this be in my hands right now? The answer is still unfortunately no, but the target date is a lot closer. You can get these robots in a couple of months or by the end of the year at least.

So, you know, fingers crossed. I hope we're entering our robotics era, our chat GPT era for robots. But right now I'm not utterly convinced. The pincers still throw me off. I need some hands. I need something over the way. It's one of those things we're like,

dude, I believe it when I see it, man. Ship the product. Let me see working demos in real people's houses. Like, give me something to really get excited about. Let me invest in the company. you know, a few other things like that. But only the good ones. It's like a lot of the, most of these companies are going to fail.

And they're going to just create fun products. And they're going to create these little like flashes in a pan. And then they're going to dissolve into nothing. Because turns out it's really difficult to manufacture these things at scale. And now they're competing with the supply chain that the entire planet is also competing on. They need the same memory that these other AI companies need. And there's only so much of it to go around. And the economies of scale are just not really practical for these companies.

So in terms of research and development, I think it's amazing. It's fun to kind of hone in on a form. of what these robots could look like. It's fun to figure our demos and use cases. But in terms of actual practicality, I don't think I'm going to have a humanoid robot in my house this year. I don't think I'm going to have it in my house next year. Maybe 2028, perhaps Optimus will be around by then, perhaps figure we'll have some cool robots by then. But it's going to take a little while to ramp up these production lines and get something that's actually viable via these economies of scale. Here's a question for you, Josh. What would be the one pain in the ass that you would

definitely allow a robot to solve. That would be, that would tell you like, you know what, I'll pay the money. I'll pay 300 bucks a month to get this robot in my house to solve this one problem. Do you have one? You know what's funny is like, I don't think there is a problem so large that it requires a robot. It's like one of the things that's kind of annoying is groceries. I have DoorDash and Uber Eats. They will go and shop for me and they will deliver it or even with Whole Foods. They have grocery delivery. You can get someone to come and clean your place once a a week and that's like a very good service and they're going to do a significantly better job than a robot will. It's like, okay, I do laundry, but it takes, what, like 10 minutes to fold the clothes,

and it's not that big of a deal, make my bed in the morning when I wake up, I like to cook my own food, there's a dishwasher to handle the cleaning of it. It's like there isn't really anything very pressing in terms of, I mean, initially my head goes to chores that I think a human robot would require. So I think that's kind of why I'm waiting just to see the unique use cases they could figure out or if they just get it down. Yeah, I think it's really down to use cases. If there's a killer use case that I see, I'm like, I need that, then sure, There's no amount that I wouldn't spend on getting this because it's such a cool piece of technology. But there has not been anything even remotely close to that use case.

Turns out we've actually automated a lot of things pretty effectively in our life. You see, I've come to the same conclusion, and that's not necessarily bearish robotics. I just think it's bearish robotics in your home. I think robots apply to more industrial use cases, such as manufacturing, or maybe even food delivery, makes a lot more sense. I think fixing that last mile problem, last mile delivery problem, where, like, you know, you can deliver everything to another country by getting it to that specific address. This is the most annoying part is a valid use case for robots.

I think the manufacturing side of things makes a lot of sense in terms of like car automation or building whatever in the future, maybe even building robots in themselves in the future, which is what Tesla's focused on. This is what Prometheus, which is Jeff Bays, this is new robotics companies, focused on. This is what Atoms, Travis Kalanix, ex-UrC-OV-C-CEO, is also focused on as well, right? They're focused on the manufacturing and industrial applications of robots. And I think if you were an investing type of person that wants to focus and figure out where the robotics trends is heading, I think it's in these particular sectors.

And that's why I'm going to be keeping an eye on. And maybe we even do, I don't know, a robotic specific episode going forward. If that's something that you guys are listening to and would like to hear more of, I think Josh and I can work that out. That would be a pretty fun to do. Yeah, let us know. Would love some feedback. Oh, I wish there was a way to participate in these companies, man. It's driving me crazy of all the cool companies we talk about and every single one of them is private.

It's like, oh, etched is awesome. Can't invest. Adams is the most incredible company. Nothing. So I don't know. Hopefully, through all of this, we'll figure out a way to get access to the upside of these amazing companies. But I think that's everything.

This is a long episode. If you're still rocking with us here, congratulations. And thank you. You are a real one. We appreciate everyone making it to the end. The outreach for sponsors has actually been so, so nice and so sweet. So thank you for everyone who's reached out in terms of getting us, help with sponsors, becoming a sponsor yourself.

We are looking for people to help support the show so we can keep the way. We still looking to do this for longer. Yeah, so please, if there's anyone you know, please feel free to reach out. We will be happy to converse either through X or in the email in the description below. If you enjoy this episode, don't forget to share it with your friends. And if you've made it this far, you're all caught up. Go enjoy the weekend. It's Fourth of July, baby.

Have a great weekend. Happy birthday, America. Happy semi-quintennial. Yeah. Happy quintennial year. for everyone. It's like, 150 years old.

Let's go. That's huge. That's huge for the gang. Like, what a country. Look how far we've come in, in a quarter of a century. Or no, two,

two and a half centuries? Two and a half centuries. What do you call a thousand years? Quarter of a, I actually don't know. An eon? That's a question. I don't know.

For fable. That's a favorite. So when we end this, use your cutting edge AI model to go ask, oh, what is a thousand years? How do we say that?

Oh, my God. Happy birthday to us. We are so cooked. At least our brains that we all float to are getting smaller. So even though we're not thinking for ourselves, the brains are smarter this week. And I'll be smarter until July 7th, until we get to usage-based credits, at least at this point. And then I'll probably still end up using the usage-based credits. But yeah, that's it.

Have an amazing Fourth of July weekend. Hope to celebrate. An exciting milestone for us, EJazz. I'm not sure if you are aware, but we are very close to episode number 200 in the most, which is unbelievable. This right now, if you're listening to this. And also, this is news to EJS too. This is episode 199 of Limelis. So we are one episode away from 200.

We're going to record that probably next Monday. And that'll be a celebration. So for the people who have been here since the beginning, like seriously, thank you so much. It's been an amazing journey. Thank you. 200 in a little over a year is crazy. We're recording four episodes a week.

And it's been, yeah. We're almost like, what, 65,000 subscribers on YouTube and many, many more listeners all over the world across Spotify, Apple Music, wherever you are, waiting, commenting, letting us know. your thoughts. To go from scratch to this has been insane, Josh. I don't know if you have any thoughts or highlights about this. Maybe we should talk about this on the 200th episode.

We might have to reflect on episode 200. So stay tuned for that one. And I have a feeling it might be about etched because I have a lot of cool things I want to talk about when it comes to that company and inference. So as always, stay tuned. Don't forget to share us with a friend if you enjoyed, if you find someone else that might enjoy. We have a newsletter dropping at the same time you are

listening to this as well. And that's everything. So happy birthday, America. Thank you all for watching. And we'll see you. next week. See you guys.