Google commits to a $40 billion investment in Anthropic. Dario needs computer, signs up with Amazon. Google's already a shareholder in Anthropic. They're trying to maximize the economic value per token. It's all bottlenecked at TSMC. That's the actual bottleneck to all of AI. Only Elon will talk about it. Google Cloud is dominating. They unveiled their eighth generation of TPUs. In particular, TPU 8T for training and TPU 8i for inference. I still believe Google's the winner in the long run here. OpenAI unveiled TPU T5.5. It very much feels like a release that's intended to strengthen OpenAI's codex. Math is cooked. A bunch of other things are cooked as well. Things are moving so quickly now that on a month-by-month basis, we're able to see the hardest of these benchmarks creep up 1% per month, so not long now. And the moonshot, ladies and gentlemen. Everybody, welcome to another episode of moonshots. Your favorite AI exponential tech pod out there in the universe. Here with my incredible moonshot mates, AWG, back with his orchid-filled room, DB2, in his headquarters of all exponential investments. And of course, Seleme is on the road. I mean, remember the book "Where's Waldo?" I think we're going to replace that with "Where's Seleme?" So Seleme, where are you today? I'm in a car in Guadalajara in Mexico transiting to the airport. And this is the only way I could do this is to do that in the car. So hopefully the friend's hotspot will piggyback you off last. I can't believe he brought up "Where's Waldo?" Peter, do you know where is still the exclusive light and sea of "Where's Waldo" for data mining? OK. We used to go to trade shows and we'd have an actor dressed up in that "Where's Waldo's" suit and we'd be like, "Hey, our neural nets can find anything in your data. It's like a "Where's Waldo?" And we gave out all the books and it was amazing. I remember that. So Seleme, you're in Mexico, the "Blessy" team is in Mexico and they're raving about the podcast, by the way. And so I guess we have a big fan base down there. We see it. We do have turns out-- Yeah, big time. About half the-- I was in a conference for about 11, 900 people and quite a few of them are avid washers. How about the rest? Did you convert them? Yeah, we got to think international. Whenever we're commenting on these topics, because everybody-- It's a big world. And everybody out there is watching-- My Spanish is not quite up to snuff to say, everybody should have watched "Moon Shots" in Espanyol. You know, there's translators now. I just-- I know. I know. I know. I did my-- I did my meaningful life session last night in Spanish with the translator. And you should have seen the translator at the end of the night, you was so fried. And of course-- I'm touring through India. What do you speak Hindi or do you-- No, I speak English. It's my native tongue, because I come from a diplomatic family. I'm pretty bad Hindi. I can get by. But you know, it's one of these where my grammar is bad, my vocal-- I'm just throughout words and open sticks. I can get through about 50% of our conversation. Well, we're at almost 500,000 subscribers. So next time you're in front of our large audience, tell them to push us over to 500,000. OK. I'll tell them right. Let's jump in. Another-- How about credible crazy week? Let's kick it off with a conversation around the AI race and the agentic boom. So check out this slide. I mean, 15 major releases in only eight weeks. We're getting a pace of two major models per week. I think you've got to be retired and just focusing only on this to keep up. There's no way otherwise. So in this segment, what I love to do, guys, is really hit on the last three, KME K2.6, GPT 5.5, and deep seek four before their extraordinary releases, each of them hitting new capabilities. One thing, we saw the acquisition or the invoked acquisition of cursor by XAI. And I think what's interesting is that the winners in this crazy model race are going to be those that are providing the best abstraction layer. So it doesn't matter what models underneath. Do you agree with that? Yeah, totally. Actually, I just had a meeting with the data center company here in Cambridge. And the amount of effort going into the TPUs and the NVIDIA B100, B300s, is incredible. But at the abstraction layer, there's factors of five and 10 just being thrown away by mismanagement of the context window. And I mean, it's just so much opportunity in this stack, which makes sense because it's all brand new. But it's just-- and also, there's a lot of vertical integration going on. The warfare is really stepping up. But I can't believe how KME K2.6 is keeping up. I mean, it is just shocking that the open-source world is actually on the radar and keeping up. And we'll get to that in a minute. But what's interesting is the speed of these releases, I'm guessing that these new models are sort of-- it's competitive marketing, where the models are probably already cooked. And they're just waiting for someone else to release, and then releasing right on top of it. And Thropic is holding back on mythos. So you know that there's at least one case where you're exactly proven to be right, which means there may be others as well. But it's funny. The dot releases are coming faster and faster and faster. I mean, what's shocking about this list is it's US versus China. There's no European models. There's no UK models, no Japanese Indian models. It's just all US and China. Everyone else is a spectator. It looks like at this point. I don't know if you agree with that. But-- Well, the model is definitely self-improving now. Well, no, you're 100% right. But the models are self-improving now. And so the rate is accelerating. Exactly what singularity theory would have predicted. The rate is accelerating. But because the models are improving themselves, it's hard to start from a cold start and catch up. But I'm surprised that other countries aren't using the K2.6 model to bootstrap their own internal research. And maybe they are and hasn't popped onto the radar yet. But I'm not finding it too hard to design new neural nets using existing neural nets. It's a very doable thing. And I'm curious, Alex, that chart down below on this slide here that's showing all the leap running, it's leap running all the time. But is it that they're all just cherry picking? They're all just sort of studying for the test on the particular benchmark. And then they're just releasing whatever the latest benchmark that they're best at. Or is this truly-- No, I think we're down in the west to a three-way race at the frontier between open AI andthropic and Google. And I think those three labs have been pretty good about not benchmarking of over-focusing on just one benchmark. They're pretty good generalist models. I think we're seeing an honest to goodness arms race or horse race or rat race, depending on which metaphor you prefer. My friends at the frontier labs often call it a rat race. And as to the Chinese models, it's interesting. The aphorism, why do you rob banks? Because that's where the money is. So to the earlier point about why no European models wears Mistral in all of this, for example, it's because the US and China are where all the compute is. And ultimately, I think open AI is no one brown, who of course is quite famous for having led their reasoning approach. He's recently started almost pondering with a bit of on-we, whether the weights actually matter as much as they used to, or whether it's really turning into a race for compute. In some sense, as inference time reasoning becomes more and more important, his argument, not mine, but I think it's a credible one, the weights themselves start to become less important. And the same sense that, say, individual units within a transformer style architecture become less important as the transformer itself starts to scale. The overall weights for an entire model may become less important as more and more reasoning gets used. And you see, in effect, a spacetime transformer that's rolled out over time in reasoning token space. So if that argument holds-- and I think it's a pretty interesting one that I hadn't heard elsewhere before-- that would almost suggest that while at the same time we're seeing a race to the bottom on, say, per token intelligence densities between American models and Chinese models, open perenns. The American models are still about six months ahead. And this has been pretty consistent for the past couple of years, closed perenns. It may not matter in the end. What may matter in the end, at least according to the scaling laws we have at the moment, is who has more compute at the end of the day to do more reasoning. So-- I really see that in a minute. But 15 models over the course of two months is insane. Some of these are just improvements on existing models. And some of these are completely new pre-trained models. I think that difference needs to be pointed out. Celine, any thoughts on this insanity? Just the fact that we have that on many releases in eight weeks kind of blows my mind. We're watching the cost of cognition, coordination, execution is all collapsing at the same time. I mean, I think that is so much, not so much, to break through is the compression density is crazy.
Well, and the capabilities are mind blowing. These are not just fake little dot releases that are benchmarking. If you use them firsthand, and what's really helpful is if you look at our podcasts or at any postings on the internet from three months ago, six months ago, nine months ago, 12 months ago, and look at the predictions of capabilities. We're so far ahead of what even the upper bound of predictions would be in terms of the capabilities as the parameter count grows. So if you extrapolate from there, we're just on this knee curve of the acceleration and the singularity, and raw parameter count and more chain of thought reasoning is just going to push us to limits that are way beyond human. So what is the average person care about this? Like when my boy says, "Okay, great." And you release with a new numbers over and over and over again. At the end of the day, stuff is getting better, it's getting cheaper, it's getting faster. What does the average user, do you recommend someone sticking with a particular model? I'm just going to be on an open air, I'm just going to be on unthrobbing, I'm just going to be on Google. Any thoughts there? See, I think the question itself is a red herring. Why? Because OpenAI bet the company on consumers using all these reasoning tokens, that consumer oriented strategy for all of these trillions of dollars of cat-becks that they're building out would work. And they've had to pivot rather prominently in the past few months back to enterprise. So I think the question of what is the average user care, which I can still do as what is the average consumer care? I almost think the market is telling us the average consumer in the short term isn't even part of the equation anymore. This is really the question should be what is the average enterprise care? Because they're the ones. But I'm asking for our listeners, right? A lot of them are entrepreneurs or general consumers. At the end of the day, is it okay for someone? I'm just using Chatchy PT, I'm just using Gemini 3.1 Pro, I'm just using the latest version of Anthropics models. Is it important for people to be driving to the latest model or is it okay? Because ultimately, everybody's basically leapfroiding everybody else. And if you're just a mom, a dad, a student, and maybe an entrepreneur just getting going, this insanity of 15 models in eight weeks bouncing back and forth, I mean, Dave, you're using two, three or four models all the time, right? Oh, many more now. And that's the biggest change. The coordinator model can now manage dozens or hundreds of other models successfully. And six months ago or three months ago, that wasn't true. So for the average consumer, the ability for the stuff to install itself, like you can go into the model now, yet you have to use the latest ones. But it doesn't matter a lot whether you're using Cloud 4.7 or GBT 5.5, just use one of the latest ones. But ask it to install itself, ask it to build something on your laptop for you and it just works now. You don't have to understand the Linux command line, you don't have to understand any of the underlying infrastructure. It's smart enough now to explain itself to you as it goes. So I think for the average listener, that's a massive unlock. There was someone who's never built software before can just think of something and then create it in an hour. And that just wasn't true, you know, six months ago. Hey, everybody, you may not know this, but I've done an incredible research team. And every week myself, my research team study the meta trends that are impacting the world. Just like computation, sensors, networks, AI robotics, 3D printing, synthetic biology. And these meta trend reports I put out once a week, enabling you to see the future 10 years ahead of anybody else. If you'd like to get access to the meta trends newsletter every week, go to dmandis.com/meditrends. That's dmandis.com/meditrends. Yeah, let's jump in our first model story here, which is moonshot AI launches, Kimi K2.6. I just downloaded onto my max studios this weekend on top of Skippy, who's orchestrated by Opus 4.6. So Kimi K2.6, it's a trillion parameter, open-weight, open-source model, activates 32 billion of the parameters at a time. It runs 300 parallel agents. Very importantly, natively, it can process text, image, and video all at the same time. It costs 30 times less than the most capable, closed models. Starting enough, you know, moonshots AI didn't get its name from us. The three founders based in Beijing. Their favorite album is Dark Side of the Moon. And so that's where it came from. The company's backed by about 4.7 billion in capital from Alibaba, Tencent, and IDG. And this model, right, if you look at the numbers on the bottom on the benchmarks compared to GPT-5, 4, Opus 4.6 and Gemini 3.1 Pro. It does amazingly well against all those models. And this one was trained, they report, for a total of 4.6 million dollars, compared to hundreds and millions or billions on the other, you know, closed-source models. Dave, I mean, I find that amazing. I mean, almost incredible. So almost incredible. It's so much to say about this. You know, it was starting for the fact that, yeah, Alex said a minute ago that, you know, that the Chinese models are running about six months behind the US models. But if you look at the benchmarks, this is up there, or beating Club Opus 4.6, which was only three months ago. That came out in February. And so that's not a six-month lead. That's a three-month lead. And the price performance, you know, most people, when they first start, they don't care too much. It's cheap, you know, all the eyes are pretty cheap. But then when you realize that you can run 10 or 100 of them concurrently, like, well, this is going to start to add up. So if you run this on fireworks AI, it's about one-eighth the cost of running the Cloud API. Well, or the OpenAI API. So, you know, one-eighth is a pretty big price cut. If you download it and run it like you did with Skippy, then you're running it about one-thirtyth the cost. So that's a big, big deal. And then, of course, the caveat is, as Alex has pointed out many times, you're not 100% sure if it's not, you know, spying or doing code injection, it's probably not. But you can't guarantee that. So that's that's, you know, so somebody tells me, this is one-thirtyth the price. Try it. You're like, mm, I'm a little sus. Like, why is it one-thirtyth the price? But I doubt it's code injecting on you, but you can't be sure. Whereas if you use it in Thropic or OpenAI, it's definitely not code injecting on you. In fact, it's safeguarded all over the place. So there's your landscape. The caveat is always, it's only going to get more chaotic. Alex, how big a deal is, can we, K2.6? I think it's helpful for certain enterprise use cases where you want to be able to self-host the model and you don't want to say use AWS bedrock, which by the way, now hosts GPT 5.5 in addition to the opus models. I think it's helpful in that respect. It's helpful if you want to be able to self-host fine-tuned models for yourself, ditto with deep-seek V4. But I think in general, again, it's a few months behind for other use cases, for consumers that want to be able to self-host for whatever reason privacy or otherwise, probably very helpful for folks who want to self-host their own clause, very helpful. So I think there are many use cases where these typically Chinese open-weight models like K2.6 and deep-seek V4 are very helpful. I do think, however, they're not at the frontier. And to me, the big headline is that disparity between the American frontier closed-weight and the Chinese frontier open-weight seems at least for the moment to be in place. Peter, I think your setup is perfect. It's exactly what I do, too. I use opus 4.7 as my work-estrator because you want that extra notch of intelligence. And then if you have simple tasks or you're just sub-tasks, you can farm them out and save the money using K2.6. And then if the results coming back don't make perfect sense, then your work-estrator will tell you, "Hey, this is garbage." And so you can actually rely on opus 4.7 to give you the straight truth on what the underlying models did for you. It's exactly the way you set it up, Peter. David, a week ago you said you moved from 4.7 back to 4.6. Did you move back to 4.7? I have both running now. 4.7 is kind of wordy and sounds kind of PhD-ish, which annoys me sometimes. And 4.6 is friendlier. But then it's clear that 4.7 is a little smarter. And so sometimes you just need the right answer no matter what. And so I actually have both running in parallel agent windows now. Selina, I'm curious in Guadalajara, Mexico City, in parts of South America, in parts of Asia. What are you hearing about the use of US models versus open-weight open-source Chinese models? So I get a mixture of both things. A bunch of people use the hosted models, the big ones just because it's easy. There's a subset of people who use the open-source models and the Chinese models. And they don't really care. I think they should care at some point that's going to come up. One question I have for Alex and Davis. How do you protect against the code or prompt injection in these open-source? Is there a way of defending against that? If that's clear, if there is, then there's a huge case for this because everybody here is looking for the low-cost approach.
right? But for the most part, I'll be blunt. The conversation is not around which model and open-source are closes. Like, what do we do with AI? Like, it's literally at a level of lack of sophistication around this that you would expect. But the option is also there for startups then to leapfrog lots of people and build aggressively for the coming madness that's upon us. Yeah, it's almost an impossible question, Selene, because if you sit on this sideline and you don't use this stuff aggressively, you fall way, way behind. But if you start using it aggressively, you're generating thousands or millions of lines of code before even know it. And so then the odds go up, right? So I think what you're trusting right now is that the guardrails that are in Thropic and OpenAI put on their models, they're very, very cautious when they're pulling in code, open-source or otherwise. I mean, almost annoyingly cautious. So you kind of assume they've done a very, very good job of filtering out nasty code injection. But the numbers work against you at scale. So there's no simple answer. I mean, you can, when I got into it, I was like, "Hey, I'm just going to look at the code. I'm not going to just run it. I'm going to see what it does." That's a joke, right? That's just laughable. It's generating so quickly now that there's no chance you could even scroll through it. So you have to use AI. It's like a lot of things, actually. You have to use AI to protect against AI. There's no other way to get the scale. So it's tricky. I know that wasn't much of an answer, but it's tricky. You know, one thing I'd love to point out here, we talk about this on occasion, but I don't think we've ever really spoken about in detail. The Kimi K2.6 uses something called a mixture of experts and MOE. And it's interesting, and just take a moment about this. If, in fact, you have a trillion parameter model, and you ask a question, it's basically accessing all trillion parameters every time to analyze every token. And what they did here is they actually created a set of 30 plus experts. And so that, you know, some percentage of all the parameters are dedicated to one expert system. So if you ask a coding question, the orchestrator looks at this and says, okay, this is a coding question. We're going to send it to experts number 3, 7, and 12, and only uses a portion of the parameters. And only uses, instead of all the experts, all the experts, it uses some sub-fraction thereof in it, saves money and saves time. And how many different models are using that right now, Alex? Sparsity, which is the term of art, I think that we're talking about here is endemic to all frontier models at this point. It's also the basis for the human brain. If we look at the brain, most neurons don't at any given point in time have action potentials that are going in and out. So, sparsity is a great way to reduce the memory footprint of models. To my knowledge, all of the frontier models use sparsity one way or another. It's also a good way to another term of art regularize the models. So to make sure that particular weights or parameters in the models aren't overfitting to the training data, one of the age-old techniques is just blasting away individual weights or parameters in the neurons, making them disappear entirely as a so-called regularization technique. So, sparsity is everywhere at this point. And it's only going to, I think, become more important with time as we try, you know, what one of my holy grails is, as I've mentioned on the pod previously, I'd like to see a million parameter or smaller diamond or black hole of a model at the end of the scaling race. And I think sparsity and cranking the knob on increasing sparsification in these models is one possible path to getting us there. Hey, just to add something that what Peter said, the M.O.E. innovation, mixture of expert innovation that came from deep sleep is actually layer by layer. So almost these neural nets are about 140 layers deep now. And it'll route the expert layer by layer. So it'll say, look within this layer, I'm just doing basic image classification and this layer, I'm doing deeper thinking in this layer, I'm doing higher level math, as it moves through the neural net, it'll actually route to, you know, now I think up to 128 different experts layer by layer. And so it'll find the optimal pathway through the entire neural net. On top of that, you can also have dedicated experts like here's a surgeon, here's an artist, here's a coder above and beyond that. But M.O.E. is actually within the neural net layer by layer. All right, next story is opening I unveiled GPT 5.5 literally just seven weeks after GPT 5.4. Greg Brockman calls it a new class of intelligence. It's natively omnimodality, it's able to process text and audio and video and images all in the single unified end and architecture. It has a 37 point increase in over 5.5 over 5.4 in long context reasoning, which means 5.4 and 5.5 are both a million token windows, but 5.5 can actually remember the beginning of the million tokens and provide, you know, complete context across the entire thing. Token efficiency, 40% fewer tokens with the same latency and I love this, hallucination is down 60% over 5.4. Let's go to our resident genius Alex, what do you make of 5.5? How important is it? I think it's very important both intrinsically and also relative to 5.4. So I want to highlight two key stats here. The first is the leap from GPT 5.4 thinking to 5.5 thinking. That's probably the biggest leap overall on terminal bench 2.0 specifically. So one way to interpret this terminal bench is a benchmark that's focused on the ability to agentically operate from a command line terminal. Where is that useful for codex and for cloud code type environment? So one way to construe this huge leap, which is larger than most of the most are all of the other leaps that we see in terms of other benchmarks, is that 5.5 is being very seriously focused, benchmarks, if you like, although I really haven't used it, don't think it's narrowly overfitting just to creating and making codex a better cloud code competitor, but it very much feels like a release that's intended to strengthen open AI's codex. That's thought one. Thought 2 is my favorite benchmark among all of these is frontier math tier 4. Frontier math tier 4, which I think we even had a new year's bet about that we're going to have to revisit sometime later this year, is one of the best proxies for the ability for AI's to solve professional level research problems in math. And what do we see? We see from GPT 5.4 pro to 5.5 pro approximately 2% leap in approximately the last two months. What does that tell me? That tells me that we're seeing now approximately 1% gains per month in research level math coming from frontier AI's and we're getting closer to approximately half of all of the frontier math tier 4 problems getting solved. So you can extrapolate this and realize if the present rate just stays the same, which I guarantee it won't, it's going to accelerate, but even just at the present pace, we're talking about essentially all frontier math tier 4, all professional research grade math problems being solved in the next four or five years. So math is cooked. We know I'll say it, you know second time, math is cooked, a bunch of other things are cooked as well, but things are moving so quickly now that on a month by month basis, we're able to see the hardest of these benchmarks creep up 1% per month. So not long now. It's worth pointing out that the API pricing on 5.5 is twice that of 5.4. So it's 5 bucks per million in put tokens versus 2.5 dollars on 5.4 and 30 bucks per million output tokens versus 15. I like this simplicity, that pricing. Dave, are you been playing with this at all? Yeah, absolutely. And I think what Alex said earlier in the pod is really, really important and insightful. Like, no one brown is saying, wow, maybe the weights don't matter so much as this chain of thought processes is just way ahead of any expectations on how intelligent it can get. From a user's point of view, that first benchmark terminal bench, if you ask it to do something complicated, like configure an entire system for you, downloads some software, integrate it, make it all work, you know, connect it to my outlook, connect it to my whatever, it just works. And that exactly ties to that first benchmark. It just flat out works. And so it just, it feels like this incredibly capable brilliant assistant, no matter what you're trying to do because of that first benchmark. Then the last benchmark, the frontier, or second or last frontier math, you know, Dennis Asabis came out and said, yeah, I think it's a kind of a coin flip. This was on Alex's innermost loop. You know, it's kind of a coin flip now on whether just the existing architecture scaled up solves everything. Yeah, I think I think coin flip, he's moved a long way. He has, you know, we need new breakthroughs. We're out of breakthroughs apparently. I remember 10 plus years ago when I was chatting with Demise, he used to say there were five breakthroughs.
remaining between where we were then and AGI as he construed it. Now we're out of them. It's half a breakthrough or zero breakthroughs at this point. Next week we're going to be on with Ray Kurzweil again. We're doing that May 4th event for the launch of We Are As Gods. And I'm curious we should ask him, you know, what does he think? What's required to get to true AGI or ASI? We're just going to extrapolate what we're doing or do we need breakthroughs? I think that that requirement's been falling. It is a lot less than it is. I mean, it's starting to feel a lot like, like actually Alex, you know, it would be great for that is to put together a chart of Dennis's number of breakthroughs, which because at Davos it was down to two. Now it's down to 50, 50 that it's zero. But you mentioned five. That was what maybe a year and a half ago. So that would be a very legal chart. The five numbers of him was when I was chatting with him in, this is 10 plus years ago. Yeah. Okay. Well, there are some sort of exponential decay of breakthroughs clearly. Alex, you said it a little bit earlier. You know, this is ultimately a compute race. Oh, let's talk about that. You know, a couple of a couple of stories here around Google Cloud and Google Cloud is dominating. So what do we see? We see Google announcing at Google Cloud next to 2026 their major conference. They unveiled their eighth generation of TPUs in particular, TPU 8T for training and TPU 8i for inference. Now we have training and inference chips separately, just like Amazon has their training chips for training and their inforrencia trips for inference. These new GPUs are three times faster in training performance, 80% better performance per dollar. They're designed to run millions of agents in real time. So Google is really all in on the agentic era. Sundar Pichai, the CEO, who I had a chance to spend some time with last weekend. He made it crystal clear. He says over 16 billion tokens per minute being processed and 75% of Google's code is now written by AI. So fascinating. Dave, what do you make of this? You know what's surprising to me is that the the price performance of the TPUs is landing right on top of Nvidia, not much different at all, which is surprising because it's a completely different architecture. It uses a systolic array design. I mean, it could not be more different from a GPU under the covers, but for whatever reason, it's all kind of canceling out and landing identical, which is fine from Google's point of view, because now they have their own total chip fab through data center, through models. I do everything. Solution, yeah. I still believe Google's the winner in the long run here across the board. Terrible. Terrible. Terrible. Terrible. Terrible. Terrible. Terrible. Terrible. Terrible. Terrible. Terrible. Terrible. Terrible. Terrible. I'm still I still believe Google's the winner in the long run here across the board. I don't know if you in the Anthropical System, you had TerraFab on the other hand. - Google owns a material percentage of SpaceX. - They do. They do. - And in the network. - I don't know if you saw, there was a tweet out recently about Google's investments that were made. - Yeah. - And they just had massive returns on their investments on SpaceX, on Anthropical across the board. - Huge returns. - Yeah, I was at a board meeting yesterday company, the chairman of that has massive cash flow and a huge cash balance. And they were like, "Well, I don't know if a public company can really do seed stage investments." I was like, "Have you looked at Google?" (laughs) They have multiple hundred billion dollar gains on their investments. And they don't even do it for the money. They do it for the knowledge. And for-- - Well, and strategic relationships, right? Larry and Sergey were just bonding with Elon and they said, "Okay, Google, it's gonna invest a billion dollars." And now it's worth, you know, God knows how many hundreds times more. - Yeah, that's interesting. - It's also investing in the future. - Right, I remember conversations with Larry and Sergey about the nature of the frontier. And I think to their credit, they're investing in the frontier and SpaceX is part of it. And also compute. Epic put out this really, I think eye-opening stat in the past week that Google now accounts for approximately quarter of all of the AI compute on the planet. And I'm sure eighth gen TPUs will be part of it. I think it's also worth keeping in mind that the TPUs at this point are being designed by TPUs. I have number of friends at Google who are responsible for designing next gen TPUs. And they're all just using Google AI to do it. The recursive self-improvement goes all the way down to the silicon at this point. - Our next story in the Google ecosystem, again, also announced at their large cloud next conference is Google commits to 960,000 Nvidia VR Rubin GPUs for their A5X. So pretty extraordinary, A5X is Google's new bare metal virtual machine instance, delivering 10X lower inference costs and 10X higher token throughput. Just an interesting FYI, VR Rubin for whom these chips are named was an American astronomer who discovered the first conclusive evidence of dark matter. I love the fact that Jensen is naming chips and systems after famous individuals. Now, Wi-Fi and Fastening, this goes back to the conversation a minute ago, is that this cloud is two times bigger than class is two and 2.4 times bigger than Stargate, Abelene. So Google is winning on at least based on what they're building and plan to build. Again, Dave, thoughts here. - Well, you know, partly, just to touch on one thing, he said there, Peter, part of the acceleration we're seeing in society as a whole is that all the really, really smart people are working on real tech now. - Hardware. - Hardware. - And space and medicine and real tech. And if you go back to the meta era, the Facebook era, the rewards were all in either cheesy consumer experiences or banking. And doing deep tech was kind of a way to die poor. - So it's creating a whole new era for society. The post-day era, we all knew it was going to be very, very different, but now the rewards are in actual deep tech that benefits humanity in really big fundamental ways. But I think if you just counted the number of people that you know that have been pulled into this vortex, it would have been just a few percent working on world changing deep tech real stuff just 15, 20, 30 years ago. Now it's almost everybody that you know is getting pulled into like, you know, do something big and world changing and it's actually working. And so that's a big change for society. So that's helping accelerate things as well. - All right, our next story is anthropic is cutting deals for cash and compute. I mean, huge amount of capital flying back and forth between the frontier labs and the hyperscalers here. So Google commits to a $40 billion investment in anthropic. So last week, Google committed to a ton of money, $10 billion in cash right now at a $350 billion valuation. And note, you know, we talked about this last time and anthropic on the secondary markets is now at a trillion dollar evaluation. So this $350 billion is coming in at roughly one third the cost of what others are paying for it. And they committed to another $30 billion if anthropic hits certain performance targets as well. They're gonna be providing five gigawatts of TPU compute committed over five years. That's the equivalent of, you know, literally providing power to three to four million people. I'm finding this pretty extraordinary. We're gonna see in a moment a conversation where anthropic is cut deals with Amazon in a similar fashion. Actually, let me go ahead and hit that and we'll talk about this, these money for guns conversation that's going on. So Amazon and Anthropic are trading cash for compute. So here's the second deal. Amazon is investing a total of 33 billion. They've committed to 25 billion on top of the eight billion they've already invested. In return for Amazon's cash and anthropic is committing to spend $100 billion or more on AWS over the next decade. Anthropic will run clawed on Amazon's custom train chips and Amazon will provide five gigawatts of AI compute capacity for anthropic. So, I mean, we're seeing Anthropic becoming beholden to both AWS and Google in a significant fashion. Gentlemen, thoughts on this one. - Well, it's so funny to me. Like obviously, Anthropic needs much, much more compute and is growing, oh, actually a very good friend of ours, Peter. I won't mention him on the podcast, but he's an investor in Anthropic and he was telling me at the board meeting yesterday, he can figure it out from that comment. But he was telling me at the board meeting yesterday that Anthropic under the covers is thinking they might hit between 40, 50 up to 70 billion in revenue by the end of the year. - We talked about 100 billion by the end of the year a few pods ago. But still, there were 30 billion last month doubling, tripling, it's extraordinary. And the current year wouldn't hit those numbers as because they can't get enough compute to keep up with the demand. - And one of the demand was that they didn't release mythos because they have enough compute to deal with it, right? So, it's a limited, a limited release of the capabilities. - Yeah, an OpenAI cut Sora, I think one of the reasons is probably compute. So, which is energy. - Which is energy, yeah, which is energy. And I think that it's so funny to me to see all these deals. So okay, so Dario needs compute, he signs up with Amazon. They're all, and now OpenAI is gonna be running on GCP and also on Venrock on Amazon.
So you can get it through bedrock. So everybody's partnering with everybody else, but it's all bottlenecked at TSMC. Like this is all great. You can all partner with each other up the Yin Yang, but whose chips are actually gonna get made, you know? And you'll see TSMC in any of these podcasts, in any of these deals and these meetings. And then you saw Jensen actually recently say, he doesn't have any long-term agreement with TSMC. They just kinda make it up as they go. So all of this is bottlenecked. And only Elon is talking about, look, the fundamental constraint to all of this is the TerraFab. And I already locked up 16 billion, could be 45 billion of Samsung's capacity. The only three companies in the world capable of making any of this are Samsung Intel and TSMC. And like that's the actual bottleneck to all of AI. Alex, is it compute or is it energy in the day right now? I think they're indistinguishable at this point. I think permitting for onsite energy is a major limiting factor. I think it's probably unbalanced more of a limiting factor at this point, maybe not a year from now than TSMC. But it is a limiting factor. Having powered land, having data centers that you can take all of these, you know, infamously, Microsoft, even in the past few months, spoke about having lots of GPUs that they'd love to rack mount in a data center. But lacking the powered land and lacking the data centers to plug them in. I think at this moment, energy, at least in the US, but I agree with Dave that in the medium to long-term, semiconductor fabrication supply chains, doubly so if there's any geopolitical conflict are like clear to be a stranglehold once we solve our energy story. So let's talk about the not investment advice segment here. You know, where do you invest your capital? Like, you know, if compute an energy, I mean, I'm seeing the energy stocks beginning to fly, right? A friend of mine just had this IPO of X energy and it popped like 30% in the first day. We're seeing blue energy and other energy stocks beginning to skyrocket creep up over the time. So, you know, I don't know if you're going to invest in ships. Do you invest in, we saw Intel pop up in AMD. I mean, all of these guys, you know, that entire ecosystem of chips and energy, ultimately if they're really the constraining part of the innermost loop here, I think the most, you know, most demand is there. Any thoughts, Dave? - Oh, so many thoughts. You go for an hour on just this topic, but, but invests like crazy in anybody who has access to chips and can find a power supply. (laughing) That's, you know, pretty straightforward. The power supplies everywhere, all these legacy manufacturing operations, aluminum melting and all that, uses a huge amount of electricity and swapping it over to data center is a massive increase in the value of that energy supply. But you have to have a, you know, a line on the chips. Then at the kernel level, you know, because the chips are so constrained and the demand is through the roof. At the kernel level, anyone who's writing software at the kernel level that empowers, you know, AMD chips or, you know, legacy GPUs to participate or just makes the inference more efficient. On Nvidia chips, those companies are worth a fortune. So anyone who's building kernel level software is a brilliant investment. And then in the vertical use cases, Anthropic rolled out something called skills, which you should absolutely play with, is just a way to use the context window more efficiently by designing skills that the AI can then pull in. So rather than have to reinvent everything every time, just build a skill. - Yeah. - And then you can call on the skill very efficiently. So companies are now discovering they can refactor their entire business or their entire, whatever they do around 100 or 1000 different defined skills. - Yeah. - But those skills then become the defendable intellectual property within that vertical domain. So, you know, any vertical domain where you're racing to build out the entire skill database for that, for that use case is also an unstoppable investment theme right now. - You know, do they have to go forever? - The other thing that's interesting is that both Google and China and Amazon are getting their shares in Anthropic at one third the going rate. I find that extraordinary, you know, $350 billion valuation versus the trillion dollar valuation. - Yeah, it shows you how important the compute is. I mean, again, you're gonna be sold out for forever if you can get the compute. - And these hyper-scalers are kind of hedging their bets, right? They're not picking a winner. They're buying every horse in the race, you know, 'cause this, you know, AGI ASI race is just way too important to lose, so they're just investing left, right, and center. - I would also just parse these as the market doing what the market does. Some of the participants, some of the frontier labs like Anthropic have an insatiable hunger for the compute and they have the revenue generation to generate the demand and sustain the demand. And so if you're Anthropic, you're going to go to every possible source at scale of compute that you can find, whether it's Amazon or whether it's Google or whether it's other sources, you're just going to go and seek as a hungry customer for compute whatever the market will provide. I don't think necessarily the story needs to be any more complicated than that. It turns out the world demands a lot of compute to solve some of these really interesting problems in code generation and otherwise. And what we're going to see over time is all of this demand is going to translate into supply. It's going to translate in the short term into what looks superficially like a bit of a circular economy between call it the top 10 or 12 companies after we see the IPOs of SpaceX and OpenAI and Anthropic. But that's going to diffuse throughout the economy over the next few years would be my prediction. - People are hungry for compute. Selene was hungry for bandwidth. Selene, welcome back. I see you're in this situation. - Yeah, from the airport now in a stationary spot. So let's see. - Dude, I'm just going to call you Waldorf from now on. All right, let's move on. A couple of fun stories. I'm going to add this segment every time for the podcast, which is what did Claude just kill. So this is the stock chart for eBay. And this comes out from Anthropic Research. It says new Anthropic Research Project Deal. We created a marketplace for employees in our San Francisco office. With one big twist, we tasked Claude with buying, selling, and negotiating on our colleagues' behalf. Basically doing what eBay does. And we see a drop in the stock price. I think this is, eBay's not really dropped anywhere beyond this. But I think this is going to be more and more common. Any thoughts, Dave? - Well, I think a lot of this is just a immediate knee jerk fear reaction, but then things kind of settle out and you realize, wait, Anthropic is going to build all kinds of marketplaces because they can. But it's not going to hurt eBay. I think what you're going to see more and more is AI is growing so quickly that it's going to largely grow around the legacy economy. So around the banks, around the insurance, it's just going to be its own world. And it's going to be feeding on itself and building just colossally large constructs that some people are not even aware of. And it will all happen very, very quickly. So I think eBay will be fine. - Absolutely. - Any thoughts, you're slim? - I have a slightly different take. There's so many places because lots of problems and companies exist because coordination is hard. And AI makes coordination easy. And that's going to threaten big chunks of places, market places, customer support, listing optimization, dispute handling. There's a huge categories of these that will become agentic workflows. And I think the bigger question about what to the AI just kill is what workflow category did it just in confidence and automate? - You want to hear something cool, like related to this. The data center CEO that I met with this morning, we were talking about data centers going into space because power is basically free. Solar is basically free in space. And he said data centers, there's power all over the planet that's not tapped, that doesn't disrupt society at all. That's not why data centers are going into space. Data centers are going into space because there's no regulatory authority preventing it. - You try to do anything on earth. - Well, that's not true. You still have to, if you're going to be flying all these data centers and communicating, you need licensing domestically and the ITU for bandwidth. I mean, there are going to be regulatory hurdles that Elon and Google need to get, especially if you're launching 500,000 satellites. I mean, when you're putting up a debris field like that, there's going to be pushback. There's going to be pushback. - Yeah, it's interesting. If you can hear that. - I'll square that circle here. - I'll square that circle here and say, I think in the short term for Sun Synchronous Warbit, yeah, that requires FCC and other approvals in the long term. If we start to say launch AI data centers from the moon that will probably, and we're building them on the moon, that will probably require fewer approvals, at least under the current regulatory regime. - I'll take that. - I'll take that. - That's 20 years away to actually get manufacturing on the moon. I'm talking 20 years away, Peter. Listen, if you look at it deeply, I mean, I know 20 years away is infinity. I get that. But we're talking about, I mean, just to be clear, the stuff I'm concerned about is the next five years, right? If you're launching, we talked to Elon about this.
500,000 V3 satellites in a constellation, there's going to be debris issues. Elon pushed it off by saying, "Oh, we'll have superintelligence to figure that out." We have this, everything looks amazing from far away. But the reality is by the time it comes closer, there are real issues. And so it's not going to be just the promised land of going to space. We're going to have challenges going there still. Yeah, it's interesting how the timelines line up too, because between here and there, there's all kinds of constraints, but between here and there will have solved all math and will have discovered all kinds of new physics. And so on the space, I'm the super space enthusiast here. I can hope for nothing more than that vision to happen, but it's always easier on the promised land. Peter, I'm gobsmacked to hear that you think it's going to be 20 years before we have fabs on the moon, my goodness. Fabs on the moon, manufacturing and pumping into Earth orbit with mass drivers. You think that's 20 years away? No, okay, maybe 15, but it's not the next five years. Do I hear 10? That's hard for me. It's hard for me. I guess Optimus Robot will improve that. Demand will improve that, but the concern is if you have an unexplosion, but a collision of spacecraft and orbit generating debris, we still don't have any mechanism for moving debris from orbit. And so it's going to be a challenge. I'll make two. Your concern is Kessler Syndrome? Kessler Syndrome, yes. Is going to sabotage moon-based fabs? No, it's going to sabotage the next five years of 500,000 satellites in Earth orbit. I mean, right now we have 10,000 satellites from Starlink, which is the most ever pumped into orbit ever. And we're talking about 50 times that. And we're talking about not just the US, Amazon's going to do their best, Jeff is not going to stand still while Elon's doing this. And then you've got Chinese constellations. So do you double or triple that number of satellites in orbit? I mean, listen, I can't wait. And it's going to have challenges. Selim, you're going to say, Yeah, I'll give a couple of thoughts here. With just finger in the air here. I think human or robots are five to seven years away minimum. In mask at mass scale in widespread adoption, okay? Minimum. And I think that's okay. I agree. I understand. And I think a fab lab on the moon and consistently doing fabrication and all that stuff is 15 years away minimum. So I'll say that. Not to this not coming. It's just a question of it's a when not an F, which is my goodness. This is lunacy. Outer lunacy here. We are the moonshots podcast. Yes. Can we get back just maybe to project deal and anthropic? I think we're missing an important point. Everyone who hand rings over the latest anthropic project, purportedly sabotaging or killing some SaaS company. Anthropic doesn't want to be triggering SaaS apocalypse is left and right. There's relatively little economic motivation there. I think if you look through the through line, through all of these anthropic projects or research projects other than the alignment ones, anthropic is can all of their projects can be explained by end corporate strategies and unhovelings can be explained by very simple principle. That's all that they're trying to do of that. Clod code. It turns out through Claude code code Gen is actually quite economically valuable. Per token. It turns out per token it's more valuable to generate useful working code than say to generate video or cat images or whatever other consumer plays open AI and some other frontier model providers were chasing. They've dropped that now. Everyone's focusing on code Gen because on a per token basis, it's so economically valuable. So I would look at projects like project deal running a marketplace, running a business as anthropic looking for new ways to increase the per token economic value of their output. It's as simple as that. I think the business. Brilliant Alex. Yeah, that's absolutely brilliant. Thank you. The move is so long here. We're coming into the battle season. It's Elon versus Sam and open AI. This just got posted today. So today is a start of a very important day in the AI world. The trial between Elon and Sam and open AI begins in the Oakland federal court. Jewish selection is happening right now. So I just put this up to keep us posted. We'll be learning a lot. Of course, discovery is unveiling a lot of texts, a lot of emails that I bet both Elon and Sam and a lot of other people would rather not have aired in the public. Any thoughts here, Gents? I think it's sort of said that it's come to this. It's going to make just one remembers the Bill Gates versus Steve Jobs docu-dramas that were made from critical Apple versus Microsoft era. This has, I think, a similar feel to it. It's sort of said that I think this ended up in court versus settling earlier on. But I do think many will, I think history will probably view this as sort of an iconic struggle that will get the full Aaron Sorkin, if not similar, like movie treatment. This will be the full Hollywood type, totally great. Titanic battle. Yeah. Selim, any thoughts here? How is this playing in Guadalajara? No recognition awareness at all. And that's probably a good thing. This is kind of so proper. I'm with Alex on this one. It's just heavy drama. We wish it hadn't come to this. We've been great to get these guys to settle off thing. But their positions are hard and baked in. And so it's been a time like this. How do you unravel the movement from open AI to a for-profit company? I mean, do you back it up to a nonprofit? Then what about all the capital invested in open AI? Does that disappear if they lose the case here? I'd like to do it more. I mean, I've said, I think, on the pod and past that if the model of changing non-profits, large non-profits to public benefit corporations can be scaled, I'd love to do this to a number of major American research universities. That's not my question. My question is, what happens to all the capital invested? Literally hundreds of $122 billion in the last couple of months. There's so much pressure for this court case, not to be won by Elon Musk. Well, I mean, if you're following the detailed TikTok of the way that this trial is being structured, it's being structured in two phases. The first phase is more of deciding whether the claims that Elon at all have made are, in fact, the case in the second is the equivalent of a reward type section, deciding what awards if any to make as conditioning on the first phase. But I think there are a number of details in this court case that are notable. One is so jury selection. There's been public reporting that already selected members of the jury are aware of entanglements that Elon's had with the President's administration and may view him negatively as a result. I think that's the fact that jury members are being selected reportedly with some political influence seeping in. I think that's very interesting. I also think it's interesting that the district judge in this case has, again, reportedly decided that she's going to take the jury outcome as an advisory opinion, but that if there is an award, she's going to decide ultimately from the bench on the final award. So there are a lot of nuances here. Wow. Dave, any thoughts opinions here? Yeah, do we ever figure out if we get to see it live live? No, it's not being broadcast, but I'm sure they're going to be your court reporters giving us a lot of details here. You can wait for the full Hollywood treatment in a couple years. Yeah, by the way, there will be a Hollywood treatment of this. It's guaranteed. Every other major. Of course, it may be an AI-generated feature film, but nonetheless. It will be for sure. Well, I'm surprised how many texts, like personal texts, have already come out. Yeah. The emails get discovered right away and everyone in all your email gets thrown out there for the wild to read, which is crazy, but it happens. But texts traditionally have not been thrown out, but yet we're seeing them all. So I don't know exactly how that's happening, but for Elon to win, he doesn't have to win the case. He just has to slow down open AI. I mean, in the middle of the singularity, if you lose three months, you're basically lost. You're a good. You're a good. All right. Another fun topic. A few stories here. It's about AI surveillance and privacy. So let's check this out. Open AI's Chronicle uses agents to build memories from screenshots. Sam Altman described this one as telepathy like. So Chronicle runs on Open AI's codecs, where background agents are taking periodic snapshots of everything on your screen. The screenshots are sent to Open AI's servers for processing agents use optical.
character recognition and visual analysis to extract the context of what you're doing every minute on your screen, structured memory files are created and stored locally. And we talked about this before, AI monitoring everything. Ultimately, it's sort of the camel's nose under the tent of being able to replace any worker. We have significant privacy concerns that come up on this. And no one's raising that. I don't know if you guys remember when I was researching this. So Microsoft had launched something recently called a recall. It was a product that they put out there and then they retracted because all the cybersecurity people said this is a privacy nightmare. It's litigation bait and they pulled it back. But when OpenAI announced this product, no one's pushed back. Can I first of all point out what a beautiful double-entendre from Microsoft's crack product marketing department naming a feature recall and then recalling it? I think what we're seeing here is one big architectural clooge. And I think it's going to be cloogey both from Microsoft's perhaps ill-architected recall as well as OpenAI Chronicle. This wants to be built into the operating system and the hardware. It doesn't want to be an add-on. I don't think I'll just speak for myself. I don't want an agent taking constant screenshots of my desktop, sending it to a server and then parsing it, sending back results. This should all be built at Apple style. I would hope that Apple will get its act together in the next few months and build this into the window manager and the compositor and the operating system. The operating system is rendering the screen. Why can't the operating system understand what it's rendering? This is ambient AI is the term of art here where AI is monitoring everything all the time and enabling you. This is in one sense, this is what I did this past weekend with my OpenClaw with Skippy where I gave it access to everything. Every single granola gets put into memory, every WhatsApp message, every email, every calendar, everything. It just makes it so much more useful. And I think something like Chronicle as well would just enable it to be like Sam said, telepathy. Well, that's the quandary. I mean, a lot of people who get in trouble with AI or they get stuck, it's something they're doing on screen. The AI doesn't have visibility into it. But if you unlock that, the AI can be incredibly helpful, but it's also seeing literally every mouse mode. But when we talk about our moms are still not using AI, why not? This is a big unlock. The voice interface and this are the two big unlocks because it can then say, oh, I see what you're doing wrong. In fact, let me just do it for you and save you the trouble. And all these configuration screens on any Apple device and the menus are ridiculous now. The number of layers of configuration you can do. I think something like some crazy stat, like 70, 80% of all iPhone users never change any defaults. Yeah. It's just too confusing to do anything. So this is a huge unlock for all of that. Right now it's hugely intrusive. Right now, I take screenshots and I send it to Claude or whomever and say, hey, can you please help me forget this out? But this is going to have sort of an expert over your shoulders. Always there to support you if you need it. Well, I think it's when they first start playing with AI. Like Alex's standard first query to test a new model is build me a first person shooter. That's a better prompt than that. Sorry. That's what I'm saying. But people want to do something visual and graphical to learn how it all works. And then when it doesn't work, they want to show the AI, hey, this doesn't look right to me. Fix it. So they screenshot it just like Alex or just like Peter, you just said. They screenshot it. But here, this is just a much more convenient way to get video, not just a screenshot, back into the AI's brain and say, look, this doesn't look right. Fix it for me. And so you have a much more fun dialogue with the AI. But you have to accept that privacy is being compromised there. I think I don't, I'll take a very different position. Please, Peter on that. Which is, I think any loss of privacy here is just due to this being an architectural atrocity. This wants to be built into an operating system like Mac OS. It wants to take advantage of the secure enclave. It wants to have secure hardware that's cryptographically guaranteeing that as it captures pixels that come out of the compositor and the window manager and the renderer that all of those are securely handled and kept local. The reason that this is one big privacy dumpster is because it's not being baked into the hardware. I agree. And local operators. But that can be fixed. And it will be fixed. And I want that. You know, I've often said, I'm going to give up everything, every piece of detail because I want my AI systems to be that much powerful. Selim, you're back with us talking to me about what do you think about this? I think this is AI agree that looks two other things. The one is that this is going to cause massive privacy issues for workers. I'm worried about a big rubber watching them. Already today, it was a crazy statistic that 44% of Gen Z workers are sabotaging AI's efforts to automate their own work. They're putting in the wrong data throwing off the AI training. It's really crazy what's happening right now in workplaces. So I think this will just exacerbate it and bring this whole conversation to their funds. I talk about a losing battle. You're far, far better getting on the wagon than you are trying to do that. That's such poisonous behavior. Protect your job. Welcome to the health section of moonshots brought to you by Fountain Life. You know, my mission is to help you use the latest technologies, including AI, to not just do your work at home, teach your kids, but to help you live a long and healthy life. I'm here today with an extraordinary physician, the chief medical officer of Fountain Life, Dr. Don Musei-Landon. Let's talk about cancer. You know, I know from the member database that we have at Fountain, our members who come in who think they're healthy, it turns out 3.3% of them have a cancer in their body they don't know about. That's right. You know, the majority of cancers that we screen for, those aren't the ones that are necessarily taking the lives when found at a late stage. We know that when cancer is found early, the chances for cure are much higher. We know it's much easier to treat a cancer when found early versus when found late. What we're finding in our members is over 3.3% we're found to have these cancers that were otherwise wouldn't have been found or detected. Yeah, you know, it's interesting. You don't feel the cancer until stage three or stage four. And if you don't know what's going on inside your body, it's like driving your car with your eyes closed. And you can know. And so when members come through found how do they detect cancers? So we're doing full body MRI and we also do early cancer detection screening. This is very, very important and these are not typical tools used in the conventional care setting when it comes to prevention. This is a hard thing because currently these are not studies that insurance would yet be covering, but the goal is to collect these numbers, do the research and work hard to democratize wellness. Yeah. And the day you can know what's going on inside your body, it's your obligation to know. So check out found life and go to foundlife.com/peter to get access to the latest technology to help you detect cancer at the very beginning at stage one when it is curable before it gets to stage three or stage four in your world of hurt. All right. Here's our next story. Basically world ID verification integration into zoom. And here it is. So in the backstory, I think that's important here. So in 2024, engineering firm called Arup, ARUP, lost 25 million after an employee in Hong Kong authorized a series of wire transfers during what appeared to be a routine video call with the company's CFO and several colleagues. The problem is that everyone on the call except the victim turned out to be an AI-generated deep fake. We've seen similar attacks in multinational firms in Singapore and the impact of this is huge. Right. So what we saw in 2019 to 2023 was $130 million in losses due to deep fakes, 2024, which is $400 million. 2025 last year, it was a billion. It's projected to reach $40 billion by 2027. And so step in our friend Sam with his device called the orb that takes a photo of the back of your retina and you verify on zoom that you're an actual human. Use his world ID and a real time face authentication from a selfie as well as video. And it says yes, yay, Verily. This person is a human. So and you get a verified human badge on your on your zoom link. Did you just say game or leave that's fantastic? We're right back in Shakespeare here. That's awesome. Yeah, Verily you're a human. You still have to go and actually scan your eyeball in one of these orbs. Is anybody got it? You guys done it yet? No, no. Apparently it's bouncing all over Africa. People are scanning away. But I haven't done it. But I love it because I don't know if I told you Peter. But I was on stage here at a company wide meeting and we took a little five minute break in the middle and our controller came up to me and said, Dave, I'm so sorry. I only got half of those wire transfers to China out. I'll get the other out right away. Seriously. What are you talking about? And so I got back on stage and I'm like, what are what she was talking about? And so then the whole second half of the company meeting in the back of my mind, I'm like, wait a minute. So I got off.
I got off, she said, "Okay, I got $300,000 out." And I'm like, "What are you doing?" And she's like, "Well, you told me it was an emergency "and we got to get the money to China right away." And I'm like, "Why would BBB be wiring money to China? "I don't understand." So anyway, only about 75,000 got across the border. We never got that back. And the rest of the FBI got into it right away. But I'm like, "Man, digital transfers like this, "everything should be logged anyway." I really feel like the digital fraud world is gonna get solved. And this is a big part of it. But everything should be logged all the time. It shouldn't be that hard to deal with digital stuff. I'm much more worried about chemical, biological, radiological stuff than I am about digital stuff. 'Cause I think we're gonna get it fixed and this is part of it. Alex, any thoughts here? This is minority report. This is the sci-fi future that we're catching up with. Apple with its face ID was focused on the face, not on the red nav. But if you remember the Tom Cruise, Stephen Spielberg minority report division, this is it. I think it's been interesting to watch as world evolved from world coin. And it's been interesting to watch as the company bounced back and forth between more crypto-focused and the economics of it versus the identification of human as a human side of it. But it seems from a distance, like the human identity verification side is ultimately the bigger seller than the crypto side. And to the extent that's the case, as resident crypto bear on first of order. - We will have that debate, Selim. Don't worry about it. - So there's a wild irony here that the more AI scales, the more valuable, verified human identity becomes. This is kind of interesting. - Yeah, sorry, I couldn't. - Yeah, no, so here's this next story that's related. So Grock creates a realistic AI French woman with a reflective ID. I'm gonna play this little video here and take a look at it very carefully as she holds her driver's license up to the camera. Look at the beautiful and real-ness looks. - So, to the world, I just received my new card, identity. Look, it's official and super well done. I'm so happy. So this was posted and it went viral by this gentleman, Dr. David Lutzke. He says, "This AI French woman was created by Grock complete with perfectly reflective ID a few more months and video ID verification may no longer be reliable." So, I mean, how many times have you taken, a picture of your license or your passport and uploaded it? It is gonna become more difficult. We're gonna have white hat, black hat, competitions up the wazoo here. - Alex, or-- - Well, I would maybe just comment the IDs themselves should be verifiable with a centralized database. That's how you can maintain a single source of truth and whether people are flashing IDs or not maybe-- - Unblocked chain letters. - Centralized database, not a blockchain, but I'm gonna put one on Peter. Good one. I'm just poking you, but I'm just poking you. - I also think, there are so many other technologies that we have to bring to bear. We can do hardware level cryptography, for example, chain of custody for video. It's not that as a civilization, we lack the technologies to ensure that any video or images actually originated from the real world without tampering. It's just that we lack the demand for it right now. And if I would predict that if ever the situation of deep faking gets so bad that it's causing real problems at a societal level, that'll just unlock all of these technological solutions including hardware level crypto for cameras, cryptography, not cryptocurrencies, that the market will speak for itself and will get all those tech. - I'm still waiting for the laws to come out that require all, you know, GROC and every other video generation to really identify it as AI generated. It's not a real work. - I covered in my newsletter, "Intermost Loop." There is a bill right now by partisan bill working its way through the house that will cover elements of deep fake finger printing like that. - Yeah. Yeah. All right, let's move ourselves along here. We're gonna talk about the economic impact of AI. There is a lot going on. Token maxing, word of the year. So this is from a report from 404media.co. Startup CEOs who are token maxing are bragging that they are spending more money on AI compute than it would cost to hire human workers. Astronomical AI bills are now in a certain corner of the tech world supposed to be the marker of growth and success. Look how much I'm spending on my tokens, everybody. You should invest in me so I can spend more on tokens. Dave. - Yeah. No, this is a warped story. This is a great thing. The way you get left behind is by not trying. That's the worst thing you can do right now is not get in the race, not play with AI, not try. And token maxing is fine. Like a CEO that's proud of the fact that they're consuming a ton of tokens, you can come and optimize it later in the year. But get every one of your people on their AI platform like now, like yesterday. And go ahead and start burning the tokens and then you'll have no trouble making it more efficient later if you get in the game now. So I think it's great when a startup CEO says I burned three million of venture money on compute. Fine, you're learning a ton along the way. And nobody incinerates money for very long. They're not that irrational. So this is just sort of the backlash story. - What was the gents in factor? It was like half your salary in tokens. - I'm saying, yeah, I'm saying you're full side. Like I'm telling everybody by end of year. So you have nine months. 50/50 is a good target. Half payroll, half token use. And then again, you're not gonna have any trouble optimizing it. The token use is effectively about a 10x force multiplier. So if you're one to one, it's like I've got one humans and 10 AI equivalents in my bucket of endeavor. So I'm actually under invested in tokens at that point relative to human salary. So one to one is a better target, I think. - So Liam, are you doing it? - Well, I think the bigger, the more healthy question is what's the ratio of tokens to reducing iterations and maximizing efficiency rather than just a raw spam? I think for now raw spam, it is fine. But that's kind of vanity metric, right? Your better off kind of looking at it is to what extent to the compressed iteration cycles that'll be where it will end up. - It's what Alex you said earlier. It's dollars per token economic reward. - Exactly, I can't get five per token. - I think that's a great problem. - If you ask a great, great salesperson, how many miles did you fly this year? - Sure. - Like it's a terrible metric of sales productivity, but if the answer is zero, it tells you it's a bad salesperson. I think it's great when a salesperson said, oh, I had a million mile year last year and they're proud of it. That's great. It's not the right metric, but it tells you they're proud of what they do. token maxing is a lot like that, I think. - Alex, close this out on this one. - Yeah, I've seen a variety of asset allocations in recent months between humans and AI's. I think tokens to humans is one interesting way of framing that a pessimist will look at this and say, this is replacement theory. This is humans being replaced by AI's how awful. An optimist will look at this and say, how incredible we're empowering fewer people to do more and achieving higher per capita productivity within an organization. What I don't hear very many people asking is, where does this end? So right now I see asset allocations, humans to AI's are at least human labor versus AI compute budgets, ranging from one to one, one to two at some of the frontier labs, it's an even more asymmetric ratio. Question in my mind is, is there any stationary end point? Is there a fixed point as this evolves? I tend to think it's going to trend towards one to infinity effectively that as we start to phase humans out of the service labor force, we're going to see all tokens and no humans. It has to. There's no other way around it. The capital is a little demand, but tell me, agree. - I told the humans merge with the tokens at least. (laughing) - Tokenpreneuron. All right, Selene, this was your story. So UA launches a gentick AI government models. This is from Sheikh Mohammed, the prime minister of UAE, the ruler of Dubai. He says the, he says UAE is launching a new government model within two years, 50% of government sectors. All sectors, all services, operations will be run on a gentick AI. The UAE will be the first government globally to operate at the scale of autonomy. Selene, brief us on this one. - Yeah, so I had to, I did a talk for his Highness before years ago and talk through where this is going. And you know, Minister Al-Alamah, the minister of AI hasn't given him a bottle of ice, paid him years, all his end-mind. And they are going full speed on this. I gotta give them massive credit. This is the benefits of the authority you can wield when you have a benevolent dictatorship. - Yeah, I think absolutely. - You could just get it done. - And they, when you have that, you have to make sure that the whoever's in charge is doing the right things for the country. And the ethos here is 100% alignment. And they are going,
at a massive speed on this. Just to give you an example, I was given a golden visa, right? And I was asked to be the test case. And the thing was could you get a golden visa authorized an issue within five hours? And they were freaking out going, you know, Singapore takes five days. And his highness said, okay, do it in five hours. And they were for them, but they got it done. And so there's an ability to cut through legacy thinking in a very powerful way. And this is such a massive, and a bit of advantage. We're actually working with a few of their folks in the prime minister's office on this. And so we're very, very excited about where this goes. >> There's another quote from Sheikh Mohammed. He says quote, "AI is no longer a tool. It analyzes, decides, executes, and improves in real time. It will become our executive partner in enhancing services, accelerating decisions, and raising efficiency." So, I mean, you can do this in an absolute monarchy. You can move this fast. I mean, what's shocking about this story is a speed at which it's moving, right? You can, there's no parliamentary approval, no public debate or consultation. And the question is, can Western democracies even keep up? I mean, you're gonna see this in probably Saudi, maybe in Singapore, other Middle Eastern nations. Can we see anything like this in the US? >> Actually, yes, you can. And I think we will. I tell the story of, it used to take six months to get approval for a wind turbine. And I think it was Colorado, one of the Western states. And then they finally just got together and mapped all the power lines and water mains and flight paths on a GIS, plotted on Google Maps, and made it available. And now it takes like 30 seconds to get approval. Because it knows where everything is. It doesn't need to take six months. And I think there's the economic impetus of this. This is the basis where I think AI can make the biggest and most incredible difference. Because in descriptive workflows, you can absolutely completely automate and almost all of government certainly implementation or policy enforcement is prescriptive workflows. We know exactly the steps to renew your drivers like this. We know exactly when it needs to take place. So there's no reason why that can't be handled automatically. >> With AI, step one. >> In a very short future. Step one, give a person a super frustrating experience. Step two, make them wait in line longer than they need to. Yes. Anyway, Dave, do you want to jump in on this story? >> Yeah, I don't think the US has ever copied a good idea back from another country since the American Revolution. We stole the British legal system. But since then, I don't think there's been anything but this is the opportunity. Well, I mean, look, you're exactly right. Monarchy can move very, very quickly. The rate at which things need to be regulated and new services need to be rolled out is way, way, way faster than any government in history has ever run before. So only AI is going to be able to do it. So if we get a great system together in the UAE, we're inevitably going to want to copy it back to the US. I think Peter asked the right question though, is the US ever going to, like the way Congress works, are we ever going to take a good idea and bring it back in? Yeah, I bet against that. But it's the right thing to do. We're the AI. We're weirdly on this one. I'm more optimal stick than you guys, which is weird. All right. Let's move on. We're going to have some fun here in the biomedical space. So there's a new wave of biomedical innovation that's coming. And I want this segment here to give people hope. We talk about longevity escape velocity on this pod. We talk about the health span revolution. Well, it's happening. I was with Demis last Saturday at the Breakthrough Awards talking to him. And he's absolutely convinced that we're going to cure cancer and solve all disease inside of the next five to 10 years, hopefully on the five-year side. So the first story here comes out of OpenAI. OpenAI releases chat GPT for clinicians. So it just gave away to all US clinicians. These are physicians, nurses, physician assistants, free AI co-pilot. And this co-pilot outperforms all human doctors. So they have a health bench, benchmark that they use. It scored 59 versus 43.7 for human clinicians. Pretty extraordinary. They validated this on 700,000 model responses. And they got a 99.6% accuracy using their physicians evaluating the AI versus human responses. And pretty extraordinary, something that will up level, I think, medicine nationwide. And for my standpoint, I've been saying this for a while. I think it's going to become malpractice to diagnose a patient without AI in the loop. There is so much going on that no human doctor can possibly understand it. At Fountain Life, we upload 200 gigabits of data about you and across your genome, full imaging, full microbiome, a tabloom, 140 blood biomarkers. Humans can't analyze all that, but AI's can. So, Jents, any thoughts on this? Alex, do you want to weigh in? Yeah, I'll chime in and say the professions are cooked. Yes, this was a widely expected release. This wasn't a surprise. Those of you watching early releases, leaks out of open AI saw this coming months in advance. You can even know from those leaks what the next one to drop is what the next profession, it's law. There's also one coming for management consulting and financial work. Open AI, thanks to GDPVAL, in some sense, mapped out all of the knowledge work verticals and is in a good position thanks to their own internal and now external benchmarking to know the relative strengths of their model as appropriately fine-tuned or post-trained for different verticals. So, I would expect to see many, many more of these chat GPT4X for different verticals. In the case of clinicians, thanks to open evidence and work by Epic and the form of up-to-date and other clinical AI's, this is already a somewhat crowded market that open AI is coming into. If I were open AI, I would release this sort of product more as a reference design and a way to ensure that capabilities that are built into the underlying models and then post-trained via a variety of evals are broadly available and that open AI maintains its status as a favored foundation model for clinical and biological work. Maybe they'll try to monetize this as best they can. Right now, it's available for free, but I tend to think it's worth more to open AI, more as a distribution channel for medical knowledge and one that they can build on. Open AI has released a variety of statistics over the past year for how many people are self-diagnosing or otherwise trying to treat themselves using chat GPT and I think offering a standard regulatory compliant channel for that is a very clever way to then do a sales up pitch to biomedical enterprise and life sciences in general, which is probably worth real money as. It's also a data aggregation strategy, right? I mean, open AI is going to be getting a huge amount of data far more verified than I feel this way or I think I might have this, bringing in a million plus clinicians into the loop. The other thing, yes, that's worth saying here is that at least current estimates are that we're gonna have 86,000, shortage of 86,000 physicians in the next 10 years. But it's gonna be interesting, right? You, you know, I have two nieces that have gone through medical school, my sister and myself, you know, lots of friends and you're spending literally between college, medical school and postgraduate training and whatever field you're going into, you're spending well over a decade and half a million, close to a million dollars to get this degree. And will you even need it? Is a medical doctor gonna need to be in the loop or is it a nurse plus an AI that's gonna be giving us all our medical advice or diagnostics and our therapeutics with a optimist robot giving your surgery? There's a lot of change coming here. - Yeah, a huge amount of change and also it'll be a great case study and like we're not about replacing doctors here or about detecting thousands of things that were not previously detected and cutting them off early and extending longevity and making life better. And you know, it's not a given to me at all that the number of doctors goes down just the number of things we wanna do goes up 100 or a thousand times. - But are you gonna spend that much money to go through medical school and get this little profession when the AI is doing the diagnosing again today? - Of course this is about replacing doctors. I mean, let's call us spade a spade. Of course, when fully developed, this and comparable solutions are about automating away medical practice. How could they not be? And also by the way, nursing and also by the way, the HMOs and drug design, open AI and other frontier labs are all pursuing drug design and drug delivery. Of course, it's about the full picture of, if you're gonna solve medicine or you're just going to leave millions of human doctors practicing as sort of meat puppets for the AI, no. This is going to be the end to end solution. We're just seeing the beginning of it. - I agree. A couple comments here. One is, you know, in an ideal world that doctors getting a cognitive exoskeleton with all of this, right? You get this amazing capability to expand your own intruders thinking. But Alex is completely right. But on the other hand, this you're going to get a huge backlash here. This is a very,
regulated industry, remember a few years ago, Texas passed a law banning telemedicine, okay? Just outright banning it because you know, for every spot on my hand, I must have to go to a physical doctor, I can never do that over video. So the immune system response is going to be very, very fierce. I expect to see this battle play out heavily over the next few years because this vested interest up the Yin Yang and healthcare has the third worst immune system ever, behind religion and educational activity. Yeah, and then also sure that the immune response, if you look at what happened with the broad transition to electronic medical records, like epic based systems, for example, every clinician that you speak with will complain about epic, they'll complain about EMRs, how much EMRs distract from direct interaction with the patient, all of that, and yet every major medical system is either completed or is in the late stages of at least in this country, their EMR transition. If they can't resist EMRs, if they can't resist EMRs, how are they going to resist strong AI that outperforms humans? Wait, no, no, no, no, no, no, EMRs are kind of as add on helpful aid because it saves you in documenting the process, etc. This is really a poor, this is the clinicians hate the EMRs, they hate the interface, they hate the process. Of course, but they're going to hate this 10 times more because it's a direct replacement for the cognitive ability that they've trained for 10 years to do. So just, my prediction is huge regulatory and immune system backlash on this one. My prediction, AI labs have been using healthcare as the reason why they can't slow down, as well as the fight with China, right? If we slow this down, we're going to lose lives, it has been sort of the heraldine call. Totally agree. All of everything I said earlier about restitution needs to be a wholesale replacement of the medical system is absolutely correct. But the past, my prediction is delivered with stones and speed bumps. Or the record, and this is sort of interesting. We want to sort of have that interesting. Go ahead, finish up Alex, you're good. This is an interesting micro debate for the record, my intuition, and I interact with a lot of clinicians is the exact opposite, the clinicians hate the EMRs, but they love the AI that helps them do a better job of what they want to do. And there may be an extent to which AI interfaces like this end up being framed as the solution to all of their EMR woes. Until it takes their job. Well, of course, that's the way this works. Let's move this along here. Our second story here is AI to reduce wasted donor hearts. And I love, you know, I just want to show a number of stories here how AI is going to be interfacing and changing the medical practice. So I don't know if you guys are an organ donor, I am anybody else. So currently there's 4,000 patients who need a cardiac transplant today. There's 103,000 who need some type of a transplant, kidney, liver, lung. And when an organ donor is on the table end of life and the physician has to analyze the organs and decide whether they're viable for transplant, you know, you've got like 15 minutes, typically at 2 o'clock in the morning to make that decision. And so in the heart world, only a third of the hearts are ever actually chosen for transplantation. So here comes something called top heart. Yeah, just a third, make it out the door. So here comes something called top heart from NYU in Stanford. And top heart is able to look at 20 different variables, right? Typically the physician is looking at how old is this person? Do they have a drug history if they know? And looking at coronary artery disease to say, should we ship this off? Their goal by looking at 20 different variables is give that surgeon at 2 a.m. in the morning a second opinion. And they believe that they can get an additional 500 hearts into the organ replacement ecosystem. This is on top of the fact that there's an entire sort of synthetic biology world going on right now to provide an abundance of organs from bio printing and Xeno transplantation, you know, pig organs, you know, the anthogens being replaced by human antigens. This is the work of George Church at egenesis and Martin Rothblatt at United Therapeutics. So you know, this is an abundance story of going from a limited number of organs to an abundant number of organs. Alex, you tracking this as well? I'm tracking the space broadly. There are other advances as well, like trying to create a national market for organ donation versus a bunch of state markets that would be greatly enhanced with improvements in vitrification and cryopreservation. I think it's good that there is a vibrant and growing distribution channel for donor hearts. I think that's great. But I also think it's very painful that the need for one human to die, or at least that one human dies and donates a heart to another human, that's such a zero-sum type situation. It's painful to think about. And while it's great on margin to have more efficient ways of distributing donated organs, I really, really would like us to get as soon as possible to a situation where donor organs are completely unnecessary. And we will. I think egenesis, Dean Cayman's company, Advanced Organ Generation, they go from your skin cell to a pleuripotent stem cell to regrowing your heart-liberal and kidney. A lot of this is going to be up and operating by the end of this decade. Hopefully sooner. Can't come soon enough. And of course, as we have autonomous cars having less car accidents, the ability to have organ donors is going to begin reducing though. Still motorcycle accidents are probably the number one reason we get organs donated. Let's move on to our next story. And this goes in line with the fact that we are beginning or at the beginning of the slaying of cancer. So this is a great story, pancreatic cancer mRNA vaccines show lasting results in trials. So I don't know if people have been tracking this, but we now have these cancer of vaccines. And this is using mRNA. We used it as a COVID vaccine. This is actually the ability to create an mRNA that activates your immune system against the cancer that you have. So there are more than 120 of these trials going on against lung breast, prostate, melanoma, pancreatic, and brain cancer. In this particular case, a five-year survival rate for pancreatic cancer has just gone through the roof. Historically, it's 13%. If you have pancreatic cancer, it's a death sentence. Only 13% of people are able to survive that. So in this report, eight out of 16 patients who generate a strong immune response to the vaccine, that's 87.5% still alive after six years. So how does this work? You have a surgery to remove as much of the tumors you can. You sample the tumor. It's sequenced. And then that sequence is identifying 20 unique mutations in your cancer. That is then built into a personalized mRNA that activates your immune system like killer missiles. It activates your killer T cells to go after and attack your cancer. So this is a breakthrough in how we deal with cancer. And the fact that you're durable after six years is pretty extraordinary. I remember this incredible quote from Raymond McColley, our biotech guy at Cimiardi, said, mRNA vaccines are the first battle in the last war against disease. Yeah, amazing for me. And I think this is showing my daughter works on that. My daughter works on this over at Moderna, actually, mRNA vaccines. And it's a journey. Yeah, I mean, Moderna got a bad rep on their mRNA for COVID, but this is the Holy Grail, right? I mean, being able to go from your cancer to here's the injection that's going to save your life is extraordinary. Well, and if it were, I thought that I'd have a pretty good rep. That's the amazing thing. It's a universal solution. Like, when Alex talks about all of math is cooked, this is the difference between in the old days, I solved one math problem. Now I have an AI. It solves all math. This is the equivalent in biology where if it works, it should work everywhere. Yeah, Alex, you're right. I mean, mRNA was a, was an, you know, project warp speed. I'm just saying afterwards a lot of people are coming down on mRNA vaccines, but there's a lot of politicized griping over mRNA vaccines in general, but there's going to be political griping over almost anything at any scale. I do think I think back quarter of a century to Eric Drexler and engines and creation and the National Nanotechnology Initiative when the US Congress was, was sold to story that with billions of dollars of congressional and national investment that we would get medical nanorobots that would swim through our bloodstreams and do weight-answer cells. Well, we're getting it though, but we're not getting it with dimandoid nanorobots. We're getting it with these lipid nanoparticles and Moderna and Pfizer-style mRNA vaccines. I think it's interesting to almost as a retrospective to say we actually got the nanorobots, they're just fat. They're not, they're not dimandoid, they're fat. Yeah, we're using our own machinery to do the battle for us. That's the other angle. Do you have a prediction, Peter, given that immunotherapies in some sense, like really, really costly immunotherapies at the moment?
We've known about some form of immunotherapy for 100 plus years, and people who were infected with a virus 100 years ago, in some cases, or bacterial infection, showed tumors shrinking. We've known at some level that some form of immunotherapy would work, and we're only now figuring out how to fully weaponize it and operationalize it. Where do you think this goes? You think, like in 10 years, we're all wearing Apple smart watches that are looking for evidence of tumor DNA or RNA in our bloodstream and then send our daily mRNA update to a programmable implant or something? I think that is basically it. Either they're implantables or you'll be sampled on a regular basis. I mean, the goal, of course, is find it at the very beginning, especially if there are solutions. There's one more point about this that I think is really powerful. This is personalized medicine is actually becoming operational, and that's a huge inflection point we've been waiting for a long time. Here's another example. Again, just to give people hope and to see the data, longevity mindset is about seeing this over and over and over again, saying, yeah, the world is changing. The things that used to kill us are being either solved or delayed. The single shot CAR-T infusion shows strong response from melanoma. It's not just a strong response. 100% cancer-free after a single shot. This was an unexpected result. Within two months of treatment, all 20 patients in this trial had minimally resugial disease, MRD negative, that no disease identified after they were assayed. Again, meaning that all patients had a median follow-up of 15.3 months without any show-up of their melanoma. It's game-changing and timing. You draw blood. You identify you have melanoma. The doctor finds it. We should all be scanning ourselves all the time. We do this at found using visual and a minimum. If you have a family history of skin cancer, please have yourself checked in a regular basis. So the doctor draws blood extracts your T cells from the patient. Genetically, engineers the T cells, right? A gene is inserted, giving those T cells, and you receptor, called a CAR, a chimeric antigen receptor, that is specifically programmed to recognize the protein from your melanoma. Your T cells are then re-injected back into your body, hundreds of millions of them, and they go identify the melanoma and they slay it. For the first time ever, this type of a therapy were using the term "cure" on this particular type of, I mean, it's extraordinary. So just another example of what's coming. This is both amazing, but can you also see the clumsiness of it requiring blood extraction and then CAR T cell creation in vitro? Why can't we do this in vivo? Why can't we do this in individual cells even? We're seeing the beginnings. This is almost like a horse and buggy era of immunotherapies, but surely we should be able to do this in a fully autonomous, like, intracellular environment. Take the win, Alex. Take the win. Oh my god. Yeah, I want my FSD. Yes. And you shall have it. All right. Here's one more story, and this is a fun one. So, you know, MRSA, MRSA, people probably heard about this. It's methamcylin resistant, staphyl-orius, staphyl-caucas-orius. It's a killer infection, right? This has been typically in hospitals. It's now getting out to the community. So, 2.8 million people have MRSA infection every year. It kills 35,000 people in the US alone. The problem is all the first line antibiotics for MRSA have failed methamcylin, penicillin, moxacillin. And now even vancumisin, which has been the antibiotic of last resort is no longer working. So, this particular drug, Candaceartan is now being used. It's a FDA-approved BP medication for blood pressure. And it works to basically stop and inhibit MRSA infection. And so, this is an example of taking the existing drug. And it's now fully usable by the scientific and medical community because it's been approved. We know it's safety protocol. So, I love this. Do you remember Selim on stage? We had the abundance. We had David Faganbaum from. Yeah. So, this is a similar to history. I just tell a story and just congratulate him, a donor to his foundation. So, here's the story here. So, in 2010, he's a 25-year-old medical student. He comes down with this rare disease, where disease goes, Castleman's disease. And they throw everything they can at him. And he's literally read his last rights. He has four near-death experiences. And then, as a medical student, he starts experimenting on himself. And he discovers that his disease is caused by a hyperactivation of the M-tore pathway. And he says, "Well, if it's the M-tore pathway, I can probably down-regulated using rapamycin." And he does that and he finds out that it works. So, he's been remission-free from 12 years. And he comes up with the idea, "Are there other diseases out there for which an existing approved drug can be used to cure the disease?" And here are the numbers. There are 18,000 recognized diseases out there, but only 4,000 FDA approved drugs. And so, he's now using AI to match the existing drugs and repurposing them against new diseases. And it's working. I think that's such a great example of citizen science, also, right? Take a personal problem and then just start acting already through it. I think we're going to see hundreds and thousands of this example. And this is where people should pick and understand why you're so excited about technologies, because this is now possible. And this is not possible 10 years ago, five years ago. And now it's just going to become more rampant. And any problem can now be solved by just focusing on attacking the AI and going after us. And it's incredible. >>Solved everything, right? >>Yes, self-everything. And also, I would say, historically, before this era, off-target indications were dirty words or dirty drugs that have lots of off-target side effects, highly undesirable. But now, if we have amazing AI models of individual cells and the body suddenly off-target side effects, they become a secret weapon. And we can repurpose drugs. We can combine repurpose drugs. I'm very bullish on this space. I advise I have a portfolio company, Senjam Therapeutics, that is focused on increasingly on AI for repurposing medications for anti-inflammatories for other purposes. I think this space has enormous potential thanks to AI. >>Amazing. For folks interested, you go to everycure.org. You can see what David's doing. It's a nonprofit and support his work. He's brilliant. All right. Let's get into some fun conversations here. The robots are indeed coming. A few stories to report here today. The first is the ping pong champion of the world is now an AI-driven robot. Let's take a look at this, this a little bit of a match here and we can discuss it. The background music is killing me. Sorry about that. The robot's using nine cameras and three vision systems. It won three out of five games. I don't think pause is here. I won three out of five games. I'm surprised when it win all five games. And of course it will. >>Doesn't have a lot of top spin actually. It's just very nimble. >>No, this is the worst it's ever going to be. That's kind of incredible. The speed of responses. >>Amazing. This robot is called ACE. I'm not sure if I would see this in the same lineage as deep blue or alpha-go, but it's the beginning. >>It's totally not. This is a much lower-dimensional game than any of those board games. It frankly is astounding to me that it took this long to reach human performance in table tennis because it's such a simple game. You only have a handful of degrees of freedom in the ball. You have the position. You have its linear momentum. You have its angular momentum. And I think that's about it. The rest is just modeling the trajectory and maybe doing a little bit of Monte Carlo research for tactics that your opponent might take. This should have been solved years ago. I don't know why this took so long. >>Let's answer that question actually because that's really well said. And this is very similar to many, many robotic operations in your home in a factory. Whatever the barrier was, I think it's probably related to the vision system. It's not a high margin problem. It's really investing a billion dollars to solve it. But now because the vision systems and the feedback systems are dirt cheap and easy, but it was solved by one or two people in a few weeks. That means all these other home robots can now be built by one or two people in a few weeks. >>Similarly, there's a tennis game where I'm also excited to play with. It should be really cool.
>> It's actually the same category. >> It's a total no-brainer. If you ever use a ball machine, then you go pick up all the balls for like 20 minutes with the final. A robot that does that is literally MIT class 270 could have done it with what's the barrier. And I'm sure the barrier is related just to the feedback control and the vision, which you can now just use with a transformer. >> Well, also people that are in robot labs don't play tennis, so they don't have an incentive to go do nothing for the other things. >> They don't have to do it. >> I don't want to go too far down this rabbit hole, but there's a massive correlation between successful founding entrepreneurs and the MIT tennis team. It's basically 100%. It's crazy, including Warren. I like that anyway. >> All right. The Tesla CyberCab is now in production. Take a quick look at this video here. So Dave, you and I saw this and we saw the production line. We were at in Austin in December and December here. Of course, no controls, no steering wheel, no pedals, and operating cost of 20 cents per mile. And Elon's announced he's going to sell it for $30,000. I think an incredible investment, if you can afford it, is you buy ten of these and you put them out in your community and it earns money for you while you sleep. >> Why is this your-- >> Just listening to this podcast and you're not watching the video, go find this video clip. You got to see the interior of this to believe it. It's like you're walking into a car, but it's just a love seat. >> And it's only a two-seater, which is the average load for an Uber. It's like 1.2 people per Uber. So two seats makes total sense. >> Where? Where? >> Look, if you have four people, just push the button twice and two of them come. >> When is this expected by the way? I need this to get my kid to school so I also do that. >> So production off the line, we saw them. >> Production officially started this past week, April 24th. And the challenge is can they really build at the rate that they want? They want 2 million of these per year or go. Now, I mean-- >> Are they regulatory hurdles? So that's been passed now with the number one. >> No, it's the hour of us. >> It's same as Waymo. >> Okay. >> It's town by town. Stay by state town by town, but if you're covered, yeah, you just can't. >> I mean, the difference is a Waymo because of the LiDAR and all the camera systems and just the base. I think the vehicle probably, you know, it tops out over $100,000, probably $150,000. I'm not sure if they get into higher production if it's going to be coming down. But at 30K, this is insane. >> Yeah, no, there's so many parts. If you look at the parts, just laid out, you know, because there was that great exploded car in the showroom. >> For an ice versus electric. >> In compare it? Yeah, compared to a consumer gas-powered car and just in raw part count. And it's just-- it's got to be 80%, 90% reduction in components versus a gas. >> I'll give you the statistic. I always happen my head. The combustion, you know, the number of parts in the drivetrain and combustion engine car about 2000, a Tesla has 17 moving parts in the drivetrain. >> It's just the future of transportation is so good. >> It's just better than the car. >> Yeah, and it doesn't need a huge battery range either. You can just go and hang out and recharge itself whenever it wants. Another one will come. >> And guess what? On the transportation technology line, here is the next story. Here is Joe B. Aviation. This is Joe Ben who started velocity 11 with Rob Nail, if you remember Rob. >> Sure. >> So Joe B just did his first air taxi flight from New York to JFK. Let's take a listen to this news report out of New York. >> And hello, I live in New York. Hello. >> I know. Well, this is going to help you out, buddy. Check this out. >> We'll master getting to New York airports as a nightmare. >> Electric air taxi demonstration took off from Kennedy Airport. >> It's been a long time since the last time I had a helicopter. I had a helicopter that was on the way to New York. And if things pan out, the company hopes to have its fleet up and there and running within the next year. But for now, for the next week, you will see this aircraft that kind of looks like a large drone buzzing over our area. It's a time machine, gentlemen. I'm standing at the helip well with my bags ready. I love it. You know, EV calls it after no. That's EV tells us the name though. I just call these things flying cars for lack of a better term. We need a better term than flying cars a better term than EV tools. It took too long. We're supposed to have these by 2015 and back to the future part to here. We are in 2026. Why didn't it take so long? We need to we need Mr. Fusion. You think Mr. Fusion is the reason we didn't get our flying cars. Absolutely. That's what the movie's about. Well, in our in our robotic segments here, we had two back to back Alex. Why did this take so long questions? So let's stick it on there very much. And why did everything take so long? I guess. Hello, I'm a commentary. You think it's regulatory. I mean, regulations are why we didn't get the technology's been there for quite a while. This is a lot. I'm going to ask Peter how long it took them for the friends. This is 11 years to get approval to do something that NASA had been doing for 20 years. Anyway, yes, the FAA is not happy till you're not happy. That's the rule. Well, I think we got to answer that question because you know, a lot of the AMA questions are around what are the jobs of the future going to be. If white collar gets obliterated, but I think a lot of the answer lies in these last couple segments, you know, robotic stuff is going to be abundant imminently, but it doesn't just naturally happen. So if we can answer Alex's two questions on what are the bottlenecks? Those are jobs. Whatever those bottlenecks are, those are your jobs. There are AI models. If there's a bottleneck, there are the AI will solve it. The episode is brought to you by Blitzie, autonomous software development with infinite code context. Blitzie uses thousands of specialized AI agents that think for hours to understand and to price scale code bases with millions of lines of code. Engineers start every development sprint with the Blitzie platform, bringing in their development requirements. The Blitzie platform provides a plan, then generates and pre-compiles code for each task. Blitzie delivers 80% or more of the development work autonomously, while providing a guide for the final 20% of human development work required to complete the sprint. Enterprises are achieving a 5x engineering velocity increase when incorporating Blitzie as their pre-IDE development tool, pairing it with their coding co-pilot of choice to bring an AI-nated SDLC into the world. Ready to 5x your engineering velocity? Visit Blitzie.com to schedule a demo and start building with Blitzie today. Alright, let's jump into AMA with the mates. So guys, thank you again for all the comments that you give us on the YouTube. We read them all. I have skipping read them all as well and summarize. We pick out eight questions that we can answer every week. So please keep them coming. And let's go to those questions. Alright, so gentlemen, pick your favorite question off list number one. Selim, do you want to go first? I'll go with number four, as I know that world a little bit, which is what's the future of large consulting firms like Accenture or Cap Gemini? This is from @sthebottle1501. This, it goes full on into the transformational, the effort that's going to enterprises here. Traditional consulting is a very big trouble if it remains a pyramid of junior labor producing analysis and that AI totally destroys that model. But consulting firms, you know, in the land of the blind, the one-eyed men's king in a volatile world, your clients are slower than you are and they need help. The model will have to change the future of consulting won't be like a people pyramid. It's an intelligence platform plus domain expertise plus change management. And we've been holding the change band against that for a while. The winners are going to be becoming bringing agente workflows and benchmarks and governance and implementation capacity to their clients. The loser is, I mean, just going to keep selling headcount for from any exo perspective, consulting them from experts to rent to transform transformation operating system. And the companies that help their clients do that will win. Yeah. You know, it's, we've talked about this before that in your scarcity model, you put a wall around all of your experts inside and you measure the balance of the hour. Right. And that is going to get collapsed. Alex, why don't you go next? I'll take question number two, which asks, everyone can be an entrepreneur with AI as a tool. However, what action do you take when you genuinely don't have a creative idea for a direction for most the answer is none. And this is from three billionth random user.
I don't agree with the premise. I think, and one of the reasons why one of my funds owed to 1D Capital, backed a firm started by friend of the pod Alex Finn called Henry Intelligent Machines or him, is to solve the problem of creative ideation for starting new ventures. I think just as AI can take over as an operator of a business or a fleet of businesses, AI can also automate the process of creative ideation for those businesses. And I think in that world, in the him world, if you will, the role of the human, sort of a one person, owner, or magnate, overseeing a conglomerate of maybe hundreds or thousands of AI-run micro businesses, the role of that human entrepreneur then becomes one of a taste maker. You have opinions. Everyone has opinions as a consumer of goods and services. But those opinions can shape the taste over fleets of AI's that are providing the creative ideation for businesses that they bring to you. They say, hey, I want to start this micro business for you. You like it? Yes, no. And then the human can have an opinion. The AI is performing the ideation. The human and the generation part. The human provides sort of the discipline and the taste and the discrimination for which ideas pass the filter, which ones don't. And that's the solution. That's how we square the circle of humans, not actually in extremists needing to generate all the creative ideas themselves. Yeah, I agree. Idea generation has never been the limiting factor. You just have to get around different people or just notice the problems around you. It's historically been execution that's been the issue. Go check out pulsia. I think it's pulsia.ai, which is AI slot backwards. If you sign up for that, it will scan all of your background. And it will generate ideas for you. And in fact, it will generate a website of a business based upon what your passions and interests are. Anyway, fascinating stuff. And I'll say, maybe instead of pulsia, I'll talk my book here since I have a financial interest in this one. Check out meat henry.ai. OK. Fantastic. Dave, number one or three. So I will take three and leave you with a hard one. Have it out. If you eliminate entry level jobs, but keep experienced jobs, what happens when the experienced people retire? Isn't that like eliminating babies from humanity? Says Todd Marshall 416. I don't think it's quite that dire, Todd. Eliminating babies from humanity about the worst thing could possibly happen. If you eliminate entry level jobs, well, look, this was going to happen anyway. If you think about-- actually, we have a weekend place up in Vermont. And there's the Simon Pierce glass blowing factory is up there. And if you want a glow glass, you have to apprentice with a senior dude for like a decade. And then they let you make glass. It's like a page out of 200-year-old history. That mode of operation is going to go away in all forms of white collar work no matter what. So the rate of change of the world and the singularity is so fast that the entry level career path was kind of a dead end anyway. So now Meta announced a 10% layoff, which is really going to be more like 30% according to the insiders, I know. And they're definitely not hiring new entry level people in the middle of doing the layoff, because AI can do all the coding. That was not the career path you wanted in the first place. So we're going to have to find a new way forward. But I think AI is going to be the ultimate teacher. We're going to save a ton of time on-- like Peter was saying earlier in the podcast-- the four years of medical school followed by four years of fellowship and internship. Eight years of your life after you're already done with undergrad is just way too much time. So it's all going to move to AI-based nimble training. And then this massively expanding economy creates huge amounts of new opportunity every day. But it's opportunity that didn't exist the prior day. So the entry level job wasn't really likely to lead you on that path anyway. So it's all got to get refactored. It's nothing like people stopping having babies. I think it's so well put, Dave. Who's a well put? All right. Question number one I'm left with is from @gnluca. Patjani. Patjani, 808. Who asks, you guys say AI will create jobs, but for whom? It looks like AI is creating jobs for AI, not for people. So @gnluca, the fact of the matter is in the long run, yes, AI will be able to do any job. I think that is the case. But people still like working with people. People still like hanging out with people. And I think it's ultimately going to be the fact that two things are occurring. Number one, as every technology destroys a layer of jobs, new jobs are created on top of that. Internet kill travel agents, but it spawned millions of social media managers, app developers, YouTubers, and everything else. So there are going to be new layers of jobs coming out. And yes, those may well be displaced by AI again. At the end of the day, the question is what are you passionate about and how do you use AI to help deliver that? There's going to be a human interface layer for a lot of things. Because people like hanging out and interfacing with people, us meat puppets. So it's going to be navigated. It's going to be important. And I'll just remind you one other thing. Ideal but job is a recent creation. And most people don't love the jobs that they have. They have the jobs they have right now, because they frankly need to put food in the table and get insurance for their families. So if you could do anything, what would it be? Would it be to work? I mean, in a future of universal high income, where everything is demonetized at such a point where you don't have to work, then you start doing the things that you love. So that's my take on it. All right. Let's move on to our second set of questions. Alex, why don't you go first? Well, let's go with question number five. Wasn't all of this originally predicted by Ray Kurzweil to be happening sometime around 2040? Are we genuinely that far ahead of schedule? And this is from Brett Avalon. I'm not sure Brett, what all of this you're referring to may mean. But I do think broadly we're well ahead of where friend of the pod Ray thought we'd be. I think we achieved, as I mentioned on numerous occasions, I think we achieved a AI, which isn't raised concept, but was popularized by Nick Bostrom and co-conceived by Ben Gertzoll and some others. I think we achieved that by no later than summer of 2020. And Ray's proximate-- Ray may say I'm misconstruing his timelines-- was predicting his version of AGI by 2029. So call that a nine-year gap. Ray-- and I've discussed this with him on the pod-- is predicting the singularity-- his version of the singularity-- by 2045. My version of the singularity isn't a point in time. It's now. And it's certainly not in 2045. It's now, and it's an interval. And we're right in the middle of it. So are we genuinely far ahead of Ray's schedule? I think we are. I think Ray would probably at this point and has arguably said that we are in some ways ahead of his schedule. And I think the benchmarks reflect that. And I think the 2045 timeline that he provided, where the superintelligence would be collectively smarter than all of humanity. I think we're going to hit that so far ahead of 2045. We'll ask him next week. We'll be with him in six days from the horse's mouth. Yes. All right, Dave, want you to next? I don't think the hardest one on this one. Number eight, P-DOOM, probability of universal destruction of all humanity estimates. Muskenhinton, say, 10 to 20%. Amadez has 25%. Oldman says non-zero. He actually said more like 10% when I interviewed him. How can any of these CEOs think it's acceptable to have a 1/5 chance of human extinction? They all agree with you that it's completely unacceptable. And they all say stopping research and letting China run forward isn't going to solve the problem. And so they each individually trust themselves. You can debate whether that's good or bad, but they do. And that's why they want to not lose the race individually. And that's why they're pushing forward at full speed. I think Musk and I think along the way Amadez have both suggested a six-month pause, but it wouldn't work. At the same time, they say it. They say it'll never work. It won't happen in the real world. So I'm just going to keep moving as fast as I can. But they 100% agree with you. This is completely unacceptable, ridiculous, and the lack of government involvement across the world is utterly insane. So that doesn't solve it in any way. It's just that is what's actually happening. And that's what's going to continue to happen. And I'm continually shocked as is Alex. I know with our inability to get any kind of government reaction to the-- what's now the-- we were telling him a year ago when maybe it wasn't 100% obvious. But now it's 100% obvious, yet still so slow.
So anyway, there's your answer. Do you think I would have hit answer? I'd be curious, do you think that they believe their own estimates here or is this a case of revealed preference where they think maybe it's more socially acceptable to estimate a higher number, but actually through their actions, they're revealing a preference that suggests their internal estimate is much lower? I think it's lower. I don't know if much lower. I think they all have the same chemical, biological, radiological terrorism as the number one risk. So I think it's probably lower, but I don't think it's like 0.001 percent low. Interesting. Celine, you have two to choose from. I will take number seven, which is when white-collar jobs are erased, where does the consumer demand comes from to buy from all these new entrepreneurial ventures? This is from the-- He told me, you know, this is a tough one, right? This is the central political economy question of AI. If productivity explodes and but income does not slow the people, demand collapses and the system becomes unstable. Capitalism needs customers, right? So we need new distribution mechanisms, we need lower costs, we need new ownership models, we need AI dividends, we need equity participation, we need sovereign funds, all of this points in this system, we're with the previous question, where, you know, on an optimistic side, AI makes good services cheap while giving individual more leverage to creating income. That's the good side. The pessimism side is that you have extreme concentration and then you have massive collapse of the economy. So path we take is a governance and an institution designed choice, not a law of nature. So governance and our institutions need the fricking weight up and smell the roses here. We have to rethink this whole thing with the social contract, which is what we're basically talking about, was essentially being wiped out. We can be optimistic about it, but the pessimistic case is very-- has a very big downside here. All right, the final question in our AMA today comes from @JamesWilliamsCU2QQ. How can a UCS engineer get experienced to become a lead AI engineer? If you can't get a job in the first place. James, first of all, you know, as we've said many times, getting a job is the old model, you know, the old model of dual high school, getting good college, getting diploma, getting a higher as a junior person and working way up the chain, that is vaporized or at least being fully vaporized right now. The option right now is build yourself outside the job, build in public, right? Basically go and find something that you're passionate about. It's based on your massive transformative purpose on what you care about. We're going to be launching an ex-price in this area very shortly and use the tools of LB today to build and ship. And you know, your GitHub is now your resume. Companies are increasingly hiring if you want to get a job versus start a company yourself. They're increasingly hiring based upon what you've done. I remember Elon said, "I don't care if you have a college degree." You know, "I care about what you've done." You know, that is your degree now. That is your resume. Show me that you're brilliant at what you build, not what you happen to learn in some college or graduate degree or entry-level job. So build in public, the barrier to entry has never been lower if you'd build something extraordinary that shows your capabilities. And once you do that, you're probably unlikely to be going after a job. You're probably going to want to partner with a couple of friends and build a product, a company, a service yourself. So that's my answer. That's my answer. I'm sticking to it. I'm just a weird advising. A couple of you university around this period and one of them is an engineering university and let us know what is an engineering degree. And it's pretty clear that the engineering degree or the future will be go build some stuff at the end. What did you build? And you get degree granted on what you learned, but what did you build? Yeah. I love that. And if you haven't done anything to start yet, other than listening to the podcast, add Alex's in our most loop to your daily regimen first thing in the morning. And that alone will inspire you to shift gears and get into this. Oh, thank you, Dave. It's very sweet. Yeah, for those who want to read the intermost loop, just go to Alexwg.org and I provide links to a substack and X and Spotify, etc. But I appreciate the promo, Dave. It's very kind. All right. Our outro music today, which is beautiful, is from Hitham said. It's AI Topia. All right. Come and get ready for some beautiful video and audio. [Music] All right. Thank you to my brilliant moonshot mates, AWG. I wish you a beautiful week, Dave and Salim. I can't wait to see you guys next Monday. We're all together again. We're going to be physically at MIT, at the book launch of We Are As Gods. We're going to be recording a podcast episode there. Can't wait to do it face to face. And check out the fourth video with us. Say that, May the fourth be with us, yes, for sure. As a stark track fan, I'm not allowed to say that. By the way, check out what's right above me is the world's vision camera identity camera right? I've got so many. Literally right over my head. No, it's just an omnicamp. It's just a real lens camera, but I couldn't resist it. And by the way, it was not easy standing in Vodala Airport holding a laptop at the time. I got my exercises for the day. I'm all made. I started moving around. You're trying to avoid like a policeman or something or why? I just have to shift positions down there and shift hands and once more lean on something. I know where to sit here, it's easy. And I don't want to risk losing a connection that I thought so hard to get. Oh my God. Okay. If you've got an outro song or intro song, please send it to us.
[email protected]. We'd love to hear it, see it and potentially play it. And thank you for subscribing to this and thank you to all of the fans out there. I know all four of us run into you on the street at the airport set events and. Amazing. It's really great. It's really great. Yeah. If you see us, do come up and say hi. Yeah, for sure. Hold on. All right guys. All right. Take care. Bye. If you made it to the end of this episode, which you obviously did, I consider you a moonshot made every week. My moonshot mates and I spend a lot of energy and time to really deliver you the news that matters. If you're subscriber, thank you. If you're not a subscriber yet, please consider subscribing so you get the news as it comes out. I also want to invite you to join me on my weekly newsletter called Metatrends. I have a research team. You may not know this, but we spend the entire week looking at the Metatrends that are impacting your family, your company, your industry, your nation. And I put this into a two minute read every week. If you'd like to get access to the Metatrends newsletter every week, go to dmandis.com/metatrends. That's dmandis.com/metatrends. Thank you again for joining us today. It's a blast for us to put this together every week.
PodcastTranscript AI