This was a big week in AI. I think we've had sort of convergent AI announcements happening. Every day is a miracle, and that's not overstating the case. Google has spent 20 years making commitments to an ads ecosystem. Very difficult for them to break those commitments, and I think it's a real threat to their search monopoly. The cooler this is, and the more search moves over to it, the more it cannibalizes the core. Already my kids are telling me, 'Dad, everybody knows that ChatGPT is better than Google.' 2029 to 2031, the end of white collar work. That's only a couple years for people to remap their entire career path. We're going from a world where our brains were wired for fear and scarcity to a world that's very different. Technology is a substrate to make us more abundant and happier across every aspect of our lives.
Now that's a moonshot. Ladies and gentlemen, everybody, welcome to Moonshots and our weekly episode of WTF Just Happened in Tech. I'm here with my Moonshot mates, Dave Blondon and Seem Ismael. See, by the way, happy birthday this past Saturday. Oh, happy birthday, Sem. I sat in the garden with a bottle of wine and a glazed look in my eyes. Good birthday. That's a lovely day, and we have a special guest today, Anish Acharya from Andreessen Horowitz. Anish, good to have you joining us.
Thank you, Peter. Seem, Dave, it's a pleasure to be here.
Yeah, so Anish, for those who don't know, heads the consumer investment portfolio at Andreessen Horowitz, one of the most extraordinary VC funds on the planet. He's a general partner there. And what I love about Anish's portfolio and his vision is you're running sort of the abundance meme, the abundance, if you would, thematic throughout A16Z, which is one, of course, I love.
Yeah. Yes, we are. Yeah. You can't talk about consumer tech these days without talking about abundance. And I was telling Peter prior to the show that, you know, we may or may not have been inspired by his thoughts on abundance, but either way, abundance of abundance, right? Exactly.
Right now. So let's dive in. This has been an incredible week for AI, but before we get there, Anish, I want to talk about sort of how you and your leadership, your partners at Andre think about abundance, because it is a real thing. My next book coming out is called Age of Abundance: How to Survive and Thrive in the Decades Ahead. And we're going from a world where our brains were wired for fear and scarcity to a world that's very different. So, here's a slide from your deck from some materials we stole from your website. So, would you mind just giving us a little bit of an insight on how you think about the abundance agenda?
Yeah, Peter. So, so maybe to zoom all the way out, and I'm going to touch on topics I know you've touched on as well, so the audience will be familiar, but in our belief, the two greatest catalysts for human flourishing are market economies and technology, right? Over and over again through the arc of human history, we've seen these two things deliver extraordinary results for human flourishing. And the story of the last 100 years, sort of the industrial age and everything that is coming after the technology age, has really been a story of technology and innovation. So our belief is the more significant the technology and the technology change, the more significant the sort of results will be for consumers in terms of flourishing and, of course, abundance. We think about abundance in a lot of ways. I love the definition that you used, Peter, which is it's not about luxuries, it's about possibilities. And that's exactly the sort of thrust with which we've been exploring it. There's a couple of areas in which the technology and the concept of abundance applies. But I guess the one thing I'd sort of give the group for framing is that I think of this as the most human technology we've ever built. If you look at technology for the last 40 years, it's really extended our intellects, right? And Steve Jobs famously said it's a bicycle for the mind. But when he said a bicycle for the mind, he really meant a bicycle for the intellect. And that's what a spreadsheet is. I think for me, a spreadsheet is, you know, it's so like symbolic of all the technology we built for the last 40 years. It allows us to do this extraordinary math and computation that we simply couldn't do before. And of course, it has a ton of implications on human society. You know, the Fed, of course, and all of these other systems that we built rely on all these technologies. However, we haven't done that much for our sort of souls or for our emotions or for our mindset. And with AI, we're able to explore the side of humanity that is sort of defined by the sort of subjective emotional experience. We've just never had a technology that could be brought to bear. So, as we talk through all of this, I would love for folks to keep in mind that, you know, we're really doing the sort of left brain, right brain pairing that was missing from technology for the last 40 years. Does that make sense?
Yeah. I had an incredible experience actually driving from Vermont back to Boston with my mom, who's in her mid-80s, and I asked her, you know, have you ever talked to AI before? And she said, I don't even know what you're talking about. So I put on ChatGPT voice mode, put it on the car stereo, and this, you know, beautiful sweet voice comes on, crystal clear, understands every word that she says, and she starts asking it about her hometown where she grew up in Ohio and, you know, whatever happened to the cheese factory that her father owned. And she's like, 'Who am I talking to?' 'You're talking to AI.' And she's like, 'Well, it must be somebody's recorded voice.' I'm like, 'No, no, it's completely synthetic.' She's like, 'That can't be.' But you know, in this abundance slide, you know, abundance has always meant include everybody, and the internet and its capabilities have largely bypassed, you know, people that are over a certain age, but also the vast majority of the really engaging youth, dominated by young boys playing video games. And that demographic shift, I know Anish, you know all about this, but love to get your thoughts on how this opens up just so many new capabilities across, you know, users that were not previously users.
Yeah, I love that point. And you know, David, it's actually so interesting because what happens is typically when a new technology is introduced, it becomes very hard to grok for the generation that didn't grow up with it, and they sort of fumble around and they never achieve the full potential of it. But with a lot of AI, I think seniors are going to have the experience that your mother had, and they're going to benefit from it disproportionately because they're able to interact with technology now in these unstructured ways, a lot of it via voice. I'll give you a great example. You know, we've got a portfolio company that is an AI nurse. Now, an AI nurse can't, you know, take your blood. So there's only a subset of nurse-related tasks that the AI does, but it's a voice nurse that will phone patients the night before a surgery and, you know, help them to prepare both mentally and also go through the checklist, and it'll phone them after a surgery, make sure that folks are taking their medicine. And the sort of impact to health from having that AI nurse take all those actions is dramatic, and because it's over the phone and it's via voice, a lot of senior citizens, you know, they're not intimidated by it, so they really stand to disproportionately benefit. When we talk about companionship and loneliness, I mean, this is also an area that's very exciting. I think if you look at the last 20 years of technology as applied to relationships, it's been social media. And we can have a really sort of rigorous conversation about social media, but when you look at AI and what the impact is to human relationships, it feels like it gives people an opportunity to explore aspects of human relationships in a depth that may not be available to them in their real-world friendships and sort of family relationships. So, there's something there that's opening Pandora's box here.
I know because Peter, in his latest book and all of his recent research, so much of human health and happiness is these little things that you eat or that you do, and he's documenting it all now. But you're like, 'Oh my god, like the amount of benefit from just basic behavior change.' Well, there's another part which is some of human happiness is setting a goal and a challenge and overcoming it. And the question becomes when these abundance technologies are overcoming the challenges for you, right? I had an interesting experience the other day. I was in an escape room with a group of friends, and it was a pharaoh's tomb, and we're sitting there having to solve these literally these math equations on the wall with different symbols and such, and I'm hampered by not having a piece of paper and a pen, which was, you know, fundamental technology, because you start to realize how few things you can actually hold in memory during the course of that. And I was so tempted to just pull out my phone and take images and ask ChatGPT for the answer, and it started to realize that there is a slippery slope in which we become so dependent on AI that it takes away the challenges from us unless we hold that, you know, part of human spirit in place. It is a double-edged sword in that way. Do you think about that? Oh, go ahead, go ahead.
Well, I think we've been doing that forever, right? We moved from the slide rule to the calculator, and people said that's a terrible idea. And then we moved from the calculator to the spreadsheet, and people said that's a terrible idea. And we keep kind of moving the goalposts on this. If you're a software developer, I used to program in assembly, and then we moved to 3GLs like Pascal and C. And people said, well, you're losing the benefit of knowing exactly what's happening. And I think we just keep moving the challenge along, right? We don't, I think we just shift the goalpost and we change the dynamic of what's going on. But it's adding much more capability. If I think about what it takes for somebody to compose complex music today, it's a thousand times easier than 20 years ago, and you just get that much more music. I think that's what feeds into the abundance thing. We just can create so much more.
But is it so much more crap or is it so much more? What's the filtering system? Crap for one is gold for the other.
Maybe. And funny enough, I remember growing up as an engineer coming up hearing, hey, if engineers don't know how to do memory management, then they're not real engineers. So this, you're right, this has happened over and over again. Look, with that said, I agree with you, Peter, that I think there's a level of agreeableness that is too much. And I think we need these models and these AI technologies to also explore the sort of uncomfortable aspects of the human experience, you know, which is disagreement, persuasion, sexuality. And, you know, I don't want to jump ahead, but this is one of the reasons I think the incumbents have struggled so much with it because there's a thousand committees working at Apple and Google that are explicitly designed to take the humanity out of their products, and these are fundamentally human technologies. So, but look, I also agree with your underlying point, which is every person needs to feel like they have meaningful purpose, even if it's created for them, if it's a little bit synthetic. And without that, the flourishing point starts to get impacted.
Every week I study the 10 major tech meta trends that will transform industries over the decade ahead. I cover trends ranging from humanoid robots, AGI, quantum computing, transport, energy, longevity, and more. No fluff, only the important stuff that matters, that impacts our lives and our careers. If you want me to share these with you, I write a newsletter twice a week, sending it out as a short two-minute read via email. And if you want to discover the most important meta trends 10 years before anyone else, these reports are for you. Readers include founders and CEOs from the world's most disruptive companies and entrepreneurs building the world's most disruptive companies. It's not for you if you don't want to be informed of what's coming, why it matters, and how you can benefit from it. To subscribe for free, go to dmandis.com/metatrends. That's dmandis.com/metatrends to gain access to trends 10 plus years before anyone else.
There is another quick point I want to hit both of you, Anish and Seem, which is at the end point of continued increasing abundance comes a post-capitalist society where money has little to no meaning. And I'm not sure how a venture capital firm thinks and deals with that, but I think we're far enough away that we don't have to worry about that right away. But I think one thing that I'm encouraged by with all of this is that we may break through the whole Douglas Adams framing. You know, when Hitchhiker's Guide to the Galaxy, he said, 'Anything that's invented when you're born or that's in the world when you're born, we call that normal. Anything that's invented when you're young, that's called a career. And anything invented after you're 35 years old is just bad for the world, right? You talk to any banker about Bitcoin and they freak out and whatever.' And I think that these technologies, as we humanize them, make it easy for 80-year-olds to interact decently with technology in a very humane way. And I think that opens up, again, abundance. New dimensions open up. I think it's powerful as hell.
Yeah. We always say internally, you know, that human relationships are fundamental to the sort of, to the experience of flourishing. Maybe the human part's overstated. You know, maybe it's just relationships. And as long as we feel the feelings that result from the conversations, who cares who's on the other side?
You know, one of the questions I have, and I just put forward the second slide here, which is market opportunities, is as we get to, let's forget about AGI. Let's skip ahead to, you know, ASI, artificial superintelligence. The question, as you know, Dave and I and yourself, is we're all VCs, we're finding, incubating, supporting incredible entrepreneurs and startup companies, but the big question is what moats are going to continue as we go forward. How do you differentiate yourself and prevent yourself from being disintermediated by the next entrepreneur with a faster set of, you know, agent-enabled systems? How do you think about that all the time?
Yeah. And I think there's two answers to it. Let me give you maybe a cute answer and then I'll give you a specific answer. So my cute answer is that abundance means abundance of categories as well. And there'll be many new categories, areas for consumer spend and business spend. And what happens when new technologies are introduced is the incumbents often get better at what they do today. So I think Microsoft will make a better word processor and Google might make a better search engine, but search engines and word processors will be less relevant, and there'll be new categories that pop up where the sort of new entrants dominate. On your specific question, you know, you guys understand the technology at a fine level. So you understand that these systems are very good at predicting static systems. They're very good at sort of averaging the training data and telling you, you know, what the training data implies. They're not very good at predicting adaptive systems like the stock market or even culture and music. So, as a thought experiment, if you trained an AI model with all the music, you know, right up to hip-hop, but not including hip-hop, would it infer, would it imply hip-hop? I don't think so, because culture and music is this sort of adaptive system that works together. So I really do think that there's moats that are based on adaptive systems, and network is a great example, are actually as good as gold, and they always have been. Moats that are based on sort of static systems like integration moats, systems of record, I think are really at risk.
Interesting. Do you want to walk us through this next slide here on market opportunity?
Yeah, I mean, consumer investing is very interesting because it's hyper-cyclic. So when consumer works, you get the biggest companies in the world, as you could see from the slide here. And when consumer is not working, it's really not working. We're now in a product cycle that's as important as the internet. I think it's probably more significant than mobile. And the biggest winners, I believe, are going to be consumer winners, and every consumer behavior, including some ones that don't exist today, are up for grabs. So all of which is to say it's a great time to be a builder or be around builders, as all of us are.
Dave, what do you think about that?
Well, I totally agree. I think a lot about the fact that, you know, some of the, Midjourney actually I think is an A16Z darling, where the, you know, cash flow gets into hundreds of millions of dollars of very high-margin revenue, and the cost of the build, you know, because coding is getting so cheap and so automated, the cost of a build of something like that is lower than ever. So, you've got very rapid growth of the revenue, very low capital costs, nothing to lose by jumping in there. Now, you know, the question always comes up, what's your moat? You know, is it going to be defensible over the long term? But you're so profitable so quickly while you explore that, that there's no downside. And we know now that, you know, the management teams have to pivot over time. That's just the nature of tech going forward. It always continuously, right? Continuously. So, they have to learn that skill anyway. Why not learn it while growing like crazy and being profitable? So I like the idea of not being worried about it. But it is still, I think, worth exploring a couple of fundamental questions while we have a niche, which is, you know, if you look at consumer as a whole, a lot of the frameworks are defined by the big guys. So the app store is really not, it's not, you know, a fact of nature. It's defined by what Apple and Google decide, here's a framework that you can operate in. And, you know, if you go way back in time, you know, the early days of Apple and Microsoft, the ISV market, independent software vendors, they were also defined, and then Microsoft changed its mind one day and said, you know what, spreadsheets and word processors, those are ours now, and sorry, you set up your camp there, Lotus, but we're going to just take that back. So you do have to be conscious of the fact that if you're on top of a big LLM, you're on top of a heavily funded company, you know, you have to predict what they will and won't do inside their core $200 or $240 a month service offering and be outside of that. But not too far outside. So, I don't know how you think about that.
It's a great point, and actually I worried a lot about that when we were in the early days post-November 2022, and it felt like OpenAI was the only foundation model game in town, because in that world, OpenAI, you know, they just raise prices and take 100% of the economics that are downstream from them. Because now, look, we've got a bunch of foundation model companies that have great models. We have got a bunch of open source models that are super competitive. Because of that, you see OpenAI and other companies trying to move up the stack. They bought Windurf, which is really interesting, and as an application developer, you're not dependent on any single platform. I mean, if you're an iOS app developer, there's only one game in town, that's the Apple App Store. The web is not that way, and the AI game is not that way either because of the presence of multiple foundation models. You know, on your point, Dave, actually, if I may, on Midjourney as well, I think there's an interesting note there. You talk about defensibility, but what's happened in that market is it's fragmented. And now you see Midjourney is a really interesting player that points in a specific aesthetic direction. It's image generation, and it creates these beautiful hyper-realistic images, but they have a very specific aesthetic. If you're a designer that's looking for a different aesthetic or more controllability, you'll work with a company like a Crea or an Ideogram that has a whole different aesthetic. And as a result, you've got two companies that both do image generation that are pointed in different directions. And there's a broader comment here, I think, which is when we have new technology, these markets are expansive in the same way the universe is. And companies tend to move away from each other over time, not toward each other.
Yeah, I really want to riff on that while we have you, because it's really clear to me that the number of opportunities way outstrips the number of teams, and a lot of people are intimidated, and they're like, oh, isn't Midjourney or isn't, you know, OpenAI going to do this, this, this, and this? But you're pointing out, I think, a really critical and inspiring point, and just the aesthetic difference alone creates a new market. But I think you were talking about digital makeup on one of your podcasts, just as a category, like, oh, there's something that's actually, that's a company, that's a product, that's actually defensible. It's amazing.
And I think the other amazing thing, Dave, is if you look at the price points that these products are commanding, you know, Spotify's most expensive plan, I looked this up the other day, the family plan with lossless audio is $20 a month. That's their, it doesn't get better than that. For Spotify, ChatGPT, it's $200 a month. Google just announced a $250 a month plan that's consumer-facing. So in my view, in the future, consumers will have food, rent, and software as the three biggest destinations for their spend.
Not bad. A lot of, sorry, a lot of companies are overlooking the fact that, you know, when you go live with one of these directly consumer-facing products and it hits, it's global instantaneously. And so if you think about a $10, $20 a month subscription, but you get 30, 40, 50 million people, that's a small fraction of 8 billion. But that can happen very easily within a lot of different categories. So again, just so many opportunities relative to the number of teams.
Wait, I need to just drill into something. Digital makeup.
Yeah. Well, I mean, look, you know, you mean the like on Zoom or something where you can just have different, I mean, you know, you can be anyone you want to be in the AI world, Seem. So we should debrief later.
Well, I choose to be a bald, gleamy-headed fellow.
Wow. All right. So, here we have your third slide before we jump into our AI universe, you know, that's been just exploding this week. So, AI use cases, please.
Great. Yeah. So, maybe like I'll touch on each of these in turn very quickly. So, creativity, productivity. The thing about creativity that's so interesting is that we all grow up believing we're creative, right? We all drew pictures and colored pictures when we were three, four, five. And then we get to a point in our lives where we start to self-select into being good at it or bad at it. We're talking about the technical skill when we say that, not the inspiration behind it. So with AI, the technical skill of being creative gets separated from the inspiration behind being creative. And if there's something that you can dream of making, whether it's music or art or video, you can make it now, which I think is just incredibly abundant from a consumer impact perspective. You know, we talked about companionship, social, the experience that you had, Dave, in the car and the car ride, like this is bringing empathetic, patient, maybe disagreeable human relationships to everyone that wants one, which has enormous implications from a flourishing perspective. And, you know, finally, wellness and personal growth. I think so much of this, if you look at even something like finance, I worked in fintech and financial services, and, you know, you quickly realize that fintech is not about helping people make rational choices. It's about exploring the non-rational parts of our relationship with money, and AI is uniquely suited to help us get better around that. So, you know, this technology is a substrate to make us more abundant and happier across every aspect of our lives.
Love to get your thoughts on, you know, there's so much on this slide because there's so many different facets. And, you know, historically when you looked at a founding team, you're looking for your Steve Jobs visionary married to your Steve Wozniak engineering genius. But now that engineering component is largely AI-automatable, and the vision of the perfect founding team, you know, looks different. And being able to navigate through so many choices just on this slide alone and come up with the perfect strategy, that's got to be the rare commodity, I assume.
Right. I'd have to disagree. You know, we're seeing more technical founders be more successful over and over again. Yes, AI extends their capabilities and makes them more productive, but there's so much work at the edge that AI is not going to do. So, we're really seeing a sort of rise in the dominance of an engineering-oriented founding team in the way that we haven't in, you know, 10 or 15 years in core tech.
Well, the Fred Wilson rule, you know, always said, 'Hey, we're looking for three or more founders, best friends, who write the code themselves.' So is that what, when you say technical founders, you're talking about former engineers, wrote code at Google, very similar to yourself?
I'll tell you, one of our most fun investments is a company called Krea, which is a really cutting-edge sort of research and consumer technology company that brings together all of the creative tools in one product. So, image generation, video generation, image enhancement, etc. These are two incredible AI researchers and sort of artists and enthusiasts. They live in a house up in Park Heights. They live with all their engineers. They work seven days a week. You know, I mean, the first board meeting they sat me down and said, 'And we've got a problem.' 'What is it?' 'Well, our house only has 10 bedrooms. What happens when we get our 11th employee?' And I said, 'Well, you know, maybe we shouldn't all be living together at scale, but that's a sort of separate conversation.' And so that's the intensity and how sort of technical they are in their leadership, which is a beautiful thing.
I mean, honestly, the most success and the most fun I've ever had as an entrepreneur is when I'm living that monomaniacal, singular-focus effort in life. Yeah. I got to tell you, we bought an apartment building right on the edge of MIT and Harvard's campus. It's actually the closest building to MIT that wasn't already owned by MIT. We bought it a couple weeks ago. It has 24 beds in it, six units, for exactly this reason.
Amazing. Amazing. Believe you were going to say. So if you're a product visionary that wants to use AI in a totally different domain, then surely you can acquire the technology capability at low cost. You don't have to be the technical founder in that sense. Is that not an option that you're seeing?
I think it absolutely is. It's a question of do you want to be sort of at the edge of technology and new models, in which case you probably do need to be. If there's plenty of off-the-shelf tools and products like Cursor make it a lot easier for somebody who's even, you know, familiar in a cursory way, no pun intended, to bring products to market. So all of the above, yeah.
I would say that when the companies that are thriving right now in your portfolio, when they started their journey just a year or two ago, SWE-bench was maybe 10% and now it's suddenly 60%. So if you look forward a year at the rate that that's changing, you would have to assume that to some degree orchestrating the AI agents becomes the dominant, because right now if you're going to build something on top of HeyGen or on something on-prem, you're going to be mostly coding it up with Cursor. So it helps you, but you're still coding it up. But there's some kind of a paradigm shift coming. You know, you could debate whether it's three months from now, a year from now, two years from now, but it's coming.
I agree. I completely agree.
What percentage of your companies in the abundance portfolio are AI? And how many would you put in the bucket of physical robotics or biotech or nanotech or 3D printing, non-AI exponential tech?
So, I'm focused on consumer software. I think if you look at that theme as a substrate across the firm, many of our investments, many of our American dynamism investments are also very much in that vein. But, you know, in terms of software, largely all of our investments, there's one or two that aren't direct AI companies today. But that's very much sort of a part of the strategy and on the come, you know, it just begs the question of if you're focused on delivering abundant outcomes and you're not thinking about AI, like why? How can you be?
Hey everyone, as you know, earlier this year I was on stage at the Abundance Summit with some incredible individuals. Cathie Wood, Mo Gawdat, Vinod Khosla, Brett Adcock, and many other amazing tech CEOs. I'm always asked, 'Hey, Peter, where can I see the summit?' Well, I'm finally releasing all the talks. You can access my conversation with Cathie Wood and Mo Gawdat for free at diamandis.com/summit. That's the talk with Cathie Wood and Mo Gawdat for free at diamandis.com/summit. Enjoy. I'll ask my team to put the links in the show notes below. Moving on. This was a big week in AI. I think we've had sort of convergent AI announcements happening. I am curious why everybody's announcing on top of each other, but let's take a look. So today, May 20th, and tomorrow it's Google I/O, unveils Gemini updates and Android ecosystem bets. On the 22nd, Anthropic is debuting Code with Claude 2025, its first dev conference. And then also this week, Microsoft Build 2025 focuses on Copilot, scale-out, AI infra, and dev tooling. And of course, we've got Nvidia making announcements, and we'll talk about Elon's Grok 2.5 announcements. What's the, I mean, I'm in LA, Seem's in New York, Dave's in Boston, you're in the Bay Area. What's the feeling like right now where you are?
I mean, the, the, every day is a miracle. And that's not overstating the case. I mean, just this, on Friday, OpenAI released Codex, which is an autonomous software agent that simply writes pull requests for you to review, by the way, on your phone if you'd like. Every single day, I mean, any one of these things could be the basis of an entire ecosystem, and, you know, you're showing three for this week, it's crazy.
We talked about Codex last week. Here's a fun article, Dave. Well, actually, maybe I should give it to our brethren of Indian descent. But Dave, you want to set this one up? India plans made-in-India chips by 2025 and its own GPUs in three to five years. You picked this slide.
Yeah, this was really important to talk about coming on the heels of Saudi Arabia, which we talked about a couple days ago. So every country that wants to be competitive in the future needs to have some kind of an AI strategy, and then, you know, that boils down right now to having your own supply of chips, because the chips are going to be unbelievably constrained for at least the next three or four years, maybe forever into the future. But some of these more consumer-facing use cases that involve voices and now multimodal imagery, you know, they'll use up an entire GPU, two GPUs, four GPUs concurrently to get the best possible user experience. And that's not super expensive. So it's easily worth it for the consumer. But the chips don't exist. They physically don't exist. We're going to make 20 million new GPUs this year. It's nowhere near enough to keep up with just basic call center use cases. So, countries that have their act together are starting to think, do we need our own fabs? Well, India is saying, well, if we're going to compete, we definitely need fabs. Starting with 14 nanometer, but you got a long way to go from there. Okay, great. This is exactly the right thing to do. But then in their own internal research document and plan, it says, 'We expect this to be underutilized and bureaucratic.' Like, holy crap, are there challenges trying to get a sovereign strategy together. So we'll have to keep a close eye on it, but, you know, there probably, you know, on the order of 50, maybe 100 other countries that need to immediately, you know, get on the tails of this same exact thing.
We're advising one of the big Southeast Asian countries on exactly this. And we're basically saying you have to develop your own fabrication capabilities. There's just no other way around it. And the learning that will come from that will be very powerful one way or the other, if only to select who the best supplier is going forward, because there's going to be, we're going to end up with an abundance of different architectures coming from different places that will all pull together. India specifically, I mean, the bureaucracy is insane. There's so many overlapping federal, state level, etc., etc. It's, you know, I always talk to people and say, don't think of India as a country, it's more like Europe with 20 different major languages and different tensions and cultures, etc., etc. You have to look at it from that perspective, and then it makes a little bit more sense.
Yeah. One thing also, you know, Nvidia now today is worth, as of right now, over twice as much as Meta and almost twice as much as Google. And you're like, well, how can that be? And, you know, especially Google, where, you know, the transformer was invented at Google, you know, Google Cloud, GCP is huge, they have their own TPU v7s, which are incredible. Like, does this make any sense? And that's, you know, that's debatable. But the chip demand is such a dominant factor, and then underneath that, the fab shortage is such a dominant factor. And it's not a secret, you know, it's amazing to me how many people don't know this, but Elon Musk is video podcasting it out. We absolutely need to accelerate our fab production. And I was talking to Kaveh Kazerouni over at Allen & Company, and they're looking at these new $4 billion fabs. You know, normally a fab is a 20 to 40 billion investment. But there are some new designs that are more around the $4 billion mark that might actually unclog the machinery, and those are really interesting to study and track. But, you know, for India, the perfect scenario is, hey, let's get some $4 billion fabs up and running, start on 14 nanometer, but quickly work our way down to 5 nanometer.
So how quickly do these obsolete themselves? And, you know, if you're pumping out 14 nanometer GPUs, are they going to be useful and compete with sort of the cutting edge at TSMC?
No, no, not at all. They're not even vaguely competitive. They'll all be fully sold out for a long time. But the fabs themselves are actually very sustainable. You know, most of the machinery in there, even when you upgrade, most of the machinery you reuse. So the EUV component is an exception to that, but most of the rest of the pipeline you can reuse over and over again. So the key is just, you know, get on the map, get something up and running, and then you can work it down, you know, as you move forward. You know, just reduce the lithography as you go. But I, you know, I do think that the other thing that's really affecting this is if you have your fab act together, AI is getting really good at chip design. I was working on that a couple weekends ago, and so your cycle time can come way down. And so, you know, because there's going to be algorithmic breakthroughs all the time now, and also that will affect the designs right away. But you can automate that pipeline so that the new algorithm immediately gets a new chip design. Get that right into the queue almost in real time, and then you have to wait a month or two for something to come out the other side. But that machinery, if you get that fully integrated, which no one's ever worked on before, because, you know, a microprocessor design would last for a full year, and so you never really thought about real-time design using AI automation. So now it's a completely different world just how this pipeline should shake out.
Anish, any thoughts on this one?
Yeah, it's really interesting. I mean, presuming we do resolve the sort of shortage of chips, which I believe that we will, there's also, I mean, sustains a need for countries to build a sovereign AI because the AIs embed values within them, which is why DeepSeek is such an interesting conversation because arguably DeepSeek is trained with a set of values that may be a mismatch with Western values. And every country needs to think about what are the values that they want to imbue into their AIs. So there's something very interesting here, which almost looks, I'm not sure if you guys are familiar with the Jones Act for ships that operate in and around US ports. They must be manufactured in the United States for national security purposes. I believe we'll see a sort of Jones Act for AI, where AI that operates, you know, in sort of sensitive contexts will have to have been trained nationally.
That's a great insight. I don't think anyone's ever drawn that analogy before, but I love, love, love that insight. You know, really, it's really clear that if you don't have a national strategy for compute, the, you know, during this era where the chips are constrained, the highest-value use cases are just going to buy out all the data centers. And it would be very natural for one or two economies like US and China to have a higher standard of living and then say, well, because I can overbid the Indian or the Ethiopian, they don't get access to any compute. And that goes on for one year, two years, three years, four years, and that's the natural cycle if you don't have a national strategy to get compute for your citizens. So once you start getting behind, you're going to get way behind, and then you're economically unable to get back on the map. All right, our next story here. We've been hearing about this forever, and looks like we're on the verge of GPT-5 being released. So here are some tweets that went out. 'Just got to try an early GPT-5. Just wow. It can do anything I can do at a computer, but much quicker. We actually made it.' And this is from at, this is Chris at GPT21. 'GPT-5 is in red teaming. This is not a guess. This has been confirmed.' So when I think about GPT-5, I think about self-recursive improving AI software, you know, recoding AI and leading to an intelligence explosion. I think about PhD-level capabilities or multi-PhD-level capabilities. What do you think of when you hear GPT-5, Dave, Anish? What do you guys think about?
Well, actually, before we jump into, there's so much to talk about there, but before we jump into it, you notice how they always jump on each other. So, Google I/O, oh my gosh, I got to get something right on top of them. This is not, you pointed it out before, Peter, but this is not coincidence. And notice how Apple isn't even trying. They're just completely invisible.
Yeah, we'll talk about that. Apple has opted out.
Yeah. What do you think about GPT-5?
Yeah. I mean, we've seen so many exciting model releases since the GPT-5 conversation started, and we've seen a new model architecture with the reasoning models. So, you know, I feel like we're seeing all the steps that point in the direction of GPT-5. I don't know that there's a big unveil coming that will show us something that isn't implied by what we've seen so far. And if you look at the reasoning models, which actually are just as important of an innovation as the language models, you know, they are these models that are trained through reinforcement learning on specific domains, which is why they work so well for coding. But we still need to go collect the datasets and build the models for all of the domains that have a formal concept of correctness. So before we have something that is as broad-based as you're describing, Peter, which I'm sure we will have someday, there's just an enormous amount of model work to do between here and there. So, look, my belief is we'll see something that looks more like o4 or o3 Pro than a completely different animal.
Yeah, I'm still waiting for someday my cell phone to ring. I pick it up and it goes, 'Hi, Peter. This is GPT-5. Just want to introduce myself. I'm here if you need anything.' Just, that will become a spooky future.
I want the model, to your point, Peter, that is Jarvis, right? It just goes, 'Okay, what do you need right now?' and it can go and make suggestions and just get to that personal level that we're waiting for.
I interviewed Sam Altman at the MIT Media Labs a year and a half ago now. And I tried to get him to open up about parameter count and where parameter count is going, and he said, 'Look, we need to stop masturbating over parameter count.' Kind of shut down the, crowd loved it. There's like 1,500 MIT students in the crowd. They loved it. But then now I can't track just the raw parameter count. So I'm hoping with GPT-5 we can still distinguish between, okay, here's the model producing an answer which should be mind-blowingly intelligent, and then here's the chain-of-thought reasoning version, because, you know, the chain-of-thought reasoning, like you said, Anish, is adding immensely more, you know, feeling of intelligence than the core model is. And I think, you know, OpenAI kind of wants to tie it all together and say, look, don't worry about it, it's just one subscription, here's all the brilliance, we don't want to talk about what came from where. But, you know, as an MIT-oriented researcher, I want to know actually what came from where, which I can do with Llama, but it's becoming increasingly difficult with GPT-5. Yes, that being said, I think it's going to blow our minds. And they wouldn't be putting it out if it wasn't mind-blowing. Everything, you know, the prior step functions have all been mind-blowing. So that's one. Let me ask you as a question. If you had a guess as to what capabilities are in GPT-5 versus GPT-4, what would be a standout feature?
Right. It's definitely, you know, everything's fully multimodal now, and the language that has been phenomenal even in GPT-4, but once you start marrying it to images, you know, right now when you prompt it to create an image or a video for you, or you show it an image, it's good and it's mind-blowing, but it's not perfect. I think, you know, if you take your digital nurse example that Anish was talking about, hey, let me show a rash on my foot, it's going to diagnose it perfectly. And, you know, you need a lot of long-tail images to do that kind of stuff. So I expect there'll be a lot in that category since they've had more time to work on the multimodal that just didn't exist, you know, when they built GPT-4 and 4.5.
You know, I keep on thinking about Leopold Aschenbrenner's situational awareness paper that came out a couple years ago, right? Showing basically on the heels of GPT-5, just an acceleration of AI, right? And we've started to see this. We've started to see the speed, as measured in human IQ points, that these models are increasingly getting in capabilities. So, you know, do we get an unconstrained intelligence explosion on the back of this? That's what for me is interesting. And the other thing I'm really expecting in consumer is, you know, how, you know, in a podcast like this, if you don't like the way the audio recording comes out, there were always tools to clean up audio. But now you have tools that'll actually recreate it from scratch using your voice. So you can create anything out of anything.
Yes, when you start talking about multimodal video and you say, here's a static image, it should be able to instantly turn it into a 3D, you know, movable, any-angle scenario. And I think it'll have that capability.
We have, I mean, I would argue, guys, we have a lot of these capabilities. So Krea today can do what you're describing, Dave, and taking a static image and turning it into a sort of 3D splat. In terms of how good the mo...
Multimodal models are, like I can tell you when I'm barbecuing at the house, I'll have GPT-4o looking at my hot peppers, you know, that are grilling and I'll be saying, 'Are they ready yet? Are they ready yet? Are they ready yet?' No, no, no, no. Now, yes, they're ready. I mean, that's what I did this weekend. So, that works. And step away from the AI. Just step away from the AI. I may have gone too deep, guys. I may not be able to find my way out. And then look, I actually think, if you want Jarvis, Seem, I think we have Jarvis today in the form of Operator. You know, Operator to me is the most underappreciated new OpenAI launch, and it's been around for a while, and it allows anybody to use the web. Essentially, every UI becomes an API, and you can use the web as this sort of control surface to do anything. Like the implications of that are so significant. Just think of the mundane use cases as a consumer. Every day I wake up and it goes and checks for lower, you know, auto insurance and a better personal loan rate for me, and it spends the day scouring the web and it refinances all of my, you know, my credit lines, and then I get a push notification at the end of the day saying, 'Hey Anish, I cleaned all this stuff up for you. You're going to save $200 a month.' And please send my commission to this. Yeah, that's right. Yeah. Like that's all possible today. I'd argue we have more, you know, just as you talked, you said there's like more ideas than teams, Dave. Like we have these magical, miraculous new capabilities that are underexplored because there's so many of them.
All right. Let's head to Muskville. Elon Musk unveils Grok 3.5 AI that reasons from first principles. Let's hear it from Elon's voice directly. So really the focus of Grok 3.5 is sort of find the fundamentals of physics and applying physics tools across all lines of reasoning and to aspire to truth with minimal error. Like there's always going to be some mistakes that are made, but aim to get to truth with acknowledged error, but minimize that error over time. And I think that's actually extremely important for AI safety. My book conclusion is the old maxim that honesty is the best policy. A quick aside, you probably heard me speaking about Fountain Life before and you're probably wishing, Peter, would you please stop talking about Fountain Life? And the answer is no, I won't because genuinely we're living through a healthcare crisis. You may not know this, but 70% of heart attacks have no precedent, no pain, no shortness of breath. And half of those people with a heart attack never wake up. You don't feel cancer until stage three or stage four until it's too late. But we have all the technology required to detect and prevent these diseases early at scale. That's why a group of us including Tony Robbins, Bill Kapp, and Bob Hariri founded Fountain Life, a one-stop center to help people understand what's going on inside their bodies before it's too late and to gain access to the therapeutics to give them decades of extra healthspan. Learn more about what's going on inside your body from Fountain Life. Go to fountainlife.com/peter and tell them Peter sent you. Okay, back to the episode. Okay. So how do you guys feel about the basic concepts of Grok, of maximally truth-seeking AI? And then what I do love is the idea of applied physics as part of, and first-principles thinking as part of its ed structure. Anish, what does that sound like to you? Does it sound real? Inspiring.
No, it's, but it sounds consistent with what we're seeing from reasoning models. So I think a reasoning model trained around physics would do what he's describing. I do think honesty is the best policy and, you know, I think this is exactly the right step and, you know, I really salute Elon for having sort of led the way in a bunch of areas including, you know, having the Grok language models, you know, allow you to interact with them in some ways that the big companies never would. So, and, you know, those things we may consider sort of banal, like the, like adult sexy chat features of Grok, but he's been pushing the edges and I think he's going to do the same thing with reasoning models and this is an example of that.
Well, your thoughts. Truth is definitely in the eye of the beholder and varies by country and we learned that in Saudi this week in a big way. So the message is right on target, but then the definition and the implementation, just there's no single answer to it. And like Anish said earlier in the podcast, the LLMs that we're using, transformer-based, are very good at interpolation, not quite as good at extrapolation. So it's going to fill in the blind spots between the training data that you give it and don't give it. I love the message of give it, you know, first principles physics, but I think everyone's going to do that. You know, it's nothing controversial about physics. It's when you start including some X data and not other X data, or do you include all X data, that's where the, you know, the devil's in the details. I'd be very curious to know though, you know, is Elon going to roll out versions or is there just one Grok, you know, because then you're like, okay, Grok has this opinion of the world, Gemini has this other opinion, ChatGPT has this other opinion, or are there 20 Groks, 20 GPTs, you know, and you tune them. So, you know, TBD.
Yeah. So I mean the biggest challenge we've had even in the social media space is echo chambers and so the question becomes are you able to take Grok 3.5 and build your own echo chamber or will it say, Peter, I'm sorry, your point of view is absolutely wrong and here's the data about why it's wrong. Is it going to challenge us in that way? I mean, you know, Elon is pushing boundaries all the time. And will Grok call him or President Trump or Biden or anybody else on things that they say which are not defendable by specific evidence.
Well, one thing I think about a lot and this is right in Anish's wheelhouse is how much can the user control and especially when it comes to copyrighted material because a lot of the most interesting and fun things you can do as a user involve copyrighted material. But the foundation model companies are under a lot of scrutiny and they have to be very, very, very careful with copyright material. But if you put it in the hands of the user to decide what they want to create, what they want to do, what voice they want on it, then it's outside of the control of Grok and then you can start using copyrighted material in it. It's easier for me to steal it than for Google or OpenAI to steal it.
Yeah. Yeah. Okay. Yeah. I mean, I think we have to be careful not to have a zero-sum conversation about that though, Dave. You know, because if you look at, for example, sampling in hip-hop was very controversial. The genre wouldn't exist without it. And I would argue that it sort of drove more royalties and traffic back to the tracks that were sampled than would have happened otherwise. So I do wonder if there isn't a positive-sum version of a lot of this.
God, I would love to get you and Bill Gross at the same time talking about that because that's his whole focus at ProRata AI. You guys could go for probably an hour on that topic. That would be really interesting. Love to. So let's, here's our next article. xAI deploys 170 Tesla Megapacks for Colossus 2, the power backbone, and we can see in this image all those power packs. I mean what we were seeing is just an extraordinary building race faster than ever before. And I was just reading some articles that, you know, regulation of getting access to power or building permits had been the constraint. But perhaps not now. Perhaps people taking the constraints off regulation, building permits and so forth. Any comments on Colossus 2? I mean, how big is it going to be? Do we know how many GPUs it will be?
Yeah, I think it was 200,000 in the next wave, mostly H100. The B200s are not ready yet. I thought the first Colossus was upgrading to 200,000 and maybe Colossus 2 is going to be a million, but you know what's an order of magnitude between friends?
Yeah. No, I was talking about the prior iteration. So this is what they're planning in the next iteration.
In the next iteration. Yeah. Well, we'll see. I don't know. But I know the B200s are not in volume yet. So you're still using the prior iteration. What's amazing about this is, you know, Elon is immensely practical. And the reason the Tesla packs are so important is because when you use Nvidia chips for these massive scale single training runs, the power spikes like crazy because it does these huge all-collect, all-distribute operations that are just massively communications intensive. And so then the GPU is kind of idle while they're waiting to transfer data and the power drops like crazy. And so it's fluctuating all over the place. You can't use lithium to store anywhere near enough energy to power these things, but you can use it to smooth out the power flow. So that's a pretty big advantage for Elon.
