Transcription
Ready to do it? >> Yeah, let's go for it. >> I'm Rufus Griskum. >> And I'm Caleb Bissinger. And this is the next big idea today. Are we living in an AI bubble? Or are these the last few years before super intelligence makes the world unrecognizable?
[Music]
Okay. So, I want to push back just a little bit on on what you just said, Rufus. Please help yourself. So, I actually think it's possible that both of those things may be true, right? I think I think right now there are people throwing good money at terrible ideas just because the pitch deck says AI a lot and that's going to end badly. But I also think it is probably true that this technology is going to change the world.
>> I love your multi-elabic probably and that's we're going to unpack that. We're going to unpack that in the >> in this uh episode. >> But you make a good point. Both things could be true. We could have a stock market bubble and correction followed by an intelligence explosion.
>> Yeah. Though I also think we should say that like that intelligence explosion even that's not guaranteed right I mean there are a lot of smart people who are not convinced that we know how to build super intelligence you know for instance our buddy Cal Newport who unlike the two of us actually has a PhD in computer science he wrote this really interesting piece in the New Yorker over the summer where he argues essentially AI might not get any better than it is today
>> I would be very happy with that outcome personally but I think it's unlikely Um, as impressed as I am by Cal Newport, one of my favorite of our guests and thinkers, I'm not with him on this one.
>> Yeah. Well, you're also AI utopia build.
>> Well, I'm definitely convinced that this is likely to be the most transformative technology the world has ever seen. It's certainly on its way to being just that. But I am surfing, as you know, the precipice between like AI utopianism and dystopianism. I'm pretty worried as you know which just feels out of character for me. Our guest today though is a great spokesperson for the optimistic view and I love to inhabit the optimistic view.
>> Yes. Well, he inhabits it fully and he is unusually wellqualified to speak, I think, to the mechanics of how neural networks, the technology that underpins modern AI, how this works, because he has been building this stuff since he was an undergraduate at MIT in the 1980s. He's also unusually well qualified to talk about the mechanics of what's happening in the business of AI.
>> Mhm. because this is a guy who built one of the first AI unicorns and now funds dozens and dozens of AI companies started by students coming out of MIT and Harvard and other schools largely in the Boston area. So, who is this guy, Caleb? Well, I mean, I almost feel like you should get to do the honors of introducing him cuz you're the person who actually found him for the show, but I I'll I'll take it. His name is Dave Blondon. He is the co-founder of 23 companies. 23 companies. That just blows my mind. He's currently running a fund called Link Ventures. He's an instructor for a wildly popular AI entrepreneurship class at MIT. And he co-hosts the Moonshots podcast with Peter Diamandis. And Dave says we are all in for a wild ride. Yes, he does. He says in the next few years we're going to see an explosion of AI capability which some call the singularity after which he says we may soon be immortal potentially getting to escape velocity and living forever which seems likely actually.
>> AI is likely to become the most powerful military weapon which has incredible implications for geopolitics.
>> If we're not winning then we're losing. and he offers practical advice for students and for middle-aged farts like us. Hey, you're almost there, Caleb.
>> Jump in. Play with all the tools. Don't be afraid to quit your day job cuz you need to focus like crazy. All right, strap in, people. My conversation with Dave Blondon right after this quick break.
As you probably know, before we started this podcast, we started a book club. Malcolm Gladwell, Adam Grant, Susan Kaine, and Daniel Pink help us pick the very best new non-fiction books of the year. Books by leading thinkers whose ideas will be driving the conversation in the year ahead. I'm thrilled to share with you our latest selection. It's a brand new book from Bnee Brown. It's called Strong Ground. And if you join us right now, we'll send you a signed copy. Plus, this is really special. We'll invite you to a membersonly Q&A with Bnee and our curator, Susan Kaine. You'll also get reading guides, audio courses, ad free versions of this show, and access to our new WhatsApp community. We have an unbelievable lineup of books coming in the months ahead. I'm really excited about it. and we're planning a series of live events in New York City with live podcast interviews. If you want to come along for the ride, head to nextbigacub.com. Use the code podcast at checkout for 20% off your membership. That's code podcastub.com for 20% off. We can't wait to have you as part of our club. Hope to see you there.
This episode is brought to you by Indeed. When your computer breaks, you don't wait for it to magically start working again. You fix the problem. So, why wait to hire the people your company desperately needs? Use Indeed's sponsored jobs to hire top talent fast. And even better, you only pay for results. There's no need to wait. Speed up your hiring with a $75 sponsored job credit at inddeed.com/mpodcast. Terms and conditions apply.
[Applause]
Dave Blondon, welcome to The Next Big Idea.
>> Oh, thank you so much for having me on. I cannot wait. It's my favorite thing to do. And uh I'm a huge fan, by the way.
>> Oh, amazing. Uh well Dave, I'm delighted to have you on the show because in my own life, in my conversations and reading and podcast listening, I've found there to be a disconnect between how most people in the tech sector see the world right now and how most people walking down the street see the world.
>> Yeah.
>> When I talk to people in tech, like we we've had, you know, people like Bill Gates and Reed Hoffman and Steuart Russell on the show.
>> Congrats, by the way. That was some big big
>> Yeah, it was amazing. And also talking to people who watch tech closely like you know you've all know Harrari and historians you know philosophers you get the sense that we're at a critical moment of acceleration of the most dramatic tech revolution in human history.
>> Oh you got that right
>> right? I mean this is just an unprecedented moment.
>> But when I talk to most of my friends they say this is cool but it may be a bubble. Maybe it's overhyped. Maybe I'll play with some tools now and then, but I'm basically sticking to my knitting. You have a unique perspective on this. You're an early AI innovator, an entrepreneur. You teach entrepreneurship, computer science at MIT, and you're an investor in AI startups, and you're a dad with kids coming up in this crazy changing world to lobby a softball. Is AI underhyped or overhyped?
Yeah, that's a softball of all softballs because you've had all the big wigs including Bill Gates. They all say the truth, which is it's underhyped. It's it's absolutely going to change the world in the next couple years more than any change in human history. There's nothing even vaguely comparable to it. And you're dead right in in what you said a second ago. Today I'm at Vesmark, which is one of the companies I founded along the way. Asset manager, manages a couple trillion dollars. It's a big big company. And all the young people here are super excited about AI, but all the old people are like, "Well, you know, I tried it. It's not so great. Nothing's going to happen."
>> Yeah,
>> you are so hiding under a rock right now. And I know it's hard to change. I know it's hard to adapt. I can relate to that. But it's underhyped and it's going to explode. And it's just the reality.
>> And I mean, you and I grew up during the computer revolution and we've more recently lived through the internet revolution
>> and and that has profoundly impacted our lives. But so so you'd make the case that what's happening now and what is going to happen in the next say 5 to 10 years we should expect that to be more disruptive than any of these these prior tech revolutions.
>> Yeah. Absolutely. Well, so you know I lived through the AI or through the internet revolution and was really really lucky to graduate MIT right when I was about to explode.
>> But you know when I teach at MIT the you know foundations of AI ventures every one of those students has to build an AI business plan. And I'm always telling them look guys
>> you were born after the internet. You don't know life before the internet. You never lived through a moment in time like this before. Stop sleeping. Rush to everything you do. Don't lose a microssecond of it. Because I was incredibly lucky. You know, first company I founded, we got acquired for about a billion dollars in 2000. But if it had been like 8 weeks later, the market came apart and then 911 happened right after that. So it came apart even farther. So you don't really know how things are going to evolve. But I can tell you right here, right now, easily a factor of a hundred bigger moment in time than the internet, which was massive.
>> I mean, it was like life-changing. So, now's the time.
>> That's a huge statement. And and of course, we, you know, we'll talk later about sort of the arguments for I'm going to try to steal man the argument that we're in an AI bubble.
>> Good luck. I'll take you down.
>> You're going to take me down.
>> Good luck, though. But I think the case could be made that actually the.com crash in 2000 which you narrowly avoided
>> that actually everyone was pretty much right in 1999 about about what the internet was going to do in terms of the impact of this technology and how profoundly it would change our lives, how much value it would create.
>> Yeah, I'm so glad you said that because because a lot of the finance types uh say, "Yeah, you know, it was overhyped. It was a bubble and then it collapsed." It's not true. It was a catastrophic collapse of confidence, but you know, the internet was everything everyone predicted and then a lot more. So, it was actually more of a retreat from reality than a than a bubble. It's like it's real. And then AI is real, too. It doesn't mean doesn't mean people aren't going to make some bad investments. Doesn't mean some things aren't going to come down there. That's going to happen because this is America. But the reality is AI is going to overperform everything that people are predicting right now by a lot. Not Not by a little.
>> Wow. Well, my my hope for this conversation is that we can get into the details of like what's happening right now, how you see the world changing and and importantly what practical steps we should each be taking in our own lives to navigate these changes because it is like when one hears these kinds of radical prognostications. You just said, you know, hundred times the impact of the internet, right? It's it's almost too much for most people to gro or process. And I think I I think it'd be interesting to kind of dig into the practical implications for everyday people. I mean, I assume that you think that there are practical steps that most people should consider taking or that it's useful and important to think about and anticipate the changes that are coming.
>> Yeah. No, I think it's going to affect everybody and if you play your cards right, it can affect you well no matter who you are.
>> Uh so it' be great to talk about what you can do to to maximize your chances of benefiting and not getting crushed by by what's coming. I think it's important to put it in context, too. I I mostly grew up in Iran, actually, right before the Iranian revolution. And the the time that our family moved over there, the life expectancy in that country was sort of uh it was 43 in the '60s and then it grew to 53 when we moved over there. So, it was it was a lot like uh the US in 1880, 1890 in terms of the state of technology, the state of life expectancy, the state of standard of living. And so when we moved back to the states, I was really shocked that, you know, there's a there's a bunch of people that are technophobes or that say, you know, the world used to be so much better than it is today and we need to go back to, you know, humans are destroying the world. We're making it warm. We're making it, you know, we're destroying it for the future of humanity and technology is evil. like, well, why don't you go live in the third world for a while and see what it's like without technology and try and rewind your mind to what things really were like? You know, we we won the American Revolution. Half the British soldiers died of malaria and yellow fever. That's why we won. Okay. It was not a pretty place before we had technology,
>> medicine,
>> y
>> shelter, and now internet to stay in touch. Uh so technology fundamentally has benefited humanity tremendously. uh so fear of technology is completely dysfunctional. So now as AI accelerates into the world, you know, a lot of people are hiding under the rock again saying, "Okay, you know, I'm afraid of technology. I don't want to be involved. It scares me. Whatever. My iPhone is confusing as all hell. I I want less of it." The wrong attitude. And I and I totally can relate too, by the way. It's complex as all hell. And you know, products pre Steve Jobs. See, if if we lived in the world that Bill Gates had originally built with Windows, we'd be crashing every 5 minutes and we'd be in this this torture world. And Steve Jobs single-handedly showed us that it can be a delightful, fantastic out of the box experience that you love. And if we get more people who have that that capability and that attitude working in technology, this can be great for everybody. But there's no doubt we're coming into this era of extreme abundance. The capacity to build human paradise is right in front of us and it's ours to either make or to or to mess it up. You know, there's a lot of scenarios where we mess it up.
>> Yes.
>> And we could talk about those too, but it's it's right in front of us. So,
>> let's start with your background. You decided pretty early in your life that neural networks, artificial intelligence was going to be a focus for you.
>> Yeah. Why did you and this would have been in the in the early 1980s. Why was it that you came to that conclusion?
You know, I had an experience when when we first moved to Iran when I was a little kid. We went to the airport, we went to our house, and then a few months later, it snowed, which is rare in Iran. It's high, very high altitude, but it's And the whole airport collapsed, and everybody in it died.
>> Oh my god.
>> And I was like, okay, well, here we are. There's no airport. They'll rebuild it. This is sort of the world without without standards, without laws, without construction, with bribery, with gra. It's a messed up it's a messed up world. And when you grow up in the US and you're in sort of a private school world, you you don't really appreciate how the natural state of the world is messed up. Uh and and vast majority of humans live in messed up worlds, messed up countries, messed up ecosystems. And so um you know, I I was not into computers originally. I was into planes, aereroastro uh really love science, but but you know, more more the the physical stuff. And then the Apple 2 came out and we got one of the first ones and I was instantly in love with wow I can see how this Steve Jobs you know said it's the bicycle for your mind. It's going to accelerate the capability of your mind. I can see how this benefits humanity
>> more than anything in history and I need to understand and study this thing. And at the same time I was reading all these articles by Marvin Minsky
>> who who was at MIT running the AI lab
>> saying you know artificial intelligence strong AI thing an AI that can do anything a human being can do and then super intelligence right after that it's going to happen in your lifetimes not in his lifetime as it turned out but in my as a kid lifetime. So I decided I think I was 14 13 at the time listening to my little LCD clock radio at night thinking what am I going to do with my life? Yeah, I I decided, look, I I need to study really hard. I need to get to MIT. I need to meet Marvin Minsky, and I need to be in the middle of AI if it's going to happen in our lifetimes. I can't miss it. Otherwise, I'd be wasting wasting my life. So, so I just committed to it right when I was a young teenager and worked my ass off and got over at MIT.
And so you basically believed Minsk's message or you intuited yourself that neural networks on silicon had the potential to be transformative for the world.
So the Minsky was saying that artificial intelligence is inevitable in our lifetimes. But he was very anti-neural network. So I got to MIT. I was very disillusioned when I got to the AI lab because I had worked so hard to to get there. But I didn't think what they were working on had any chance of working. But uh the neural network was invented when I was a sophomore and I picked up the paper right away and I said, "Holy crap, this is the future." And so I had to lobby very very hard to get access to the biggest supercomput on campus, which was the biggest in the world actually at the time, non-military. And uh they said you can use it between midnight and dawn when nobody else none of the grad students are using it. So as long as you're willing to be nocturnal, you can start building your neural nets on the on this beast. So I did that for I I did it officially for two years and then I they didn't turn off my key or my password. So I I kept doing it for three years after that and nobody seemed to shut me down. But I I really wanted to be early to understanding the capabilities of neural networks and have been you know working on them following running actually I'm still running source code on my laptop that I started working on in 1993.
So
>> wow.
>> Uh so I'm yeah it's it's been a lifelong passion for me. And so here you are now many a few decades later right back where you started at MIT teaching an AI entrepreneurship class and and running a fund next door link ventures that invests I think exclusively in AI startups many of them founded by recent graduates of Harvard and MIT maybe sometimes dropouts even it's kind of embarrassing to graduate these days isn't it means you didn't come up with a good idea
>> I'm yeah I'm not supposed to comment too much on that as schools get a missed, but it yeah, you know, hey, Bill Gates and Steve Jobs set the trend years and years ago and now it's like it's all the rage now. And and I it's hard to argue against it because again, the window of opportunity is relatively narrow. Uh it's a great time to be starting your company. If you don't happen to be a senior, what are you going to do about it? You know, you can try and race through and graduate early or you know, yeah, it's it's a tough choice right now and but so I'll bite my tongue before I say anything more. timing timing matters is is the thrust of what you're saying and and c can you describe some of the most extraordinary examples of some of the startups that you've backed that are accelerating right now? I think some of these companies are run by folks between what 18 and mid20s who have I mean not only are they achieving just these extraordinary valuations in shorts amount of time but their revenues are growing at paces that I think are historically unprecedented.
Yeah, they're they're running 4x all-time highs in terms of revenue growth and and time to the first million of revenue. Um, so it is a Yeah. So, you know, when the internet was in this kind of phase, those companies were largely unprofitable, no revenue, very little revenue. So, these companies are very very different. They have a deep foundation. Demand is off the charts.
>> What's a specific example of like pathway in x number of months, what kind of revenue? What what's the product? C can you paint a picture here for listeners of like what's happening?
>> You know, just because I I literally just had a board meeting at Farsight earlier today. Farsight is uh three best friends from MIT. Uh started the company a little over a year ago. Uh just hit 2 million ARR. I I hope that's not confidential. And uh you know, most of that revenue growth was in the last 3 months. You took them a while to build the product. They launched it just a few months ago. Uh I think they said uh they went from 3 to 18 employees in the last four months. and valuation, you know, went from seed, you know, near zero to, you know, not not 100 million, but, you know, mid double digit millions valuation. Uh, they have tons of cash. Uh, so that's that's a pretty typical story, but like Meror is a next level. You know, Meror is only coming up on its third birthday now. $500 million revenue run rate, $10 billion valuation, and and the secret to their incredibly fast, I think probably world record-breaking revenue growth got to be easily by far the fastest revenue growth. I can't imagine anything getting that big that fast before. But they're selling to uh Meta, OpenAI. They they have five of the seven largest AI labs as their customers. And you know, these guys are also growing like crazy. So they have infinite needs. And so Meror is selling into this same ecosystem that's growing like crazy.
>> And and what does their product do?
>> Meror is the coolest thing in the world. Uh so they started with this idea that AI can interview you. exactly like we're talking right now. It has a face.
>> It has a voice. It can speak any language all over the world. And it can ask you really tough interview questions. And more importantly, it can hear everything you say in gory detail, including recording the audio, but also transcribing it to text. And then later, if you want to evaluate everyone you've ever hired, you can look at every word they said. Like, oh, this guy turned out to be amazing. Like, what did he say in the interview? Everyone usually forgets like, well, the AI doesn't forget. Knows exactly what you said. So then they can try and correlate that across other candidates. And one of the one of the thesis there was there is talent all over the world that's undiscovered. And nobody has time or language skills to go find it. So the AI will scour the world and find all these people.
>> So the discovery might be that people who rode unicycles in high school become 10x developers or what whatever the correlation is. But there but there's a lot of value in finding those correlations.
>> Exactly. Let me rewind that story one one more notch too because it's a really important pre-story to the story. So there are three best friends which is a very common theme. In fact, we will almost never invest in a company if it isn't three or more best friends. And I'll tell you exactly why. If they're going bad, stumbling along, it's no big deal. If it's going really, really well, you know what's going to happen?
>> Mark Zuckerberg is going to call you and he's going to try and hire one of you for a billion dollars. And you have to say, "No, I'm working with my best friends. I'm sticking with this. That's a very very high bar. So, if you're kind of loose-nit, uh you're very vulnerable to that. And that's our number one failure mode in investments. They almost never fail unless you give up.
>> Interesting.
>> And so, we don't want them to get ripped apart. So, they have to be super super tight best friends. So, these, you know, two two of them went to Georgetown, one went to Harvard. Because one of them went to Harvard, they were eligible for prod. And prod is this MIT Harvard Stanford joint incubator. And in prod they go bungee jumping in Mexico into sink holes. They go skydiving. They do all these extreme bonding things which is a great great way to really tighten up your relationship. But what I heard from the prod guys is that they had a plan before the merkor plan. It was terrible. Didn't matter because they're smart and they met the other teams in prod. And once you meet the other teams they tell you your idea is terrible and you fix it in in no time flat. So that's where the Mor do the AI interview idea came from. And then they they refined it even more after that. Then they said, well, you know what? We can actually really focus in on these RHF engineers, very specific type of engineer that all the AI labs want to hire by the ton in order to tune and train their AI and make it perfect at all these different use cases. So they really honed in on that even further. And here they are today just uh I think probably the hottest or among the three or four hottest startups in the country right now.
extraordinary. And you have a building, I think, next to the MIT campus, and you've got what, dozens of of startups there. And I mean, you're you're living in the white hot center of of the AI startup scene. And do you anticipate that this will continue? I mean, I guess you see a window of a few years where it may be possible before before AI agents are are are spinning up their own companies. What what is the constraint on the window here?
We can talk about that last sentence in a minute because that really gets that we'll go way down a rabbit hole that down there in a second. But um yeah, we have uh I guess 28 now. Uh one of our companies just moved to Silicon Valley which is a common story. So I guess it's 27 now uh in the lab. It's right next to MIT. We're we're just incredibly lucky in that, you know, MIT is now 50% computer science majors and the other half that's not computer science is usually math with computer science or biology with computer science. So, so most of those are also some form of computer science. You know, Stanford and Harvard capped the computer science program at about 15 to 18% of the student body because they don't want to become kind of monopolar, you know, unilateral campus. They want to be diverse. And so they're like, "Look, the literature department needs some students. The architecture department needs some students." I just found out there's only four architecture majors left at MIT in the senior class. Everybody wants to do everybody wants to be Tony Stark and build an AI company and have Jarvis. And so MIT kind of gave up on fighting against it and said, you know what,
>> we have cross registration with Harvard and Tufts anyway. So let's just let it flourish and grow. And so the byproduct of that is the talent pool coming out of MIT is about five times bigger than Stanford. And then Boston as a whole is not just MIT, it's MIT, Harvard, Nor Eastern, BUBC and Tufts. So collectively it's about 20 times more than the combination of Stanford and Berkeley. And so the amount of talent is like nothing else on the planet just in this really small area. Like you can you can walk or blue bike to every one of those campuses. and the campus is mostly touch now. You know, they've grown so much. So, it's almost like one massive university environment that is, you know, unmatched by anything else in the world. And so AI is so big that it's only constrained by talent. There's no capital constraint. There's no opportunity constraint. The backlog of business plans is like nothing I've ever seen. So, it's all talent constraint. We just happen by more or less luck to be right in the middle of the epicenter of the biggest talent in the world. It's it's just it's just a wild time.
>> And do you think this will change? I mean I mean San Francisco has had has so dominated the tech scene for a couple decades now. Do you think that's likely to change or you you think that once once these companies get above a certain size in order to scale they'll move to San Francisco or might how do you think this plays out?
>> The sad truth is the second thing you said that that San Francisco is is accelerating its lead just by trying to mess it up as hard as they can. I mean doing everything you possibly can at the government level to screw it up but the tailwind is so strong they can't all the you know all the foundation model companies are right there you can other than Google you can walk from one to the next uh and then there's $4 trillion of fresh liquidity from just from Nvidia Apple Google growth
>> $4 trillion of liquidity flowed back into the valley
>> and into the city and so there's nothing in the world that's even close to that so the sad truth is most of the companies will in Cambridge, Boston or you know other hot beds India and elsewhere and the really hot ones will usually move their headquarters to Silicon Valley or San Francisco more San Francisco these days up in the city. It's the cool place to be.
>> Yeah.
>> And that's cuz that's where the capital is. You know, the term sheets will often say, "Hey, look, we'll give you twice as high a valuation as anyone in New York will give you, uh, but you got to move your headquarters out. You don't have to move all the people out here, but you got to move your headquarters out here." So then people go out and the weather is perfect and like you know everybody's talking about AI. There's no deniers out there. You come back to Boston or New York or anywhere else in the in the country and there's deniers everywhere and so you feel like yeah I really want to be here because everyone's like-minded, you know.
>> Yeah. Yeah. Well, so this this is a good segue to what's happening in the broader macro AI marketplace, right? I mean we're seeing as you know just investments in AI uh that are just unprecedented. We have the $500 billion dollar Stargate project, you know, with OpenAI, Oracle Soft Bank.
>> Um, Nvidia just invested a hundred billion dollars into Open AI. Meta and Google are investing hundreds of billions. They're they're trillions of dollars of capital flow flowing into building, you know, new infrastructure to support the next generation of AI models. The near-term constraint seems to be chips, right? Nvidia is clearly the market leader, first company to hit a $4 trillion market cap, 130 billion in annual revenue, more than doubling every year. Right. Again, no precedent like you you described startups, you know, going from zero to several hundred million in revenue in two to three years, which is unprecedented. Uh, but but having the largest most valuable company in the world double its revenue annually is absolutely without precedent.
>> Yep. Mindboggling. A and then every major tech company in the world, as far as I can tell, is trying to build their own chips to compete with Nvidia's. Google's TSUs seem pretty impressive. Where do you see this going? And and how long do you think this chip shortage is is likely to last? What's at stake here?
>> Oh, the chip shortage is going to be incredibly bad. Uh just just crazy bad. Um and we can walk through why, but uh you know, people tend to think that Nvidia, Broadcom, AMD, these are chip companies, right? Well, no, they're chip design companies. They don't actually make them.
>> Yeah.
>> So, then you look one level deeper and um I'm I'm killing it on Intel stock, by the way. But the only thesis on that is look, the US is going to make chips. And the US is never going to let one company be the only chip company in the world. Right now, TSMC, Taiwan Semiconductor Manufacturing Corporation, makes twothirds of the world's chips. Or, you know, the real chips, the AI chips, the ones that the ones that matter.
>> The real chips.
>> Yeah. I mean, they're really the only ones that that get really huge margins and that but you can't have twothirds of world supply of the most critical asset in the world right now be in the hands of one company. The US would never live that way. So then what are the other choices? Well, Samsung, but that's Korea and Intel is the only other one. And so, you know, I think one of the highest priorities for America right now is is actually rebuilding Intel and getting the getting the the volume of fabs up 10 20x as quickly as humanly possible because Nvidia will absolutely sell every single thing they manufacture at a huge premium like 70 to 90% margins on everything they make and that'll last for years. but they can't make more and they're fighting with Apple and Google over what is TSMC going to make. Yeah.
>> And it's it's it's not a pretty picture there. You know, it's it's
>> and and we've been trying to do something about this for some time, right? Like the Chips Act was meant to help kind of repatriate or build build fabs in the US. I think Taiwan Semiconductor has been committed to starting to build fabs in the US. Intel had, you know, was a beneficiary of the chips act, but then we just had in August the Trump administration, I believe, accelerate that funding of, I think in total close to 8 billion
>> in order to advance the Intel you fab improvements uh and and and in a weird unprecedented twist take a 10% the US government's now taking a 10% stake at a major US technology company, which I think is sort of without precedent
>> and raises some eyebrows, but It sounds like your view is this was this was necessary geopolitically. It's important that the US have fabs domestically that are competitive.
>> Yeah. Critical. Uh and it's funny. I'm kind of torn. These moves that Trump is making are absolutely the right thing to do and they're they're brilliant. And David Sachs is incredible. Like getting him convincing him to join the the federal government is like it's like the coup of all coups. How would a guy that successful, that wealthy, and that right in the middle of AI go down to Washington and join the federal? But for whatever reason, Trump can pull people like that in, which has been great. Uh, on the other hand, uh, the federal government making commercial investments is a terrible precedent, right? It's it's absolutely frightening where that will this is exactly what Eisenhower warned us against, you know, years and years ago. If the military-industrial complex becomes interlocked with the government, it's going to get corrupt in a heartbeat. And so, even though we need these moves right now, it's critical to do these things right now. I do worry a lot about government involvement in in private companies. I'll also say that the Chips Act was a piece of dung, the original chips act, and I met with Greg Lavender, the CTO at Intel, and he said, "Look, they theoretically gave us $20 billion, but we haven't used it." like why haven't you used it? Because it has 4,000 pork things attached to it that make it negative value money. Like okay. So I I think they threw that out and they said, "Look, we'll just make a straight equity investment, a simple $10 billion. Here's the money you need to finish some fabs and that started, you know, then right after that Nvidia invested and now they're talking to Apple and it put them back on the map in terms of credibility and I think that's what what they needed." I did I didn't realize that this is exactly what Ezra Klein and Derek Thompson wrote about in their book Abundance, but just all all the complexity of all the things that get written into these bills that make that that sabotage them. It's it's a pity because that was certainly not the original intent.
Listeners of the show are on both sides of the aisle and we we make it a a point to be inclusive of a diversity of opinions. I think I've heard you say that US government buying 10% of of Intel that your conservative friends applauded this as a great deal. Your liberal friends were horrified by the US government effectively nationalizing pieces of pieces of companies. So it sounds like your own your own politics here is complicated. I mean I I find myself I mean on the one hand we we have some very shady corrupt things happening with uh UAE investments in Trump family crypto funds and some real kind of ugly things happening. On the other hand, uh there does seem to be a a a strong real politique argument that we are actually in this critical moment in which AI supremacy maintaining an advantage for the US in the development of artificial intelligence is critical to the the security of the world to some degree.
>> Yeah. Well, I look, AI is going to be more powerful than any military weapon in the history of the world.
>> Um, and if we're not winning, then we're losing. And I think everybody understands that. I think, you know, there's no real threat other than China, but China is a very real threat. And I don't think China is trying to win it as an arms race, but they're trying to win it as a power broker in the world and an influencer in the world. And and China, you know, feels like the long term, you look at the 2,000year history of the world, China's been dominant for many, many years, a big fraction of that. America is just like this newbie. and they feel like that's not the rightful order of the world. But the difference is going to be who wins the race to strong AI. It's it's crystal clear. And if you're not winning, you're losing. And if you lose, it's not good.
>> Yeah. And and another piece of this story when we talk about China and the United States is energy, right? Because because we talked about uh chip shortages, but there's also energy constraints, right? How important is that? and how and how do you think this is likely to play out?
>> You know, I was talking to Eric Schmidt, you know, the former Google CEO, and he thinks energy is more important than than chips. And uh I I kind of disagree. I think that the chip shortage is imminent, acute, and massive. Uh we will catch up in energy if we get going right away. We need to find about 90 gawatt by 2029. That's about a 10% expansion of our power supply. If we don't successfully build it, it's going to have to come from somewhere. Well, it's not going to come from households. It's not going to hurt. You're not going to It's going to come mostly from big manufacturing like steel and aluminum that they use huge amount of electricity and aluminum manufacturing.
>> And so we may have, you know, some problems in the aluminum industry while we divert all that power to to data centers, but that's not exactly tragic. That doesn't like destroy America or anything. It's It's not great, but but we'll get going on the energy creation, I think, in time to catch up with the fabs. I think that the scenario where you lose to China isn't the one where they out energy you cuz they're they're already miles ahead in energy, like like 10x, like miles ahead in energy.
>> Um, the scenario is where they start building their own fabs. They're not as good as the US fabs,
>> but they out algorithm us. And you know the the fabs generally you're looking at 2 to 10x differences in speed of the chips. The algorithms a good idea like one brilliant person thinking late at night can add a 10x just with an algorithm change. And so there's a lot at the software layer where you know enough very very smart people working in very focused labs can outthink you and that's where you potentially get blindsided and that that's why uh you know Quen and the other models open source models coming from China were such a wakeup call to the US cuz yeah these things are great I mean they're absolutely great. How did that happen? Running a business comes with a lot of whatifs. But luckily, there's a simple answer to them. Shopify. It's the commerce platform behind millions of businesses, including Thrive Cosmetics and Momaf Fuku. And it'll help you with everything you need from website design and marketing to boosting sales and expanding operations. Shopify can get the job done and make your dream a reality. Turn those whatifs into. Sign up for your $1 per month trial at shopify.com/sp specialcial offer.
This episode is brought to you by Tic Tac. Your boss is going off at you for missing a deadline and you swear the clock just moved backwards. Suddenly you remember the pack of Tic Tacs on your desk. Now you're riding horseback on the beach, the smell of citrus in the air. And what's this? Your boss is sitting behind you breathing heavenly fresh mint. Refresh your mouth and mind and share your good vibes with Tic Tac.
[Music]
GIF peanut butter. It's that tiffing good. It doesn't really need advertising. So instead, enjoy the calming sounds of the ocean.
[Music]
It's that jiffing. Get your GIF at tiff.com.
[Music]
Taking a step back. So, I'm now going to fearlessly try to steal the argument that we're in an AI bubble, which which I'm sure you're familiar with. Like The Atlantic had a had an article recently to this effect that there's been a a growing skepticism. I I I think probably as a result of some sobering news that came out in recent months about AI productivity claims, right? There was this July meter study showing that developers working with AI completed tasks 20% slower despite optimism internally. There was an MIT study released in August that found that something like 95% of Gen AI pilots at companies are failing to increase productivity or profits. Then people are pointing to kind of this massive investments of of approaching a trillion dollars and soon trillions of dollars into of capital into these projects with I think so far 35 billion of AI related revenue this year and there's a concern that hey this looks a lot like what we saw in 99200 we're maybe getting out over our skis here. What do you think?
>> Well there's two different thoughts you know there's out over your skis and there it isn't working. Those are different thoughts. But if you if you really dig into the archives, there's a paper that's absolutely worth digging up by the way. Uh I think it was Anderson Consulting, now Accenture, but Anderson at the time.
>> Yeah.
>> Uh computerization of offices actually reduces productivity,
>> right?
>> So you have people who have pencils and paper doing accounting and stuff. And when you computerize it with these early computers, your office actually is less productive because the IT guys that are constantly rebooting it and then no one can make it work are actually more expensive than just doing it on paper was in the first place. And it's worth digging up that paper because it says the exact same word. It's almost word for word what you just said. And obviously insane in hindsight. And so I know exactly what you're talking about because you know here at Vesmark where I am right now you know we we're trying to automate a whole bunch of white collar jobs and by and large it's going really really well but if you don't know where to use it how to use it where you know where you're trying to force fit it. It's very easy to draw the opposite conclusion and say look you just wasted my whole day and I don't even see how this benefited us and now we're net behind. So yeah you could you can draw that conclusion. Couple things you're missing in there. First of all, with a little more effort, it seems to work every time in in my experience here. Second, it's improving at a rate that's mind-blowing.
>> Yeah.
>> You know, it's funny. I I put all four of my kids, they they hate me for this a little. They don't hate me for it, but I mean, I had them all come into our lab in Cambridge and
work with GPT2. And so, they're like, "Dad, why are we doing this? Dad, why are we doing this?"
"Well, you're doing this because I want you to see the evolution from from this new capability because it's going to change really, really quickly." Now they use it every day and they absolutely love it. GBT5 now >> but they had that experience of of watching the evolution >> and I think that that is you know invaluable.
So now you got all these corporate executives trying to write papers saying look it's it's a bubble it's net negative you know it's not actually it's it's it's fake it's fraudulent whatever actually I'm getting it from both sides now. You've got the people who are the deniers saying it's not real. Then you got all the doomers saying it's imminently going to be superhuman and take all it's going to run us like in two years. It's going to be managing us. It's going to be our boss. It's going to be our our god. You're like, okay, you're both wrong, but you're both you're both panicking and you're actually going to miss the opportunity to take advantage of this moment in time because you're you're you're off in delusion land on both ends of this spectrum.
So I I write code all the time. I write write code every day. Uh uh it is interesting to me how if I'm trying to write code the traditional way and I'm trying to use cursor or you know I usually use claude sonnet max which is the most expensive you can get and it still makes insanely bad mistakes and it pains me to see it make these stupid mistakes but it's easy to forget that it's still accelerating my coding by probably about 5x I'd say. >> Mhm.
But then on the other side, I'm doing this pure vibe coding project where I'm not even looking at the code and I'm purely telling it how to evaluate and improve itself. And that's working incredibly well. There's an uncanny valley in between where you're kind of half in the vibe coding territory, but you're half not. And that's where people are getting the bad experience, >> right? >> But if you if you get out of that >> and focus on the end points, you get a very good experience.
But yeah, let's go at it, man. Let's let's let's what's what's your what's your counter argument? >> Well, no, I'm I mean I'm actually I'm steel manning because I'm I'm really on your side of the argument here. But I but I I was actually very very amused to see Sam Alman say a few weeks ago, when bubbles happen, smart people get over excited about a kernel of truth. Are we in a phase where investors as a whole are over excited about AI? My opinion is yes. Is AI the most important thing to happen in a very long time? My opinion is also yes. So, you know, so, so Sam Alman's here basically suggesting we may be in an AI bubble, which is kind of extraordinary given rumors that he's expressed an interest in raising 5 to7 trillion dollars from from Middle Eastern sovereign wealth funds. >> Yeah.
Um, so it seems to me that either he's trying to make sure the market doesn't over steer so we can enjoy kind of healthy growth or he's trying to make it harder for startups that might displace OpenAI to get the funding they are looking for. >> So I was at OpenAI headquarters twice in the last 40 I spent about 20% of my summer actually in Silicon Valley just because everything's happening. But I was at the headquarters twice this summer and I can tell you exactly what's going on. I don't know how you want the truth truth. >> Please, please. >> Uh so here here's what's happening. Um Sam and the AI models are are truly self-improving. They're diverting a lot of the compute back into their internal research and internal self-improvement. They've demoed to the world some incredible capabilities. You know, Google V3, the video generation. Everybody was like, "Wow, >> can I have this?" >> No, you can't have it. What do you mean? like try and make a video. You have to wait like 6 minutes. That's becoming like 8 minutes. Like why can't I do it? It's like cuz cuz they're they're in enough chips. We need those chips for our internal research. And uh you know making your funny little video is not a priority.
So you think you you think more and more of the compute inside of open AI is being applied to recursive self-learning that so you you think that process of self-improving AI is beginning and that may partially explain why some people were unimpressed with with GPT5. >> Well actually uh very specifically you know the latest iterations from all the big model companies they're not running the model that you're signed up for. they're diverting to the lowest caliber model that's good enough to answer your question to try and save a bunch of compute. You know, because these things as your question gets more complicated and your context window gets longer, they slow down super linearly. So, it's it's a nonline. So, it's really expensive to answer hard, you know, super complicated nuclear physics problems or whatever. It's very very comput inensive. So, they're like, well, if you're asking about tomorrow's weather, we're not going to run that model. We'll divert it over to something much smaller. And so occasionally or actually in my experience quite frequently you get a dumber answer than you should than you kind of expected and you're like I was asking a pretty hard question there. Your model routed it to the wrong like can you try again and try a little harder. Actually you can often go back and reprompt it and say wrong you idiot. Try much harder and it'll come back and get it right. >> Interesting. >> So that tells you how it redirects to the to the more expensive model. But that's the beginning that's only going to get much worse because of the the chip shortage. And yeah, they're diverting a lot of that compute back into into self-improvement. Now, in fact, I I know from just talking to Noan Brown, who's like the the brainiac of all brainiacs at OpenAI, that their backlog of ideas to test is purely compute constrained. It's not it's not PhD constrained anymore. Like they they have a whole bunch of ideas that are highly likely to work and they can't get them to market quickly enough because they don't have enough test compute.
I'm long invidia. Should I should I be long invidia? This it sounds to me like like I should. I mean there's also I mean I mean you know Google and others are are are creating what appear to be very good chips but the intensity of the pressure on the chip manufacturing with these dynamics that you're describing particularly as it affects the ability of these companies to improve their products yeah >> is is pretty profound. Well, the the bet on Nvidia at this stage is a bet on Jensen's ability to use this massive capital leverage that he's got to continually acquire and improve mostly the interconnect, but the data center use, the cooling, you know, the liquid cooling. There's a whole bunch of things that there are other chip companies that are just as good or catching up quickly, but they don't have the whole data center infrastructure. So the Melanox acquisition, you know, which is the NV link, you know, that connects him is a brilliant move on his part. But if he keeps doing that and he stays mentally young, he can keep accelerating his lead.
The I I'll do the steelman side, take your side of the table for a second. What's the scenario where the bubble burst and we all lose a ton of money? The most likely way that could happen is because Nvidia is sold out for years to come, they can't actually make more chips. So, they can't beat their numbers by 2x or 4x again because they can't get TSMC to make more chips. Right now, they're they're commanding an 80% gross margin on their GB200s. That's unprecedented. Like, to maintain that kind of gross margin. No chip company has ever been able to do that over the long long haul. So, I I think they'll be able to do that for years because they're they're completely sold out. but where's the supply of manufacturing going to come from to allow them to blow out their numbers? So, that gives their competitors some time to catch up. So, the the disaster scenario is the one where they miss numbers uh because they can't get TSMC to make the chips and then people catastrophically run away from the stock >> because it is kind of priced to perfection >> and then that ripples through the whole ecosystem and the whole economy and then everybody panics and then it's like the internet bubble all over again. And even though AI is the biggest force of abundance in the history of the world, everybody freaks out.
Well, and as you pointed out, Taiwan semiconductor is is making most of these chips in a very small uh country off the coast of China. Right. That's the sovereignty of which is in dispute. So there's there there's a vulnerability there as well. >> Oh my god. Is there ever I mean I think if you want to do some good investigative reporting, you know, get your get your buddies to dig in. You know, TSMC is saying they're moving a lot of their manufacturing to Arizona, to the US. Um, I think it's not nearly as far along as as they claim it is, and that if if China invaded Taiwan tomorrow, uh, it would be a huge fraction of the chip supply would go to zero. TSMC says they can turn off the fabs, the Chinese can't just take the fabs and redirect them to China. We can control them remotely. Like, I don't know how much I believe that, but that's what they claim. But, uh, we cannot get, you know, the Intel fabs in Texas and Ohio and the Arizona fabs, those have to get up and running much faster. Uh, and yeah, like you said, China claims to own Taiwan. So, not a great situation. We we never should have gotten here in the first place, too. It shows you how asleep at the wheel we can we can be.
Let's shift to the topic of how AI is converging with other powerful technologies. Um because I think part of what makes this moment so extraordinary. I mean clearly AI is the sort of uber technology that is driving innovation right now. But it it overlaps in very interesting ways with other technologies that have been accelerating for decades. >> Uh and I and I think a fun place to start here is is with robots, right? which which um when people question the potential future economic impact of AI uh when you look at the size of the labor market and if you believe that what's missing from the ability of robots to do just about anything a human can do is is this kind of very sophisticated granular dexterity and movement that would appear to be exactly the kind of thing that neural networks are good at at at learning how to do if you have powerful enough AI embodied in these robots. So, is that the way you see it? And what do you think is likely to happen in the next?
It's exactly what you said and then more. Um, the the self-driving car is a really interesting case study on this because when when Elon started working on the self-driving car, it was almost all CC code with a little bit of neural net for the vision processing. And then as it evolved, they're like, well, we don't need as much C. Why don't we use more neural net? Now it's over 90% neural net, which means nobody programmed it. It just learned to drive by driving. And then it's down to 10% soon to be 0% handwritten CC code. So that means it's a completely learned system. And it's the exact same neural net that you use in the robot and it's the exact same neural net you use to do your physics problem or to write your English paper. It is 100% the same exact code at the core. Which means we now have this central asset. And this is why when Mark Zuckerberg says there is no amount of money, $600 billion, fine. There's no amount of money that I won't spend to win this AI race. Why? Well, because it's this kernel is universal. It's used for all of these things.
So I was at 1x Robotics this summer. Um, they had the really, really cool uh robots for inside humanoid robots for inside your house. soft skin, so they don't hurt you, but very strong. They weigh 70 lbs, but they can carry 70 lbs. So, they're really, really strong, helpful vacuuming and whatever robots for your for your home. What what blew my mind is, they they're getting more nimble and more capable every day. >> And they learn and and improve without any human involvement. Like, every day they get better by themselves, which is so cool. But the voices and the personalities are perfect already. That really surprised me cuz you know when you see C3 you know like in in Star Wars you see C3PO or whatever you know R2-D2 can't even speak for some reason like I don't know why that would be but C3PO C3PO is kind of robotic and that that's the way you always portray a robot it's kind of like clunky in it speech the actual real world is they have perfect voices like perfect radio voices like you and their their knowledge and their empathy is is incredible and they're still a robot. They're still like you know you know bumping around a little bit. uh but they're going to get very very nimble very quickly and you know then they'll be again superhuman in their dexterity very soon and it's all the neural net just improving >> and so can we anticipate hundreds of millions of humanoid robots I mean there's a you know there's a question of I mean do they begin as humanoid robots and then become more specialized but the the implications of this I think let's see what do I have here Morgan Stanley predicts that by 2035 there could be over 930 30 million humanoid robots performing structured repetitive work. Does this sound realistic to you?
Yeah, that that number in 2035 sounds what was the number? 130 million. >> 930 million. Almost almost a billion humanoid robots >> by 2025. So the number that's disappointing is if you look at 1X's forecast and Tesla's forecast, it's going to be on the order of several hundred thousand robots two years from now because he just can't get the supply chain and the manufacturing up and running fast enough, but it's on an exponential curve. So could you get to almost a billion by 2035? And I think the answer is probably yes, cuz somewhere in there you get the robots making the robots. So, I think that supply chain will get spooled up, you know, over the next four or five years. I I think people have the timeline a little messed up in their head, right? Because, you know, they're thinking about self-driving cars, that's soon. EV tolls, that's going to come. Uh they're thinking about robots in their home and and, you know, maintaining everything, and that's going to come, too. But the AI that's beyond super intelligent is just a couple years from now. That's going to be much much sooner. So your your laptop being a genius is going to come two or three years before the robot is available to come into your home just because of the supply chain and manufacturing right order of operations.
So um yeah, I'd have to think I don't I don't know if that number is viable. That seems like a but it's it's it's technologically easily viable. It's just a question of whether we can get enough parts to get the price down to get the you know get the whole manufacturing thing spooled up. I I just got a note from our producer who told me that he mistyped. Morgan Stanley actually predicted 930 million plus robots by 2050, not 2035. So maybe so that that that might make more sense, but but for people listening, I mean it does seem like think about what you said 950. So almost a billion robots by 2050, but there's there's almost by then there'll be what? 7 8 billion. Well, there's already seven billion people on the planet. That means only one in seven people get one >> in 2050. That's right. >> Yeah. So, this is not going to be in your house as soon as everyone might want.
If you were to bet personally, you think you think that you'll have one of these in 6, 8, 10 years? >> Oh, for sure. You know, I love to drive around suburban neighborhoods around New England here and you look and you look at the fences and or look at the paint on the shingles and you count what fraction of all fences or all shingles need to be repaired or painted and it's like 80%. You're like, well, why don't why don't they fix their fence? It's like, well, it's expensive as all hell. We'll just let it rot for a couple more years, then we'll fix it later. That's a perfect robot. Like, there there are so many jobs for these robots. And and uh they they just tireless, man. They they they solar charge and then they just work.
That will be gamechanging. And obviously that leads to questions of, you know, labor markets and unemployment, but also of like how radically different the world may be, whether one thinks it will take 10 years or 20 years or 30 years, right? This notion that all human physical labor or the vast vast majority will have the option of of doing all that labor at a very low cost with clean electricity. The way that would change the world and people's experiences and whether it's you know 10 20 30 years is astonishing and and the productivity impact of that economically. So, do you think we're that we're talking about more like, you know, 20 year time frames and and do you see this as one of the more profound ways that our economy and our and our physical lives like h how they will change is going to evolve?
No doubt. I think there are two things that are absolutely profound about the not too distant future, certainly sooner than 2050. It's healthcare potentially getting to escape velocity and living forever, which seems likely actually. Uh, and then it's this experience where >> I don't know if you've ever been to the Breakers down in Palm Beach. It's like the most expensive hotel in the freaking world. >> Yeah. >> And you get to this place and you're just like, "Wow, where am I? Holy crap." >> And that experience, you know, very few people, you know, very few people in America get to have that experience, but then globally it's like, you know, 0.001%. That experience can be universal. you can build these like gorgeous, spectacular experiences for literally everybody and it can be dirt cheap. And I I really believe that's imminent uh and extremely possible. So you mesh those two things together, it's like, wow, I really want to live forever because this is awesome. And what do you know with medical breakthroughs, I can live forever. As long as I don't as long as I don't make a mistake, you know, I I can I can get to longevity escape. So from from a sequencing perspective, your thought is let's first make living forever desirable by improving the quality of people's lives and then let's make it medically feasible and before that like I think what the what the big labs are doing right now is uh you know reinvesting in self-improvement. That's absolutely the right thing to do because that's the unlock for everything else. So if we have to wait a little bit longer for our virtual girlfriend or whatever the hell it is uh that's okay. we should be redirecting all of that compute back into self-improvement. So then those AIs can work on the designs for the robots and can work on the designs for for healthcare uh you know the drug drug discovery and the medical discovery all that thinking will be done by that self-improving AI that is imminent. So >> in fact if you know if we could be a rational society and go back a hundred years we would have invested a lot less in physical things you know buildings in Manhattan and and we should have put all that into discover the computer sooner >> then accelerate the GPU sooner and then research AI earlier which you know I >> I think you know people who've worked in AI since way back in the 80s and 90s have thought that way for a long time and now I think the world's kind of woken up to yeah that would have been better because you know on the order of half a billion people are going to die unnecessarily because we can't get our health act together quickly enough because we didn't get AI done 5 years sooner. The health implications as you say may be the most profound and I've been fascinated to hear we've obviously had like Ray Kerszswhile and others talking about you know uh mortality escape velocity and and and living forever for for many many many years. But for me to hear people like Daario Ammedday, you know, who runs anthropic and I think was a biologist and a physicist and is very soberminded. He's very soberminded about all the risks of AI, but also, you know, talking about the benefits of potentially doubling the human lifespan in in in a period of 5 to 10 years. and Demisabus another, you know, who runs obviously Google Deep Mind, as you know, another player in the space who I think of as being less of a salesman, right? Less hyperbolic. He's fundamentally a scientist in terms of his orientation. He's sober minded. He's not he's not trying to sell you something. He's saying like this is what's coming down the pike. And both of these guys are are saying something very close to what what you've just said, which is in a 5 to 10 year time horizon, we could expect to see a majority of human physical ailments solved.
Well, so Dennis is a god. I mean, he's just he's just the most uh he's exactly the way you just described him, just awesome guy, but he's actually softelling what he really believes is going to happen because he's, you know, he's a Nobel Prize winner now. he needs to be, you know, circumspect and not overpromise and underdel, but he actually thinks it's going to be a lot more a lot sooner than what he's saying. >> Same is true with uh, you know, we were talking a minute ago about Sam Alman, you know, being cautious all of a sudden after hyping the heck out of this thing for years. Why is he being cautious? Well, because he can't get enough chips to deliver on the on the promise. So, the last thing you want to do is, you know, he's got all the money he needs now, so he doesn't have to hype it. And he's got pickers out in front of the building constantly. you know, people like AI is going to kill us all and it doesn't need any more of that. And so, it's actually good strategy for all those guys to kind of downplay expectations right now. But when you talk to them personally one-on-one because, you know, Dennis was in Davos with us last January. We had him in our in our dome as a speaker and he he's incredibly optimistic about the full cell simulator. There's about a million chemical reactions going on in every one of your cells at any given time, but you can simulate all of them using the neural nets are incredibly good at this stuff. and and the concept of living forever. If you walk down the street and you say that to a hundred people, there isn't you're not going to find one out of a hundred who say, "Oh, yeah, that's going to happen." But the math is really compelling cuz if you look at the rate that AI is improving, you know, two years is like forever. Well, how much medical technology do you need to add two years to your life expectancy? Actually, not that much. So if you can stretch your lifespan by just two years, but then the AI improves tremendously during those two years, you'll probably add another two years. >> So there's a there's a date coming soon where we can all hit escape velocity, which effectively gives us infinite lifespan because the technology is improving faster than we're aging. That's not that outlandish at all, but very few people have gotten their head wrapped around it and are like, you know, working out and eating like you should be preparing for this cuz there are a lot of scenarios where you're kind of locked in time. You're not going to get younger, but you may not get older either. And so, >> yeah, >> being physically as young as you can be on that day is a very good investment.
[Music] My marathon isn't about time. It's about my 26 friends with intellectual disabilities I'm dedicating each mile to. I'm Joseph. I'm running my first Bank of America Chicago Marathon for Special Olympics Illinois to help give my friends a place to shine. >> Join Bank of America in supporting Joseph's cause. Give if you can at ba.com/support Joseph. What would you like the power to do? Bank of America. References to charitable organizations is not an endorsement by Bank of America Corporation. >> This episode is brought to you by Progressive Commercial Insurance. Business owners, meet Progressive Insurance. They make it easy to get discounts on commercial auto insurance and find coverages to grow with your business. Quote in as little as 8 minutes at progressivecommercial.com. Progressive Casualty Insurance Company. Coverage provided and serviced by affiliated and third party insurers. Discounts and covered selections not not available in all states or situations.
So turning to this question of of practical decisions people can make in their own lives given these changes that you see underway. Let's start with education and how we think about for for folks listening who have who have kids in you know in school. Is our education system evolving quickly enough?
Are you kidding me? >> Another another softball. There you go. How about at it? >> Holy crap. Actually, I was meeting with a nameless uh administration official, one of the top people at MIT. And he said, "You know what? We will build a nuclear reactor here in Cambridge, which they're already working on, like an exothermic real nuclear. We're going to finish that nuclear reactor before this curriculum gets updated by a year." Like, what are you talking about? How can you be that cynical? I was like, "Look, look, like look look at the physics class. It's the exact same textbook from like 90 years ago. There's no way there's no way we're going to be changing this. We we're not set up to keep up with an exponential curve. We're barely set up to keep up with flat, you know, but maybe a line. We'll give you a line, but we're on an exponential curve. So, the amount of knowledge that's useful is accelerating. The curriculum is not even vaguely useful relative to what's new information that you should learn." Now at the same time very soon AI as a interactive teacher is going to be nearly inf like a TA today it's dramatically better than any TA I ever had already. >> Y >> and so very soon it'll be better than any professor I ever had like within a year. And so if you're self motivated and you want to learn the problem you run into is most people go to school for credentials. They don't go there to actually learn.
And so how's that going to play out? And this is why things like the teal fellowship and you know link studio our our venture studio but also Y cominator those credentials now actually are starting to matter just as much or maybe even more to people than their college degree and if that trend continues then you're going to end up with credentials are over here learning is AI what's the role of the school and that's going to create a little bit of a crisis I'm sure but I think that's the way it's going to play out and you know my kids already learn far more talking to AIs than they do in their curriculum. So, >> yeah, >> I I have three teenage boys and I I apologize to the 20-year-old for referring to him as a teenager, but in my mind, they're still all still teenagers. And my 17-year-old son, who's about to apply to colleges, including MIT, was was saying like he's like, "Yeah, you know, if I don't get into one of my top three schools, I think I'll just go start a company or go work for a really innovative company." And he made the comment like, "What is the purpose of college? Maybe it is primarily credentiing so that if I can if I can get into the school then the purpose of continuing to stay at the school becomes less important. My counter was actually the people you meet you know and the community you build and the friendships you build are can be transformative in your life. What would you say to my 17-year-old son?
I'll tell you exactly what I would say and I'll say it to you right now. Um, if you don't get into one of those top schools that you're really, really desperate to get into, defer for a year for sure. Um, we're working on a bunch of programs in Cambridge where you can get onto the campuses even though you're not a registered student. We've got a lot of little things we're working on there for exactly that scenario where you're trying to start a company, you're trying to meet people, you're trying to network in, come into Cambridge, or you can do it, you know, in San Francisco or Silicon Valley, too, or where wherever, but there's only a couple hotspots where there are lots of schools. Cambridge is phenomenal. But setup shop there. Uh we just bought an apartment building for exactly this reason. Actually, we've got six six units, four bedrooms each for exactly this, and it's right next to MIT. was the only it's the closest building to MIT that wasn't already owned by MIT. Harvard and MIT are basically taking over all of Cambridge and then you can network into all the students from there >> and then make a bunch of friends then apply again the next year and if you don't get in don't don't worry about it maybe your company will take off in the meantime you don't care or maybe you develop a lot of credentials that go onto your application it looks that much better the next time and you know defer for two years three years but then eventually get into one of those hotspots but what you don't want to do is go to you know Hamilton New York or you uh I won't name sputs, but you don't want to go way out in the BBS to some school that wasn't exactly what you wanted and then because then you end up there for a long time for four years and you also meet a bunch of people that are all going to scatter all over the country and you can't start a company with them. And so there are many ways to live your life. This is just one game plan, but that's exactly what I would tell your son.
Right. And and ecosystems, as you point out, are so important and for all of us, like the places we choose to live. I mean, for me, New York City has been I think I was almost a co-founder of all the companies that I've built. uh because it's been this ecosystem that's been that's been wonderful for for for me. But if you can I think it's the ecosystems of the universities that are powerful in ter for many students and and you're suggesting that there are other ways and certainly collectively we will find other ways to build ecosystems for kids. I mean given the colossal expense I mean not to obviously you know I think it's it's an incredibly beautiful luxury to you know go to a great college with wonderful professors and don't want to throw shade on that but I think you're delivering a very specific message around there are specific windows of opportunity historically when when we look at technology and historical opportunities to do new things that are powerful we are currently in one of those windows >> and you should be aware of the specificity of that. You should be aware of the importance of building companies with friends and placing yourself in an ecosystem where you can do that effectively.
Yeah. That that building companies with friends aspect of it. Really, really think about it thoughtfully cuz people are very reluctant to say, "Look, I don't choose my friends because they're co-founders. I don't choose my friends because they're they want to make money. You know, I don't choose them that way. That's that's crazy." Like, well, I can tell you it works. And and you know, you'll become friends with the people you're with. Uh, so change the environment you're in. You're going to make friends with the people you're in the environment. It's just just human nature. Uh, and so if you get yourself into one of those really dense, talent dense moments >> and places, I always challenge my kids to say, give some thought to how you made that friend because you always forget. You take it for granted. Oh, we're best buddies now. Like, well, how did that happen? It takes a lot of time. You have to be pushed together. You don't just randomly call someone and say, hey, you want to go hang out? Sounds kind of weird. like no no that doesn't happen. You're either on a team together and that pushes you together or you're in a class together and that pushes together. You're in a club together or you're sharing a your roommates and you're sharing a you know room together that pushes you together. there's some force that compresses you and if you can synthesize that like so if you don't get in and you want to go just hang out in Cambridge or hang out in New York be very close to it but then get into those bonding opportunities with those people and it'll be the same net outcome that you would have had as if you were a student.
So let's say that our our young entrepreneurs have now started a company with a few friends. You advise dozens of startups. What counsel do you have for people who are have already begun building companies? What what sets apart the successful ones?
Number one is what you said, get into uh an ecosystem where everything is growing, everyone's thriving, and and no one has any time. So, they'll just buy bye-bye. You know, whatever you're creating, they'll buy bye-bye. And if you know, very often, you know, people will have a preconceived notion that I want to build something, I'm going to sell it to banks or I'm going to sell it to universities. Their decision cycle is so slow. Uh so you don't want to be in those kind of slowmoving worlds. You want to you want to be in a fastmoving world. So so so number one is ecosystem of other fastmoving companies, people ideas that can all buy and sell from each other and and interact together. Second persistence you know if you look at the CEOs of tech companies Satia Nadella Sundar Pichai uh my friend Sid Banerjee Sanju Bansal what jumps out at you about all those names? Every one of those is an Indian name. >> Yeah. Do you know what? Do you know what fraction of America is ethnically Indian? >> It's got to be really small. >> 1.5%. >> Wow. Yeah. >> Wh Why are all these tech companies run by Indian CEOs >> and there's a lot of reasons, but a big one is culturally their parents are entrepreneurs. All of India is a bunch of entrepreneurs. Interesting. >> And they love it. They don't they don't vomit on it. M >> and so if your parents are bankers or they're you know you know third generation car dealers whatever and you say I'm going to do I'm going to take a flyer I'm going to do a new startup it's so easy to demotivate your kid but the Indian community loves entrepreneurs and it's it's self-reinforcing it motivates them and and the work ethic is incredible so if you can instill that in your teammates and say guys we got to we got to be gritty survivors working our asses off I literally the failure rate is near zero if you have that grit and focus and you don't get, you know, disassembled by somebody getting a job offer from McKenzie. So, as long as you have those dynamics, it seems to work every time. And then the other thing is play with all the AI tools. You know, whether you consider yourself an AI person or not, it doesn't matter much right now. Anyone can write code. You can vibe code anything. You know, use Lovable, use Replet, but don't be afraid to try all these things. They're so engaging and they're so eye opening. So, you know, the more things you try, the more your ideas will be like right in the sweet spot of what's actually happening right now. Because I I run into, you know, a lot of people who have an idea that they pull up out of theory land, you know, in a dorm room or whatever, and they're very often related to like apartment search. You know, apartment search should be much much better. Well, why do you think that? Well, because I'm looking for an apartment right now. Okay, you got very li limited view. Like, expand your mind to all the opportunities. You're going to find much lower hanging fruit. So, you know, blow it out, man. Get get exposed to a lot more ideas. Get yourself in an ecosystem, an incubator. There's so many great uh incubators now, incub incubators and accelerators. When I graduated, none of this existed. Now they're everywhere. Get into one.
And what you just said may give some hope to some of the middle-aged people listening who I'm guessing are the preponderance of people who are listening to us right now, which is to say that that the um I mean, you know, and I think a lot of well, I think I have these teenage boys rising. And so I think I think we think about our kids, but but also I think a lot of people are asking themselves, well, well, h how should I be navigating my own career? Um there's actually a fair amount of data showing that a strikingly large number of successful entrepreneurs are actually in their 30s, 40s, 50s, 60s. What kind of counsel would you give to people who are mid-career?
Same thing. Jump in, play with all the tools. Don't be afraid to quit your day job because you need to focus like crazy. And then if you if you rewind, you know, to old information from 1990 to 1995 to 2000, you'll see like me and Mark Cuban and a bunch of young guys, you know, David Pho, >> but you'll also see a bunch of guys that started other internet companies like Vignette, Interwoven Broadvision, and a bunch of others that were enterprise or had domain knowledge that the young guys just had no clue. And they succeeded like wild, too. So the correl like study those use cases first. go pick up some bios and study some people. But then um think about okay what specific domain knowledge have I accumulated by age say let's say I'm 45 and I'm in a in an industry in a market. AI is going to come into that market. Now the 20some year old will know the AI inside and out but won't know anything about that market. How can you bridge that gap? Create your own AI company that takes advantage of your domain knowledge. you know, very often it's either sales relationships or or regulatory relationships >> and the the youngster will benefit tremendously. So, you either set it up as a separate company and work with the young company or just work with young founders. Either way, it works fine. That is like a a really really good road map for how you get involved. And the same core message, you know, be fearless. Don't be intimidated. The the young guys use all the terminology in the world and it all blows over, you know, it's just the terms of the day. Don't be intimidated by that. It's, you know, oh yeah, the 2017 Transformer paper, like, you know, the guys at MIT are all wearing the the text of that paper on their shirts now, the back of the shirt. So, it's like, okay, don't be intimidated by the cultural aspects of it.
How about disrupting uh from within the company you're working for or innovating as an employee inside of a company today? I mean you you've spoken in interesting ways uh in the past about about the differences in different sectors right now like in medicine we're seeing incredible acceleration of AI it's maybe different in law or insurance or other sectors what what are the variables that you think are determining where AI is most effective today
you know, it's interesting I I was talking to a very senior executive of Barclay's Bank uh who's based in San Fran uh this is just this week and he was insanely frustrated with their headquarters in the UK because he wants to be AIing everything tomorrow. And that to me is a no-brainer, right? Because one, it's going to work, but two, it's going to give your team exposure to AI and they'll know how to do better banking, better IPOs, better whatever because they have the exposure. So if you're in a company and you're like, "God, we got to get on the AI bandwagon here." And then you go to the people around you, the boss or whatever, and they're like, "No, go back to work. >> You just got to find another job." Like, like find an AI forward equivalent career path, you know, same stay stay within your career path, but find the AI forward version of it. You'll probably find it in either the West Coast or in Boston or maybe New York or >> but find that version of it. and get into like-minded souls that are open-minded about getting on the bandwagon because I know I know you know in some of the companies I'm involved with. It's agonizing, you know, if you're the you're the person who's vibe coding at night and everyone around you is is a denier. >> Yeah. >> And so you get trapped in that environment. Uh you just got to move on. Like get away from those people, find new people.
And so so turning to investments, I think a lot of people hopefully are putting some money aside. I I've I've heard you uh uh speak about um tech ETFs, NASDAQ 100 ETFs, as from a from a marketplace perspective, you find it sort of not the most effective way to deploy capital when we're looking at elevating companies. But if you're if you're an average Joe who's looking to to build build savings for retirement, the NASDAQ 100 has done pretty darn well in the last in the last 15 years. It's hard it's hard to see that that's not going to continue or accelerate. Would you would you disagree with that?
It's turbulent, but it's it's a I mean, look at it over the long haul. It outperforms everything over the long run. So, yeah, it's it's >> it scares people because of the turbulence, but if you're thinking about retirement, yeah, that's a no-brainer.
I think within that, too, uh you said, I'm going to take, you know, a bunch of stocks. I'm just going to sort them into two buckets and maybe pick the top 10% of the management teams that you think are thinking about AI for real and just put your money in those and then, you know, don't put your money in the other 90%. I think you're going to kill it.
Very quickly, lightning round. We're not giving any investment advice here, folks. This is not investment advice. Everyone, everyone should make their own decisions and uh seek their own counsel. But what do you like? Uh it sounds like you had a question mark on Nvidia. Google would seem to be a pretty safe play in this in the AI space. What do you think?
Yeah, I love the Google TPUs. Google's a quandry because their core search business is a victim of AI, but they're also the AI leader and the TPU chips are incredible and >> uh so that's a little bit of a of a quandry to me. Um >> is it that is Apple in trouble? What do you think about Apple?
I I can't figure out Apple. I mean, I love the company. I've been a huge Apple fan boy. It's the first stock I ever bought in my life when I was a little kid. Yeah. >> Um, and I love that company, but how they could be this far behind and not realizing the AI moment. It's nuts. And now their game plan is, oh, we're going to fix Siri. Like, come on, man.
How about Bitcoin? >> So, uh, I I am long Bitcoin. You know, Mike Sailor at Micro Strategy was my roommate as a MIT student and one of my my favorite guy to talk to of all time. Just absolutely brilliant. And he thinks Bitcoin is the fundamental store of wealth for the alltime future. At the same time, it's also very valuable as a currency for AI agents, talking to AI agents. So, I love Bitcoin. Uh, you just have to, you know, be ready for turbulence along the way. But, yeah, also, you know, there
There are a lot of countries in the world that have don't have a currency you can trust at all. The economies are, you know, Venezuela types. Ultimately, they're going to rebel and go to Bitcoin and they'll work for other overseas countries and get paid through Bitcoin and just get out of their local economy. It's so corrupt. So, uh, so yeah, I I do like Bitcoin a lot. I'm I'm not a big fan of other cryptos other than transactional cryptos that, you know, for agent agent transactions, those are those are really cool.
>> Yeah. Yeah. Well, so this brings me to a topic that that uh that for me is uh a little more challenging. I I think a factor that's off-putting to many people is this notion that we're on the verge of an intelligence explosion that some people refer to as the singularity. It has an almost messianic sound to it. Right. This notion we're about to hit an inflection point after which we'll live in a utopia of super abundance. Everything will be free essentially. We can all do what we like. Or or some would say that this could be a dystopia. The robots will take over. I think that sounds far-fetched to many people at first because it's a it's just so divorced from the reality that we know and understand. But second that it has that that there's a pattern that we've been familiar with of people believing in either messianic perfect futures or apocalyptic dark futures, right? But having having said that, we do appear there does seem to be a sizable percentage of people inside the AI space who who argue that we're going to hit this recursive learning inflection point and that there's a high probability that we will we will see a dramatic hard takeoff of capability of these of these systems. So So where do you fall on that?
>> I've been thinking about the singularity since I was 14 years old. So you're asking the right guy. And now I think a lot about how it's going to actually unfold. So it's clearly going to happen. But the way it was described by Ray Curtzwhile, you know, you hit this this singularity, this inflection point and then it goes to infinity because it becomes self-improving. And once it's self-improving, there's like, you know, a year becomes a millisecond and it just goes infinite. It's not actually going to happen that way. And we talked earlier about like there's all these constraints related to parts and supplies and robots and it's not going to happen that way. Here's what's actually going to happen. Imminently, like very soon, the AI is going to be as intelligent as an AI researcher or maybe even more intelligent, but there will be hundreds of millions or billions of them. So, the idea flow will go up through the roof. Then all those ideas will be tested on the compute. That's why the comput is so severely constrained. And I think we're looking at at least a 100x and up to maybe 10,000x improvement in the software alone. So Nvidia chips stay exactly as they are, but the AI is now suddenly in a very short window of time, like a matter of a few days,
>> 100 and then a thousand, then 10,000 times bigger,
>> which means probably smarter, you know, maybe not proportional, but it's much much smarter, superhuman intelligent in virtually no time. Mhm.
>> And then it gets stuck because now it's out of, you know, it's like it's tapped out by the Nvidia hardware. So then it goes back to slower growth as it tries to improve its own chips. But those chips need to be manufactured and that takes time.
>> And so then it's not actually a singularity. It's a step function and then some slower growth rate
>> that's very manageable and very predictable and very understandable. And after that step function, uh the virtual world is going to be incredible. you know, videos, all kinds of software, automation, but the hardware is going to lag like crazy. You know, the the paradise of buildings and lawns and, you know, country clubs and all that is going to take time, but it's going to be dirt cheap and very doable. So, it'll it'll fill in over the subsequent, you know, 15 to 20 years. Uh, the health improvements will come much sooner because manufacturing drugs is actually easier and faster uh than, you know, building buildings is. And so, it'll all unfold. I'm trying to map out the exact dates, you know, as it as it unfolds. But the singularity is that first step function. The world right after the singularity is going to be very different than the day before. It's going to happen very soon.
>> And when you say very soon, you think in the next several years, five years.
>> Yeah, definitely less than five years. Uh if I had to take a wild guess, 2026 is the year. I've always had major events in my life on the sevens, so maybe it's a 2027 thing. Uh but it's in that range. It's 2026 or 2027.
>> And what is and how does the world look look different after that?
>> Well, I mean the biggest question I have is, you know, a lot of power is going to go to Mark Zuckerberg, Sam Alman, Elon Musk. It's a huge and and Sergey and Larry too. So a massive amount of power, wealth immediately accumulates there. And so a lot of the future of the world depends on what they choose to do with it. Um I'm optimistic, but it is very concentrated power and wealth. Government is completely unable to keep up with this. It's going to be a disaster area of like out of touch, which it always has been. So, it's just going to be more of the same. So, then the question is, are more of these guys going to end up in government? Are they going to effectively choose the next like you saw what happened in the last election? You know, did Elon Musk choose Donald Trump to be president? Then it backfired on him almost. You know, it's not quite that, but the next election will be pretty much guaranteed
>> that that's the story, right? like the the AI will control the election outcome because the media is going to be all AI generated. So is that messed up? Potentially it could be terrible. Could it be great? Yeah, it could be great too. It depends on the intentions of those people. So this 100x to did you say 100x to thousandx dramatic improvement in in capability of these AI systems. After that inflection point, we can basically solve all math, all I mean this is the inflection point that we have all these leaders at these companies saying this is when we could potentially cure all disease,
>> right? Uh extend the human lifespan, endless free energy, right? Like we're already making real progress with cold fusion and these, you know, different technologies. But once you have endless energy, you can also then start to, you know, correct for the atmospheric challenges resulting in global warming. you have endless robots. I mean, we have B basically we have kind of infinite capability, right, to solve all the problems that that are afflicting humans or constraining humans. But
>> only thing I'd push back on there is that that if you as soon as you say infinite, then you give up on trying to actually get the detail.
>> Yeah,
>> it's not infinite. It's tremendously big. But it's worth noodling through like if you want to invest or benefit from it's worth like really dig through what will and won't happen cuz it's not infinite. It's mind-blowing, but it's it's actually contained and bounded. And it's it's a it's 100 to 10,000x. I could walk you through the the details of what causes that, but it's not infinite. You know, there's a big difference between 10,000x and infinite. And it doesn't need to hurt humanity if we if we put a lot of effort into really predicting what it will and won't do and how to contain it and all that stuff,
>> right? But is it necessary to like we you know we we've had you know many guests on the show and and folks like uh I'm thinking of Stuart Russell who who runs uh Berkeley's AI lab right and he's and one of his comments is if you look at the history of the relationship between more intelligent species and less intelligent species typically it hasn't worked out very well for the less intelligent species right so there so there's one question of just controllability and right now I think that like Daario Emmed and others would say the lack of interpretability like we don't understand exactly how these systems are working.
>> If you I if they hit this dramatic acceleration of capability that we're calling the singularity, is it in their interest to divulge how much smarter they've gotten? Right? Who who who's running the show? I mean I mean to to me that is a profound question.
>> It is a profound question. And this is where I diverge from most of my uh podcast mates and and and AI research mates cuz they they're all very eager to give it a sense of self-preservation, a sense of consciousness. I'm like, we do not need that. Like this thing can be super intelligent, super human in every conceivable way, but still be a feed forward neural net that has no desire to survive. It doesn't need that capability to solve all medicine, to construct all buildings, to to create a human paradise. It absolutely does not need consciousness and it doesn't need uh self-preservation. But you know the reason species that are smarter have generally crushed other species is because they have this this survival instinct from self-preservation from evolution. AI doesn't have that and doesn't need to have that. There's no need for it to have that. But the researchers are so eager to add that. I think Daario's got his head screwed on right on this. Yes. He he doesn't feel like it needs it, but a lot of the other community is like, "Yeah, yeah, it's got to have that like it has to have rights." Like, no, it doesn't. It does not need to have rights. Absolutely. It doesn't even make sense for it to have rights because they're going to hundreds of billions or trillions of them instantaneously. If they have votes, then they have all the vote. It doesn't make any sense.
>> So, anyway, I got a lot of thoughts on that, but
>> yeah. Yeah. Yeah. Well, I I I certainly uh hope hope that we can steer this in the right direction. And I think as you say the the people in these positions running these companies are very important who they are. And we talk about I mean we had Reed Hoffman on the show. You you all have spoken with Reed. He expressed real confidence in kind of the fundamental kind of prudence and good judgment of the people across the board who are running these frontier model companies. So I
>> yeah I that's why America needs to win this race because I I the ones I've met which is about half of them are great. And you know if you have to make a bet on humanity surviving these are great these people are great choices. I don't know about China and that's like these are politicians you know that control their AI region. I don't know I don't feel so good about that one. So very good idea to win the AI race and then and then that's not enough. Then you have to figure out how does the global order like become stable. You know there's a lot to think about there.
>> There certainly is. Well, we're certainly I'm certain that you and I will do everything that we can to to make sure we have the best possible outcome here. And there's certainly a lot to be excited about. Thank you so much, Dave, for your insights and time with us today. Just a fascinating conversation.
>> Oh yeah, my favorite thing to do. Anytime you want to talk further, just just give a shout. This is just so much fun. Energizes me all day long. Keep me up all night writing code, man. Amazing. [Music] Would you like more technological prognostications from Dave Blondon? You can catch him every week on Moonshots with Peter Demandis, a show that I can't stop listening to. If you want to hear some of the other episodes we referenced at the top with Bill Gates, Reed Hoffman, You've All Know Harrari, check out the Spotify playlist in the episode notes. Today's episode was sound designed by Mike Tota and produced by Caleb Bissinger. Are you still on, Caleb?
>> Yeah, I'm still here.
>> Caleb, what do we have coming up next week?
>> I am going to be chatting with Nick Thompson. He's the CEO of The Atlantic. He hosts a wildly popular LinkedIn video series called The Most Interesting Thing in Tech. And he's the author of a forthcoming memoir called The Running Ground: A Father, a Son, and the Simplest of Sports. Can't wait for that one. Hope everyone will join us then.