The Tech Quotes That Defined 2025

27:13Everything I do is recommended by ChatGPT, and then I check with my doctors for safety. I always ask them if they disagree, and if they disagree with ChatGPT, I ask another doctor. Jesus. That is Vinod Khosla on how much he uses AI for personal health decisions at the Deus Ex Medicina Summit, another newcomer co-hosted event. This was in September. Yeah, I felt like, you know, Vinod was laying it on thick there for how much he's using ChatGPT for health benefit. And that was, of course, the theme of the conference that day. But, you know, early investor in OpenAI, he's talking about OpenAI. It also has become a massive use case for OpenAI for people to check it for healthcare questions.

27:55And these startups like OpenEvidence that are going gangbusters right now, raising tons and tons of new funding, seem to be basically ChatGPT for doctors. So while I see the benefit of that, you know, it wades into this whole question of how much should we be using AI for healthcare, but that is clearly a use case that has evolved this year. Yeah, I think doctors, we should have some on the show at some point. I'm interested to see how it's changed their job, not like in how they diagnose patients, but how they deal with questions. Oh, I have a good doctor friend we can have on. Yeah, like how they deal with questions from patients, you know, because like I'm sure like the advent of WebMD, which is, you know, like the hackiest joke from the late nineties of like, I went on the internet and found I had cancer kind of thing.

28:36But I bet that did really change the medical profession, uh, in how you interact with patients. And this is probably only accelerated it. Well, I think patients come at the doctor's appointment being like, I put my symptoms into Chachupi T and it said this, this, this. So I think I have this and the doctor has to, you know, respond in some way, but maybe the doctor's also using Chachupi T to check the symptoms as well. So it's an interesting Chachapiti to Chachapiti happening in the exam room. The, uh, yeah, I mean, it's a first having a baby Chachapiti is the first line of defense in is this medical thing normal or not?

29:11Um, uh, there was a brief second where we were worried that open AI was going to crack down on health stuff, but I think it was a sort of false rumor. And And yeah, like you said, I think it's a use case they're really leaning into. I have problems with the AI industry talking about chat GPT diagnostics as a revolution, as something that's going to truly change the way people receive medical care. I think it's such a poor comparison, a poor stand-in for actual medical care and like a professional. And I know they do talk about this a lot, but like we have major issues with the way people receive medical treatment in this country and access to health care and open AI and chat GPT.

29:57It's like a drop in the bucket of improvement toward what actually needs to happen to give people better quality of life. So I. Well, a lot of what we need is humanity. I mean, you know, you go to a hospital. You know, obviously my wife gave birth this year and nurse to nurse performances feels much more about bedside manner than medical knowledge. And in some ways, I hope ChatGPT sort of makes that even clearer. It's like, okay, if we have easy to access resources for the actual medical knowledge, then we're picking and elevating people based on their bedside manner more and more than necessarily how much they can retain medical knowledge.

30:41I'm interested in how AIs will change people's approach to bedside manner. And if it's going to cause people to become more argumentative with their nurses and doctors, because these AIs are trained to please you and to back up your questions, confirm your suspicions on things. And so if a nurse or a doctor is telling you one thing and you put it into chat GPT and it's like, no, you're right. You know, don't trust the medical professional industry. You know, you should be willing to question professionals when it comes to your health. But it sounds like it's just going to create more antagonism between people in there and their doctors.

31:21But doctors, I think, are embracing it more more than you seem to think. Like, I don't know. We'll see. We should do an episode on that. Quote number 10. If you were 13 years old, you should spend all your time vibe coding. That's how you should live your life. That's Alexander Wong on TVPN in September. Obviously, he sold his company scale this year for a bazillion dollars. Yeah, obviously, he was around, you know, a$13 billion exit right there. So good for him. But yeah, well, I picked this quote in part because one vibe coding obviously took off quite a bit this year. We've all tinkered with Lovable and Cursor and the various vibe coding tools on this show and the newsletter.

32:02And also I think it's very funny to hear Alexander Wang's dictating how 13 year olds will live their life in the future from now on. I think there's a good Silicon Valley culture of like, you know, we got to tell the youth how they're going to change their entire education always. And now it's vibe coding. So is he a college dropout? He has that college dropout profile. Yes. Right. I think I don't know if he's a college dropout. I think he might have graduated from college when he was 19. I think he was some sort of prodigy. Yeah, I think he's a prodigy. He was a math prodigy. So yeah, I think what he did at 13 is not going to be what everyone does.

32:41Not your typical 13-year-old. He did drop out at MIT, looks like. Oh, yes. He has LinkedIn. Grade 5.0. Dropout. The dream. The dream still comes out of the resume, right? there. Yeah. I think vibe coding is TBD as to its actual effectiveness. There've been a lot of interesting trend pieces about the vibe coding scene in San Francisco specifically, and the whole startup culture built around people that have no real technical skills, but have created their app entirely through vibe coding. I guess I defer to you guys a little, but like, I don't know what's the last significant startup that has raised money that was built on top of vibe coded app.

33:24I don't think there have been any. They all seem kind of lame. And I had fun vibe coding, something that I used this year. Oh, I built a... People make fun of me about this. When I go to the gym, I always forget what locker I put my shit into. And so I built an app through ChatGPT that it kind of used OCR to recognize the number. So I would hold it up to the number of the locker and it would store the number. And so then at the end of my workout, I would look at it and it would have the number in there. The obvious answer to this is why can't you just take a picture with your phone and look at that?

34:00I was going to say, I love the notes app. That's a great app that also exists. You can write it down. But then you have to write it down. I wanted an app that I could just hold it up to the number. And look, I wanted to vibe code something. And that was the idea that came to me. And you want a purpose. It's nice to have a purpose built app, you know? Yeah. I also learned that when you vibe code something, it only lasts on your phone for 30 days through test pilot, the iPhone kind of beta testing service. So it's now gone. Locker Buddy is a usable app. RIP. Yeah, I think I can somehow re-upload it there.

34:37But I am actually pretty skeptical of vibe coding as something with real utility. The fact that you can do it doesn't mean it has that much value. And so beyond just like little home projects like what I did, I don't really know this is quite the revolution that a lot of people are wanting it to be. Tom, do you still think we should all learn how to code? All of us? Well, you know, they say you don't ever have to learn how to code anymore. And the refrain was always learn to code, learn to code. But now we have vibe coding. The fact that we can code now in plain language, in plain English, and that, you know, obviously there are limitations, but it's getting better and better.

35:16The fact that people who are themselves genius coders and math geniuses think that you should just learn to code in English, like that's hugely significant. You know, I like raising my daughter. It's like, OK, I'm spending a lot of time right now trying to babble with her, eventually teach her English. You skip the step where you have to go learn some code based language to interact with computers. You can just use the English you learned. And all of a sudden, you know, the humanities majors among us are going to be empowered to build whatever we want on the internet. And that's validated by the math geeks this year.

35:52Like, that's huge. I think it's landmark. I think we're way underestimating how much it's going to be a boon to the word cells among us. I believe that, you know, you did it, Tom. You built the app. But what value did it truly add? It was a fun little project. It's like, I think everyone should learn how to like build model airplanes. didn't have enough ambition. It was sort of a silly little project, but like, yeah. Well, I don't know if like vibe coding allows you to like have much more ambition than what I did. Right. But we're near like one or two of vibe coding. We're in the beginning. Like this is where tech people dominate the rest of us.

36:26You see like, oh, it's just okay. It's like, yeah, because there's a rate of growth because they're just starting. Like we're just, this is just happening. This is all so fresh. But what are you going to really build? What would you do? You have the ability to vibe code right now. Like what could you build in your life that would change it in a way that you're like, oh my God, everyone fucking needs to vibe code right now. There are whole things around the newcomer business that I would like to build. We are literally talking to some of the vibe coding companies about trying to build stuff. Why wouldn't you just hire a developer to build something for you?

36:57Well, somebody like us, we're not going to hire unnecessarily a great developer. It's not our expertise. I don't know how to manage a developer. you know, and some of the things we want, you know, it's like the beauty of vibe coding, right? And this is why we had, you know, Dylan Field at Figma on stage at one of our Cerebral Valley events in London. You know, it's just like you get to think when you write a piece, you think about the ideas and it makes your writing better, right? When you're coding the thing, you're building the thing and you start to understand what goes into it, what trade-offs have to be made and so i think giving the person with the idea for the product the ability to do the actual thing without having to know how to code just allows you to sort of think through it and yeah it just empowers you can ultimately tell a developer what you want when you want like an actual functional high quality it's best for like prototyping right now i i agree but like yeah i think it's the trend it's ever going to be i do think that's true to your point yeah yeah i also I'll gang up on you on this one, Tom.

37:59I think it's pretty big. Let's circle back to this one at the end of 26. We'll see where my coding is going. Sounds good. I'm betting the under. All right. Quote 11. But make no mistake. What we are dealing with is a real and mysterious creature, not a simple and predictable machine. This is Jack Clark on the significance of artificial intelligence development during a speech at the Curve Conference in Berkeley. And that was in October. Your former colleague at Bloomberg, Jack Clark. anthropic co-founder and my former colleague like jack jack and i have hung out and he is we used to go biking together remember we biked on skyline i went to like a punk rock backyard show with him yeah wait no way i didn't know this yeah he was just a regular guy he covered google guy it's the most normie beat it was a little too wonky it's like jack this is bloomberg like worry about money less about tech and he like really loved technology and then he goes off to open ai and And yeah, it turns out you make a lot more money in Silicon Valley thinking about tech than thinking about money.

39:00I remember when he quit Bloomberg to go to OpenAI and we were all like, well, that's stupid. Why would you leave a great job like Bloomberg, which you ended up doing? I didn't think it was stupid. I just, you know, it was in the reporter Kool-Aid cult that you spend your life, you know, it's very hard to get into reporting. It's hard to give it up. But kudos to Jack. And Jack, you know, runs a very successful, he writes a very successful newsletter, Import AI. So somehow he's able to co-found Anthropic and do the newsletter hustle and be a top thinker on AI policy. Anyway, Madeline, we've talked about Jack, but what stood out to you about the quote itself?

39:38Right. Well, I felt like, one, he was fully anthropomorphizing the AI, which in this episode alone, we've debated if that's legit or not. And it also felt really representative of the Anthropic take. You know what I mean? Like all of the foundation model companies have sort of their stance on how much we should regulate this or not regulate this or limit the contents of the speech or be concerned if it's spewing dangerous ideas or what the power of the great machine is. And yes, AGI fears have fallen somewhat out of the conversation, but Anthropic is still doing it and they're still very concerned about safety.

40:13So that was sort of the signpost of anthropic posture, I felt like, as a quote that we should put in the list. Yeah, I feel like it fits in with the slotkin point of view, which is, yes, 2025 is the year where Doomerism's out of favor and Tom's view is ascendant. But there are still holdouts who think this is a thing heading to general intelligence. It is a real and mysterious creature. I sort of agree with Jack. I, what I actually take away from this is less like we, I've already said my piece about, you know, the anthropomorphization of technology. I am really interested in like the separation of anthropic from the rest of the AI field right now, just as a company, like kudos to Dario for, he's just a very earnest guy.

41:00Like he is stuck by his guns as I'm very worried about AI. I want to talk about regulation. Uh, he has been caught in the crosshairs with, with, um, David Sachs and the Trump administration. He truly does not like Trump, which causes him potential business concerns. And I think going out of this year, Anthropic looks like the better bet as a company right now, just in terms of its financial stability. Its aim is very true. It's not trying to do a million different things. And a lot can change. I saw Elon shitting on Anthropic at some point during the year saying like Anthropic winning was never in the cards or something like that.

41:41And they might go public next year, which is by no means success. You know, that's not like an end point. It's not like you've reached the end of your journey and you've won, but like they're building something that feels a lot more reasonable right now. Well, it's also funny Elon saying that when he's running, you know, XAI as well, as if he's just, you know, commenting on this, like he doesn't have a direct competitor stake. Yeah, we live in the world where you pick your philosopher king. And I think everybody on this podcast would prefer Dario to Elon Musk and Sam Mulvind. Plus, we all agree that Claude is the better writing partner, I have to say.

42:15Yes, although it's starting to get a little sycophantic to me. Jack is the man behind the philosopher king. So, all right, quote number 12. Overall, the models are not there. I feel like the industry is making too big of a jump and is trying to pretend like this is amazing. And it's not. It's slop. That is Andre Carpathi, my man, during his interview on the Dwarkesh podcast. And that is from October. Yeah. So I think you can tell it worked out quite well that these were chronologically next to each other because I did pair these together. If I was able to just throw them all on the page, you know, we come from anthropic.

42:51Oh, my gosh. The consequences of our actions. how will we live in the future to where the models are right now is bad in a kind of that kind of broke the internet in tech world, at least when that episode came out of having, you know, Andre Carpathie basically say the quiet part out loud that these models are great. There's things that they can't do as well as we'd like. And they're not. Are they God kings at this point? No. Yeah. I mean, this was the interview that shook Silicon Valley. Everybody takes what Carpathie has to say extremely seriously. He is both technically capable and speaks in plain English.

43:25And I don't think anybody is disagreeing with him directly. He's a little bit more cautious, maybe Daria, but he was more sanguine about the rate of progress than I think Silicon Valley had been up until that point. Yeah. This is, I think the first time we've mentioned it so far, but the word slop, which was the actual word of the year, although I like your sycopancy as another choice for word of the year, Eric. I think slop is a fascinating concept. It gets misapplied a lot. People are just calling everything they don't like as slop, which is unfortunate because I think it's a very specific thing, which is content that is unmoored to general taste or interest.

44:04And it just sort of gets produced and thrown out into the world. But AI will have to, at some point, reckon, the industry will have to reckon with the prevalence of slop that has come out of these tools. I think Meta's launch of Vibes personifies the slop. Is that a fair statement? Like when Vibes came out, when they were rushing to meet Sora, that to me was the slop of 2025. Just a fully scrollable AI generated content app with no taste discernment, no rhyme or reason, just scrolling through the endless feed. that to me more so than just, you know, slop being something I don't like to your point, Tom, that was like the definition of what really we're talking about here.

44:48Yeah. And is it a replacement for entertainment? Is it going to get, you know, infused into quote unquote, mainstream entertainment or will it exist as this parallel industry? The idea of someone just tuning out and like scrolling through slop is very nihilistic to me. I, I, I, and I'm not even saying that like i believe ai can have many positive uh effects on entertainment but slop if slop feed even sora the idea that someone could like go through sora as a social media tool not just a creation tool but like just checking out other people's like create that's weird that's really weird to me i don't love that idea sycophancy and slop 2025 all right quote 13 for our top quotes in 2025.

45:33Having listened closely to my fellow San Franciscans and our local officials and after the largest and safest dream force in our history, I do not believe the National Guard is needed to address safety in San Francisco. My earlier comment came from an abundance of caution around the event, and I sincerely apologize for the concern it caused. That was Mark Benioff on X after asking Trump to send the National Guard into San Francisco ahead of Dream force. This is someone Tom has written a profile of Mark Benioff in which Benioff sicked the governor of California, Gavin Newsom and Yo-Yo Ma on Tom to keep the profile upbeat.

46:11So I'm sure Tom will have lots to say about Benioff. But Madeline, you want to take a first crack at why we picked this one? Yes. So many will ask, why did I not pick the send the National Guard quote in and of itself. Fair point. However, I felt like the five day, like maybe not five days, but super quick apology to the mass backlash was such a like tail between your legs moment. And it really personified to me also, as we've talked about the techs move to the right, how quickly they'll easily swing back when they realize that it's not working for them. Yeah, no, it was astounding that he did this, that he called for Mark Benioff, the good, loyal Democrat, the one who was to the left of a lot of Silicon Valley Democrats calling for more taxes in San Francisco.

47:03And then he calls for the National Guard. It just seemed totally – He called once for a boycott of Indiana because of its anti-abortion law. He threatened to pull their employees in Indiana out of the state because he was so offended by a. Right. He was like the moralizer in chief. I mean, he was he embraced sort of woke ism. Right. I mean, he was definitely pro like sort of the push to D.E.I. I mean, the business case is just Salesforce is suffering and could use a boost. and maybe he's like, I need some of this Trump favoritism I hear so much about. But I don't know, Tom, get in Mark Benioff's head for a second here.

47:50I mean, I think I always used to use Benioff and Ron Conway as examples of when tech people get involved in politics and actually know what they're doing because they tend to win. Benioff supported Prop C and it won. Whether or not that's been good enough for the city is a different conversation, but he backs winners. He knows when to get involved in stuff. That was the revenue tax race. Yeah, on businesses, which was to help fund homelessness initiatives. And he spends, he's a big philanthropist and all this stuff. Ron Conway essentially picks the mayor in San Francisco every time there's an election.

48:24And you saw a split between them after this comment, after his National Guard comments to a point where Ron Conway, I don't think is really buddies with him anymore and stepped down from the board of one of his foundations. Kudos to Ron Conway. The only tech guy with a backbone right now. Yeah, in San Francisco. And Dario. I think Benioff is... We can do our newcomer's good list. Yeah, we've got Ron Conley. Nice list and naughty list. That would have been fun. Yeah, I think Benioff has got a lot of problems to deal with as a business right now. I mean, when I get this story... Maybe he's just out of touch.

49:01Like, he didn't know. Like, what's wrong with him? Like, I guess he just seems to not be able to read the moment. Or maybe he's reading it better than anyone else and things will come back around. But I thought you saw between the National Guard comments and then he went to a Trump state dinner, which is fine. A White House dinner. Like it's important for these guys to go there. But he was like taking selfies with Pam Bondi on it and being like, hey, the fun kids just showed up. And it's like no one likes Pam Bondi, dude. Like I don't even think Trump likes her right now. Right. Like the political instincts are way off right now.

49:34The only thing that's more often Benioff's political instincts is Salesforce's stock price. Benioff, was he on the podcast this year, by the way? Our podcast? Well, and at the end of that podcast, if you stuck around to the end, I still saw him like Ron Conway, where I was like, what's going on? You won't stand up to these guys. And there were early hints because I was like, you don't have more of a backbone than this. You've been such a loud Democrat. And he was pretty cowardly. Like I was definitely disappointed on on that episode. But yeah, he's in a weird place. I think people lose touch at a certain point, even if you have a great feel for, you know, the instincts and the politics.

50:15At some point, you just the scene moves on without you and you start making these kinds of mistakes. And Benioff had so many of them this year. I think the general billionaire world and successful business world did not clock the shift rapidly against Trump that happened as quickly as it has. And I think that they thought that they maybe had a couple more years of this before the pendulum swung. But the economic fallout and people don't like the immigration crackdowns and basically public opinion has shifted remarkably quickly away from Trump in the last few months. So I think a lot of people were caught a little off guard.

50:51We're going to get real fascism and we haven't gotten the memo yet. And these guys know what's. Yeah, that's what I'm saying. He might have better instincts than all of us. All right. Quote number 14 for 2025. A rejection of any kind. I should like shift in my chair a lot as I'm reading this. Do you have any Adderall over there? Yeah. Is that what it is? Okay. You can do a push up. Adderall. I think I won't get in trouble for joking about. I don't know which what it is. my son is is neuro atypical so i feel okay calling him out for this um a rejection of any kind of shared and defined sense of common culture in this nation and others has significant costs we should indeed must return to a shared national experience an embrace of common identity that by definition puts forward certain ideas values cultures and ways of living at the exclusion of others that's alex carp uh in his palantir's q3 letter to shareholders from november not deal book where, you know, he, we're referencing the crazy physical antics of Alex Karp at deal book, though.

51:54I don't know if the actual pros was as good. It was hard to quote the pros from the deal book moment. If we could include a video clip in this post, I would have included a video. However, I wanted to get the sentiment across of the twitchy conversation. Yeah. I think Alex Karp has been, you know, a really prominent CEO this year in tech, obviously Palantir stock has gone up quite a bit and down and fluctuated, but you know, it's been a stock to watch and his, you know, embrace of the, the West and the Trump administration and kind of defense tech as a larger theme. The West in the, as a synonym for white people, or I don't know the West.

52:35What does the West mean? How they say it's like Judeo-Christian values, you know, it's a catch all for all that, you know, and how best we should do business and live our lives. And his sort of embrace of that kind of went mainstream. And it was enough that he put it, you know, in his quarterly letter to shareholders that like, this is the philosophy we are going to follow at Palantir. And I think it's a philosophy that guides a lot of these defense tech companies that are, have propped up that are all kind of children of Palantir in a way, or children of Anderrill, which is also came out of Palantir, um, with a lot of ex guys there.

53:07So I felt like this was a way to capture the defense tech moment. And maybe we can embed a little twitchiness somehow. I take great umbrage, embrace of common identity, that by definition puts forward certain ideas, values. The great thing about being a country organized around the Declaration of Independence and Enlightenment ideas is that we're organized around ideas directly. Like I, I feel like they're trying to sort of, yeah, they're the, you know, there's a lot of identitarianism, uh, obviously on the rise this year. And this seems to try and have it. It's trying to have it both ways. It is.

53:53It's trying to say, you know, yes, we're rah-rah America, but if you're being rah-rah America, you have to embrace, you know, American values of being a country founded on ideas, not identity and, you know, nationality and race per that narrative. And he's also trying to do it pretty heavily. But yeah, our identity is what matters. And that's the right one. So it felt like, you know, really captured the spirit of those guys. Yeah, Karp's a weird one, though, too. He doesn't really fit the molds of a lot of these people. I mean, the dude voted for Kamala. And so he's also like, they always bring him out as the like, token Democrat defense tech guy at all these events.

54:31Yeah, like he's kind of a lib. He's just sort of all over the place. I never know what to to draw from him i like the idea of him as like an avatar of uh kind of the defense tech moment too because palmer lucky is the same sort of thing he's kind of hard to categorize um but that's true these guys matter but palmer you know is much more socially of the mega world than yes well i mean carp is close to peter teal so i mean they're all i mean it's complicated they're all complicated i don't have a great read on alex carp like i used to go to sun valley uh The big media tech. Not as a man with a badge, as a reporter clinging to the edges.

55:08Yeah, I would sit at a morsel. Please, please, CEO, can you give me a scoop about what coffee order you're making right now? Yeah, yeah, yeah. That's not even an overstatement, right? It's like Tim Cook had a coffee with Mark Zuckerberg. Scoop, you know. No, no. Tim Apple was at the coffee shop outside Sun Valley, and he tried to pay with Apple Pay, and they didn't accept it. And, uh, we wrote a little item about that for the information. I got picked up by business insider. Oh, that's a great wheel of content. Yeah. It's really one of the true humiliating experiences of my life is going to sun Valley, but, uh, I would see Alex carp walking around, uh, you know, in between the, the lodge and the duck pond and the amount of nervous energy emanating from this guy could power a fucking data center.

55:57That guy is, there's a lot, there's a lot going on there. Let's move on. All right. Quote number 15. I like this one. This is where we're looking for an ecosystem of banks, private equity, maybe even governmental, the way governments can come to bear, meaning like just first of all, the backstop, the guarantee that allows the financing to happen. So that was Sarah Fryer, of course, OpenAI CFO on whether OpenAI would need a federal backstop against its spending. And that was at the journal, Wall Street Journal's tech live event in Napa. And that was in November. Great, great pick. Backstop. That's another word of the year, honestly.

56:32Backstop really rang around Twitter. I mean, that one speaks for itself. You know, OpenAI CFO calling for a federal backstop in words that she quickly walked back after this live event. But that's the power of live journalism and why we have a thriving events business, because you say stuff on stage that maybe reveals what you're really thinking about. There was a lot of hand-wringing from OpenAI after that, where they were like, that's not what we meant. My words were taken out of context. I was just agreeing broadly with the premise of this. And they had to announce that they never wanted to do it.

57:03And then it turned out there was some letter that OpenAI had actually sent to the administration in which they had floated this. I don't have the reporting to back this up. This is purely based on intuition. They fucking want a federal backstop. There is no question in my mind that OpenAI wants to be sure that when they start to take out enormous loans to pay for their compute needs, that they want to be sure that they will not default as a company. and they got called out for it. They kind of got a little too close to the sun. They saw everyone on all, you know, from left to right. They had a policy memo document that floated this idea.

57:35This was not an errant thing she said. And it really feels like if they'd done it more artfully, they might've gotten away with it. Like it's not totally outside of reason that they could convince the Trump administration to say, Hey, it's very strategically important that we have investment. We're going to guarantee loans to open AI. There's no reason they'd fail. But I mean, thankfully, I think, as we said at the top of this episode, there are enough AI skeptics that that I think would be very politically unpalatable. And anyone who believes in free enterprise and risk and thinks bailouts to store the entire economy and that capitalism means capitalism should be against a fucking federal backstop.

58:18That's the risk investors are meant to take. They're betting on whether AI is a safe financial investment or not. And that's the provenance of the free market and investors and not the government. So yeah, I'm happy to root against the best stuff. If you're going with the premise of what venture capital is, it's inherently a risky investment. That is where these companies are birthed from. That's kind of the whole deal that you're taking a risk and it may not work out, but you may get super rich otherwise. So it's funny to see, you know, OpenAI, the most valuable venture-backed private company in the world.

58:53Maybe, is it compared to SpaceX? Where are they? Are they neck and neck right now? But anyway, it's, yeah, you know, we'll do our business in private. It's not even public, but yeah, we need a federal backstop. But it's not even venture capital, it's debt. Like debt comes with risk with it. Like that's the reason that people that take debt, make money from taking debt, taking on debt, like buying out debt because it's a risky endeavor. And so it was funny to see David Sachs, you know, and in response to this whole thing, being like the government will absolutely not be bailing out any AI companies.

59:24There are lots of model providers out there, which is an idiotic thing to say, because the whole point of being too big to fail means that like there are wider implications to a failure. There's a fallout for it. We said this on the show in stories on newcomer, OpenAI strategy of taking on these massive deals is a too big to fail strategy. Sam Altman has been intertwining himself into the broader economic environment such that if they were to default as a company, there would be a pretty large, maybe not catastrophic, but a large effect on tech stocks more broadly. And I don't know. I think he's engineered it.

1:00:04This is where we're at right now. OpenAI is wound into the valuation of multiple publicly traded companies. And if they were to collapse one day, it would have some sort of larger economic fallout. Okay. Brief thing here. It's like equity is much safer than debt, right? You know, it's like most of the money OpenAI has taken on so far is in the form of equity. So if OpenAI's value collapses, Microsoft or somebody could buy it at a low price. It doesn't mean it collapses to zero. The real problem is if it has all this debt that outstrips its equity value, then everybody could want to avoid buying it because they would take on all this debt.

1:00:47And then you have a huge problem. So right now, I don't think we have the level of debt that we have that sort of house of cards. But the debt exists at the peripheries. Like Oracle, which is a major partner of theirs, has taken on a lot of debt to pay for data centers. CoreWeave has taken on debt. CoreWeave has a ton of debt, yeah. Yeah. I mean, all the neopods float on a sea of debt. But no, those companies aren't going to get saved. That's why it's good that it's sort of externalized from open AI. So I don't know that I see the catastrophe you're painting, but Sam Altman has an insatiable appetite for more and knows that he's in the business of scale, right?

1:01:24These foundation models seem still to get smarter the more they spend on training runs. And so he'll take as much money as he can get. And if the government giving him a backstop to get cheap debt was going to be one path, he'd try it. But he's also happy to try things that don't succeed and move on to the next one. And so hopefully the backstop idea didn't take. To Sarah Fryer, it never hurts to ask. You never know until you ask. Exactly. All right. Quote 16 for 2025. I think democracy is fucked if we do this. That is from Law Adventures Andy Konwinski, interviewed by me at the Sierra Valley AI Summit.

1:02:05And this was about what would happen if the U.S. seeds open source on AI research advancement to China. Yeah. So, I mean, I picked that one in part because I feel like the open source debate hadn't been touched on as much in this list. But the the the increasing competition with China around the AI race was a very major theme of 2025. And they are obviously maybe they're cheating, but there's some evidence that suggests that they're taking some stuff from the US. But they are, you know, kicking our ass and open source in a lot of ways. So I think it's a fair point. Andy, you know, founder of Perplexity and Databricks, very importantly.

1:02:42So a serious guy and has the beard to reflect his seriousness. I don't know. Tom, you asked me the question. What was your take? The Rick Rubin of tech. Very difficult for microphones, by the way. Yeah. I think open source is a fascinating topic that was so big at the beginning of the year because of DeepSeek and this idea of whether an open source model could actively compete with what the proprietary closed source models can do. We've kind of forgotten about it. China really does seem to be way out ahead right now. Not even does seem to be, they are way out ahead. And there's a lot of open source models that are being used by American companies that are all made in China.

1:03:26And that is concerning to a lot of people, that dynamic. And you're seeing a few companies coming up like Reflection and maybe Meta is actually abandoning open source. So that's kind of interesting. But I actually, I would place my bet that five to 10 years from now, the most important models probably will be open source. It just is the trend of that kind of technology. So I guess if you're someone like Andy, you better hope it's coming from an American. What I don't understand is if it's open truly, then who cares if it's Chinese, we can still have us companies that host 100%. What's it matter?

1:04:02I mean, is he more interested? He's more worried about like the capability and just that if you're producing the open source research, you understand it better and therefore you're more capable. I mean, it's hard to get into like the geopolitics of it all, but I think it's, he was equating open source with democracy that like it's, and that's a kind of core belief of certain technologists is that technology should be something that is democratically available and open source is the only real way of getting there. And if we, you know, we believe in a future entirely run by AI, the public should have the right to, you know, be able to adjust it as they want and not pay a price every single time you want to access this technology.

1:04:42It's an interesting idea. Quote 17, home stretch, second to last. We should have like corporate music playing in the background, like those like training videos, you know? Like da-da-da-da, da-da-da-da. Yeah. We are delighted by Google's success. They've made great advances in AI and we continue to supply to Google. NVIDIA is a generation ahead of the industry. It's the only platform that runs every AI model and does it everywhere. Computing is done. NVIDIA offers greater performance, versatility, and fungibility than ASICs, which are designed for specific AI frameworks or functions. This is a tweet from NVIDIA's newsroom after the release of Google's Gemini 3, and it got a bunch of public praise.

1:05:20And also, I think it's specifically in response to a story from the information that Meta was talking to Google about purchasing a huge number of TPUs from them. Right, yes. So this, I would call the quote that personified the oh shit moment for a lot of people about Gemini 3's launch and how successful the model was while being trained on Google's in-house TPU chips to where everyone kind of realized Google has the full package under one roof. And we should not count out Google at this moment in the AI race. They have top scientists. They have a really strong, high-performing model. They have a great image model.

1:05:59They have a search engine business that prints money for them to power all of this. So they aren't necessarily as reliant on debt as maybe some other people will be. And they have the chips themselves. So that's kind of, you know, we can debate if they're a monopoly or not, since they kind of are told they are, but it doesn't matter. But they have an advantage here. Google stock is up 65 % year to date. It certainly has been the not not a surprise to us, as you've heard on this podcast before. I guess we should make more money betting on stocks we believe in because we've been Tom came into the job betting, betting on Google or narratively betting on Google.

1:06:37and the stock is doing well. But NVIDIA, I think as Tom is going to get into, this is extreme defensiveness in PR from NVIDIA. Like you doth protest too much situation. No, we're doing great. Our business is doing awesome. And don't forget that you'll need a lot more than just one type of chip. Exactly. We're specialized. Yeah, I don't want to go into the Lulu world of telling other people how to manage their comm strategy, but um sometimes not saying anything is the best response uh to an event and not even a particularly damaging one like nvidia is a four fucking trillion dollar company you can handle like one or two bad days of trading uh because of a story saying that like one company is going to be buying your competitors chips like you guys are still basically owning the whole ai chips game um so maybe maybe don't talk too much about that but and look in terms of google i have not actually messed around with the Gemini text model that much.

1:07:36I haven't seen it back and forth with that sort of thing, but Nano Banana might be my favorite app of the year. I think it's an incredibly fun image generation tool. It's so quick. It really understands what you're asking for. I think really is the, that was like the sort of lead piece of the Google comeback, quote unquote, comeback story was the effectiveness of Nano Banana. And they still got it. Demis He's still got it. NVIDIA is up 36.5 % this year. So not bad, but it's not as sexy as it had been. Over five years, it is up 1 ,350%. So yeah, we got so much growth in what, 2024? The 2025 for NVIDIA was just not as exciting.

1:08:29But Google has been insane. and we're getting a little defensiveness from NVIDIA from their public relations department. All right. Our final quote for 2025, quote number 18. I cannot imagine having gone through figuring out how to raise a newborn without ChatGPT. I'm too afraid. Do it again. He's not that bad. I cannot imagine having gone through figuring out how to raise a newborn without ChatGPT. That's Sam. Is that a better Sam? Sam, I don't know. I mean, it's usually a little more muted and kind of confused. He's totally reinvented his voice. That's Sam Altman. We refer to him as Sam. We have literally gotten questions from YouTube commenters being like, who is Sam?

1:09:13I think you can default on this podcast. I'm sorry. Sam Altman. But yes. Sam Ganges from Lord of the Rings. Sam, do you think we're talking about? Sam Altman on The Tonight Show with Jimmy Fallon, December 8th, 2025. Madeline, why was this our last quote of 2025? It was, you know, Sam Altman, go on TV and act human challenge. Oh, it was, you know, oh, every time we have these guys, you know, come and defend. I mean, so much of this list was really the AI industry coming out and saying why they're important, why they should matter. Political leaders saying why they should matter, defending them as well.

1:09:51This was, you know, to the normies, the Tonight Show, we're going on Middle America network talk show, Let's Chat. Why is AI so great? Oh, this thing, to your point, Eric, that humans have been doing for millions of years, can't do it without AI. Real relatable content there. Which is funny because on this podcast, I often invoke using ChatGPT for my newborn. So I'm sympathetic that it is a useful thing, but just saying it's the hyperbole of the fundraiser, Sam Altman, right? He gives these maximalist things to get people to give them his money. And then sort of suggesting that you like needed ChatGPT to do the most essential of human tasks, raising your child, just shows a level of disconnection from real humans that doesn't make you a great spokesman on The Tonight Show.

1:10:46Yeah, I think 26 is going to be a very critical one for the AI industry as we are well past the point of fun, novel, growing technology. OpenAI is probably going to reach, ChatGPT is going to reach a billion weekly users at some point early next year. And their big battle is going to be getting people, the normies, comfortable with this technology and wanting to use it every day. and finding utility in it so that they're willing to pay money and make this company at some point financially stable. And if you can't do better than Sam did on probably the softest possible interview in media, then it's not going to be an easy task.

1:11:28Yeah, Jimmy Fallon really known for his hard-hitting interrogative questions there. Yeah, the guy who must Donald Trump's hair. Maybe he should have done that for Sam. Yeah, he should have. But yeah, that would have been good. Should have ruffled his hair a bit. I don't know. All right. Those are the 18 quotes that defined 2025. Thanks for listening. See you in 2026. Bye. Bye. Thank you for tuning in to this week's episode of the podcast. If you're new here, please like and subscribe. It really helps out the channel. Listen in for new episodes every week, wherever you get your podcasts.