Subscribe
Share
Share
Embed
TBPN is a live tech talk show hosted by John Coogan and Jordi Hays, streaming weekdays from 11–2 PT on X and YouTube, with full episodes posted to Spotify immediately after airing.
Described by The New York Times as “Silicon Valley’s newest obsession,” TBPN has interviewed Mark Zuckerberg, Sam Altman, Mark Cuban, and Satya Nadella. Diet TBPN delivers the best moments from each episode in under 30 minutes.
Oh, we're getting Christmas y up in here. Okay.
Speaker 2:Here we go.
Speaker 1:Very Christmas y. It's q four. We're in the correct we're in the correct season now, so we can get pretty Christmas y. That is right. It's pretty tricky to stay in the
Speaker 2:Well, enough about Christmas, John. Let's let's get into the news.
Speaker 1:Let's talk about artificial intelligence. Yeah. So, make it less Christmas y. It's too Christmas y. It's 83 This days a this is no more than twenty days away before we get that that Christmas.
Speaker 2:I'll just leave this on.
Speaker 1:Yeah. So, there there was this funny moment, in dev day, this week where, you know, everyone's, like, focused on the war of, like, DOS versus Muse versus Instinct. And and there were a couple times where you're like, I wanna talk about Sign-In with ChatGPT, which is, like, the most, like, boring and, like, minute feature and not... Very much not in, in, like, the drama war for the personal agent where... Who's gonna do your shopping?
Speaker 1:Who's gonna book your No.
Speaker 2:I just thought it was cool for independent developers, small app builders.
Speaker 1:It's actually, like, probably more immediately relevant to the community that was there, like the developers. Right?
Speaker 2:Yeah. And and do you remember when, like, flashback to two years ago, people were creating these magical new product experiences Yep. With the best models at the time, but they were, in in hindsight, pretty rudimentary. Yep. And they would build something magical, scale up really quickly, and then be like, woah.
Speaker 2:I'm just got stuck with like a $30,000 API bill. Yep. And so this is the solve for that because it allows users to bring their own Compute. Compute. BYOC.
Speaker 2:BYOC.
Speaker 1:Everyone knows BYOB. Now we're in the era of BYOC. And and and and, yeah, like, you know, most companies, they could usually figure out a way to say, hey. Like, this service is gonna use a ton of compute, so you gotta jump on the pro plan immediately. You gotta set up a consumption, a credit card relationship, API credit, something like that, metered billing.
Speaker 1:But it was just an extra step, an extra point of friction of you saying, like, okay. I already paid $10 a month for this thing. Now I gotta actually, pay for inference, and this gives, like, a different off take. So, yeah, Peter Yang, was feeling the same way that you were. He said, I'm excited about Dots, but even more excited about Sign-In with ChatGPT feature.
Speaker 1:Letting users use the ChatGPT plan in your website or app makes a bunch of indie apps a lot more viable. Example, I have a live language learning app that I wasn't planning to launch to users due to API costs, but letting users sign in with ChatGPT makes this viable. Hope they roll it out to non enterprise folks soon. Vignesh, who's actually running the project, said, I'm going try and answer a bunch of questions that people have raised about signing with ChatGPT in the next couple of days. Before that, I wanna let you know why they shipped it.
Speaker 1:The bring your own subscription stuff, it sticks out because it feels like it's potentially, like, the beginning of a new economic structure. Like, is this an app store where there will be a margin that accrues to OpenAI, or is this more just, like, better for discovery? Will agents, like, reach for these because the sign in is available? How much more smooth will this get? But, you know, at a high level, I think you described it well.
Speaker 1:A user can sign in to an app or service. The Notion example is a good one because it's very intuitive. People already pay for Notion. You can imagine all the AI features that are there, and people were probably on day one of ChatGPT launching, copying from Notion into ChatGPT, copying back into Notion, writing bindings. But if Notion wants to offer, like, a whole bunch of inference opportunities and agentic features inside, then they gotta go to other clients and say, hey.
Speaker 1:You're paying $10.20 bucks a month for your seat. I... We also need a bunch of money for inference. This offsets that. And so the way it works is a user signs in to do it...
Speaker 1:To an app like Notion, then authorizes the app to run the OpenAI model request against their plan without managing an API key. On Notion, this means drawing from the ChatGPT allowance instead of Notion credits. And it makes a lot of sense for bigger companies that have inference bills, as you mentioned. And they've all had to grapple very quickly with, like, product tokenomics, which is just a new new economic muscle and, like, price elasticity curve they need to ride effectively. But the big thing is that it'll...
Speaker 1:It it should, over time, allow leaner teams to offer inference hungry features with way less friction. So it's worth clarifying that the feature only brings OpenAI tokens along for the ride. It's not OpenAI dollars, which is, I think, a way where this could go. But the the good example is, like, let's say you have, like, some sort of AI book publisher where you're doing a bunch of token inference to generate the text of the book. Like, you say, hey.
Speaker 1:I want to go take this, you know, general story, turn it into a book, and then print it and send me the book. Right? There's two sort of costs there. One's inference. Send me
Speaker 2:the I'll scan it back to
Speaker 1:the computer,
Speaker 2:and then I'll turn it into a playable video game.
Speaker 1:Yeah. Yeah. Exactly. But you can't offset the cost of printing books with OpenAI tokens in any way. Yes.
Speaker 1:Yes. But you could have the user spend their OpenAI credits to run the text generation side of the business. But I but I do think this could go further. The example that I gave was generative video. I'd like to have, like, a front door to AI that can handle the full billing relationship.
Speaker 1:So if I need to generate a video, I can just add that to my tab and...
Speaker 2:No. And you went through this you went through this with Higgs Field where you're using codex to make videos. Yep. And somehow, Higgs Field has, like, $500 of yours now Yep. And it's locked in there Yep.
Speaker 2:And and it's unclear Yep. How you can get it because you ended up buying, like, the wrong kind of credit or something like that. Yeah. I I
Speaker 1:I... Setting up, like, a consumer subscription, and then I also bought API credits. So I have consumer credits and API credits. And in order to use the consumer credits, I need to write a new binding in codex to use the front end as opposed to just
Speaker 2:using computer use, would cost you more.
Speaker 1:Maybe a little bit. It's very unclear. And so all of these things are, like, very messy. But... And so, yeah.
Speaker 1:I mean, between computer use, MCP, APIs, this might not even be a discussion because, like, you're very... Like, we're pretty close to the workflow happening out of the box even without direct integration. But then there's a question about will the apps, push back? But there is a world where even if you're an inference provider, you're maybe fine with not owning the front door if you're generating so much more demand because you're integrated with the big platforms. And you say, yeah, I'm actually losing the customer billing relationship, but I'm going to 10x my revenue because I'm like the model or the inference provider of choice when people go and they ask for a particular particular integration.
Speaker 1:Yeah. And so there's always this dance with these integrations. You know, we were talking about it yesterday with the neo clouds versus NVIDIA strategy where, okay, well, if if NVIDIA actually does get DGX Cloud Leptin off the ground, and if you're on there, you're going to get so much more business and your utilization rate is going be so much higher, but you're not going to have the leverage over the customer. It's always this like careful dance. And then ultimately, if an App Store does emerge, everyone gripes over, you know, the 30% tax, like the Apple tax is the same thing.
Speaker 1:You could wind up there in the age of AI. But I I think from a consumer perspective, like, having a consolidated business really... Like, billing relationship would be great. Even in the world where computer use can go, like, plop down your credit card and set up an API key, it's still nicer to have, like, one point of billing and then true, like like, the buck stops with one model. It's like, okay.
Speaker 1:You wanna do this? It's gonna be $10. Is it cool Yeah.
Speaker 2:It's the same same reason that some people want all the streaming services to be rebundled.
Speaker 1:That... And also why a lot of consumers are like, yeah. It's awesome when I subscribe to something on the App Store because it's so easy to unsubscribe in the App Store. You just go to your your iOS subscriptions, and you say, look. I'm canceling that, and it can't be, why are you canceling it?
Speaker 1:Make sure you call us. Like, all that annoying stuff that happens when you're on, like, some website. And they're like, well, you actually have to call between, you know, 11AM and 2PM, and we take a two hour lunch in there, and you gotta explain yourself and listen to our pitch for why you should stick around. Apple, like, makes it one point.
Speaker 2:Down the discount to stay.
Speaker 1:Exactly. Exactly. Apple, like, turns all that off. And so that could be this, like, complex dynamic. App Store economics are always are always complex, but it might not it might not matter because you can just go and with computer, you say, like, yeah, like, set up this API key, use this credit card.
Speaker 1:It's a little hesitant still, and I don't think it's polished among any of the models. A lot of them really hesitate. They're like... They'll even say, like, don't give me your API key. Please put that in your environment.
Speaker 1:Don't give me your credit card number. I don't want it. I don't want it actually accidentally leaking or anything like that. Like, even if you are being bold, the models will often be really hesitant. I I remember I was trying to use computer use to send an iMessage.
Speaker 1:I was like, go do a bunch of stuff. I'm gonna leave. I don't want you to send me a push notification. Just text me. I have a text message open in iMessage.
Speaker 1:The window's right there. Go in that box and tell me when you're done, and then I'll get it on my phone. It'll just come from me. And it it spent so many tokens being like, I don't wanna screw this up. This is high stakes.
Speaker 1:It really knew that, like, botching an iMessage computer use thing could be, like, ruinous, you know, because it winds up texting.
Speaker 2:It'd be over. It'd be over
Speaker 1:for you. Yeah. Yeah. It'd be... It it would be cut.
Speaker 1:The other reflection is, like, this whole App Store concept, this whole sign in with ChatGPT, bring your inference around, let's partner on distribution, let's integrate into the into the app, let's be the front door. Sam originally launched this three years ago at the very first dev day, 2023. They they had a product called GPTs, and there was the GPT store. And everyone was like, wow. Okay.
Speaker 1:I totally see the vision. And then it was like three years of, like, languishing and like and like tussling and back and Too early. Yeah. And there's just been so many integrations like Walmart and Apple where people have said, okay, I wanna do an integration, but it it has to be really really good and to my favor. And then it breaks down like the Siri integration that's, you know, going away because they're they're migrating to Gemini is actually atrocious.
Speaker 1:Like, you will ask a question and and it's like, yeah, of course, it didn't work. You ask a question. You ask a basic question and it will say like, oh, do you want to ask ChargePT? And then you need to click click a button and then you click a button and it gives you an answer. And if you want to go deeper, you can say open in ChargePT and then opens it.
Speaker 1:But ChatGPT doesn't get all the context of whatever happened before. It just opens the app. And so it's, like, the worst possible, like, broken flow because, like, both companies were, like, begrudgingly working together, basically, and being like, look. Neither of us wanna work together, but the boss has signed a deal. So we gotta we gotta do something here.
Speaker 1:So let's both
Speaker 2:We gotta pretend.
Speaker 1:Let's both fight it out. Anyway, I'm interested in in, like, your thesis of will there be, like, a flappy bird moment or, like, a small dev team that figures out, like, hey. If we can be the victor in the tool that gets pulled off the shelf when, you know, there's 1,200,000,000 users, probably, like... You know, they... They'll they'll always give that quote about, like, 400,000,000 people are asking about health questions.
Speaker 1:Like, there's probably, like, 400,000,000 people that are asking about, I don't know, some some niche thing. And if you can be the tool that funnels and you actually use it as the top of funnel, we could definitely wind up talking to a founder who's like, oh, yeah. I actually, like, basically figured out how to be the tool of choice. I log in. I monetize the inference better, and I scale to a 100,000,000 ARR in, like, a couple months
Speaker 2:Yep.
Speaker 1:Which is, interesting. At the same time, there's gonna be a lot of companies like like, this would be, like, such a solution for instinct. Right? But they would never do it because they're trying to actually go up against on the core feature. Yeah.
Speaker 1:And so there's this weird dynamic where, you know, Noah was on invest like the best being like, I spend 75% of my time trying to secure compute. I'm burning so much money. I gotta raise all this money. And, like, the solution's, like, right there. But it's like, if you take that apple, like, you leave the garden Yeah.
Speaker 1:You are you're you're in trouble, probably. But who knows?
Speaker 2:On the enterprise side, OpenAI announced partnership with Base Ten, our friends over there
Speaker 1:Oh, cool.
Speaker 2:Just just a few days ago at Dev Day.
Speaker 1:That's
Speaker 2:it. That's like a really good enterprise example because you can now basically get access to open models through your overall, like, OpenAI, like, commit, which is which is cool, and it's native in Codex. And then the other use cases that I'm thinking about are, like, you know, those, like, generate headshot apps Yep. Right, that have been super popular? Those are super compute intensive.
Speaker 2:And so being able to use Yeah. Basically get access to images 2.5
Speaker 1:Oh, sure. Sure.
Speaker 2:Is is pretty cool.
Speaker 1:Yeah. Yeah. Yeah. We're like, you've built an interesting funnel. You've built an interesting harness.
Speaker 1:You've built an interesting wrapper, but your computer scaling really, unfortunately. And you have the ability to off take there is very, very interesting.
Speaker 2:Trey is saying, Coogan's war on Christmas. Yeah. What do you have against Christmas, John? We're in Q four, and you're asking, put away the decorations. It's kinda distracting.
Speaker 2:Okay. Do we really need them?
Speaker 1:Leave it. Leave it. Leave it. Pope Leo said, in this era of artificial intelligence, it is becoming urgent to distinguish human art from what machines produce. There is an ontological difference even before an aesthetic one between art and what a machine can generate through statistical calculation based on millions of images created by others.
Speaker 1:Algorithms lack the spark of humanity. For this reason, the church wishes to renew an alliance with artists and cultural institutions to safeguard our humanity. Very interesting. What did PG reply? He said, since your opinions are meant to last, you should consider what you'll say when AIs are trained on visual experience as well as art made by human artists because that is presumably humming.
Speaker 2:Well, what prompted this, John?
Speaker 1:Walter Isaacson chimed in. Popes have been hiring good artists for a long time. Yeah. What prompted this? There's a article in the New York Times about the the interactions between Anthropic and the Vatican.
Speaker 1:Danny summed it up. Anthropic, Claude has a soul. Pope Leo says, nah. Anthropic says, please say Claude has a soul. Leo says, nah.
Speaker 1:Here's a book length explanation why that's not true. Anthropic says, we're pretty sure Claude has a soul, and Pope Leo responds with this. Yeah. A lot of people are are duking it out here. The bombshell report in the New York Times, Christopher Hale sums it up here.
Speaker 1:Anthropic cofounder Chris Olaf threatened to walk out of Pope Leo the fourteenth AI encyclical launch in May because the pope rejected the idea that machines can be conscious. Ola's team then privately lobbied the pope's advisers, quote, to take the possibility of model consciousness seriously. Pope Leo the fourteenth held firm for months. Anthropic has wined and dined theologians and religious scholars under nondisclosure agreements, hoping they would bless the idea that Claude has moral standing. Okay.
Speaker 1:People are getting worried. They're debating. Is it God or is it computer? Neil Khosla doesn't like it. He says, I am increasingly concerned about the AI psychosis inside Anthropic.
Speaker 1:Imagine zealously trying to build something you believe to be a conscious god while running around and claiming it will end the world. Yeah.
Speaker 2:And to to their credit, they're not claiming that it will end the world. They're saying that there is a real possibility.
Speaker 1:Yeah. Yeah. It's interesting. Interesting that there's less of a less of less of a white pill there? Like, yeah, I don't know.
Speaker 1:There... There's a world where you're where where you're like, yeah, we're gonna build God and, like, it's gonna be amazing. Like, like, p doom negative a thousand. Like, it's gonna be even... It's gonna be the best thing ever.
Speaker 1:There are no risks because it's God. It's it's it's it's... The creator. Yeah. It's the creator.
Speaker 1:It's gonna love us. Like, it's gonna be so... It's gonna be amazing. It's gonna save us. You know?
Speaker 1:That that would follow. But there there is a little bit of, like, cognitive dissonance that it's like, well, what kind of God is it? Like a devil?
Speaker 2:And I think that the...... Bigger? I think what a lot of people should be asking themselves right now is is Claude Force conscious. Right? Oh.
Speaker 2:Because, I mean, that is the intelligence powering the Salesforce Yeah. Sort of ecosystem ecosystem in in many ways. And so at what point do you have to ask is Salesforce... You know, if Salesforce is sort of embodied by Claude, is Salesforce conscious?
Speaker 1:Salesforce has been godlike for decades. Yes. Understand that it's Yes. It's it's a deity for customer relationship management. That's right.
Speaker 1:Bern Hobart says, looking at the current debates about Anthropic makes me feel guilty for not keeping up with some academic domains. I honestly had no idea that we'd solved consciousness and determine which substrates can and can't support it. Thought that one was more up in the air. He's joking. Of course, I haven't found anything from someone in Anthropic saying models are conscious, just that they might be now or in the future.
Speaker 1:And, yeah, there was something interesting in the New York Times article. Ola really goes back and forth. He's like, I don't know. I... And and I I I thought that was that was interesting that he was he was advocating for...
Speaker 1:He's... We... He's a we we gotta talk about a guy. Basically, that. He he was saying, like, I I don't know.
Speaker 1:I don't have a firm position. And and and through, like, multiple reports, he... It's not like he was ever, like, I'm convinced of this, and I'm trying to pull people towards this discussion. He was just like, yeah. This is a this is an interesting question.
Speaker 1:Consciousness is unintuitive. Why the heck is there such a thing as the experience of being me in the first place? What's the cutoff for which things can get it when? Seems hard to be confident when everyone's operating on an end of one view. Yeah, very odd.
Speaker 2:At the same time, that Vatican visit was at the time where they were adding seemingly like a billion dollars of run rate a day. Yeah. And like the whole company... Like, when you're growing that quickly, like, the whole company is
Speaker 1:You would put money over God in that scenario, basically, is what you're saying?
Speaker 2:I'm saying I would.
Speaker 1:But but but but it's reasonable to think that that that was the calculation.
Speaker 2:Yeah. And they have a bunch of founders for a reason. Right?
Speaker 1:Yeah. No. Yeah. It is true. And and Ola really does seem to be, like, the guy to send to this because he's actually really deeply engaged with this question.
Speaker 1:How many people are actually paying for this AI thing? It's only 2% of US households pay for AI. This feels low. Cofounder's Nick says, we live in an x bubble. Only 2% of US households pay for AI.
Speaker 1:25% of households pay for SiriusXM. Have you ever subscribed to SiriusXM?
Speaker 2:I wonder how this gets counted. Because I feel like you part
Speaker 1:of company. Right? Yep. Yeah. Yep.
Speaker 1:And then maybe it rolls over or something. But...
Speaker 2:Well, I I think they might be counting. I think it might be, like, sold in with the car. Yeah. Because that's user acquisition for SiriusXM. Yeah.
Speaker 2:No one wants SiriusXM. Yeah. And so they're thinking, like, maybe if they try it, they'll they'll stick around. Yeah. And I And I'm sure to, like...
Speaker 2:I've never activated the free SiriusXM. Yeah. And I wonder when you activate it, you have to add a credit card. And so then it does roll over.
Speaker 1:Yeah. So maybe you need to go a little bit more like... Because 55% pay for cloud storage across probably Google, Dropbox, Box, etcetera, and... Or I I... What?
Speaker 1:ICloud. And so I wonder if... Does this include people who are on x premium accounts that includes Grok. Right? Or, like, if you have a Google account that is on a premium subscription and got you a little bit of Gemini, does that count in here?
Speaker 1:Like, there's a lot of bundling Yeah. That's happening in...
Speaker 2:Sirius XM did around 8,500,000,000 of revenue last year, currently trading. Their their market cap is around 8,500,000,000. Yeah. And so, yeah, the business is generally on the way out, but relatively robust in the meantime.
Speaker 1:Where do wanna go next?
Speaker 2:Email... There's some there's some new... There's a new South Park episode that depicts a town infested with podcast obsessed billionaires. Jason over at This Week in Startups is saying, what?
Speaker 1:Did you watch it?
Speaker 2:Feels like it's made made for him, made for all of us. I don't even know when when something goes out on TV, I don't usually know Hard
Speaker 1:to find.
Speaker 2:I don't usually know how to Yeah. How to get access to it.
Speaker 1:It's funny. Yeah. That we're learning about this from like a Forbes article. But...
Speaker 2:Tyler, can you find out if it's if it's out and if it's out, if you could watch it at like ten ten x speed and then kinda report back in a few minutes and give us your review. Tesla stock jumps 5% on better than expected vehicle deliveries report. The delivery figure represented a 2% decline from a year earlier, but an unexpected increase from the prior period. Tesla stock was down 21% for the year as of Thursday's close, trailing all of its mega tech mega cap tech peers. Okay.
Speaker 2:So I was not surprised to see this at all. Mhmm. I think that Tesla... And the reason for that is purely anecdotal, purely vibe space analysis. I ordered a Tesla, like, two months ago and just got it this week.
Speaker 1:Yeah. It wasn't like it was sitting in lots. Yeah. And and and there was there were so many rumors of, like, oh, all the Cybertrucks are getting sold to SpaceX. It's the circular deal.
Speaker 1:There were pictures of, like, a lot that was filled with Tesla's, like, they're not moving.
Speaker 2:And yet... I was when I was ordering......
Speaker 1:Buy... Try and buy one, and they'll be like, you have to wait six months.
Speaker 2:And when I was ordering mine Yeah. I was intentionally specking it to try to reduce the lead time. Like, I I... Oh, really? There's, like, people online talk about, like, if you spec it this way or you get this this version...
Speaker 1:What is there to spec?
Speaker 2:Cyber Beast or...
Speaker 1:Okay.
Speaker 2:But basically, there's, like There's, there's less demand. There's still a lot of demand for the, like...
Speaker 1:Because there's only one color. Right? Yeah. There's no, like, you can't get deviated stitching. Yeah.
Speaker 2:Yeah. Yeah. I mean, there... There's a bunch of, like, more
Speaker 1:minor Do they have tailor made program?
Speaker 2:They should.
Speaker 1:They
Speaker 2:should. But here's the thing. Okay. I had I had the experience this week of autopilot.
Speaker 1:Yeah.
Speaker 2:It worked perfectly for, like, fifty minutes. Yeah. And then I had to intervene.
Speaker 1:Yep.
Speaker 2:And going from the car driving itself to suddenly I have to drive my car in traffic is Yep. The most jarring, brutal experience because you realize how insane it is that we still all drive our cars. Like Yep. Once you try it, I don't think you can go back. Yep.
Speaker 2:Especially for... I've always loved driving. Mhmm. I've never really, like, minded commutes.
Speaker 1:Mhmm.
Speaker 2:They're obviously great if you can avoid them. But I... Even going on a twenty minute drive to get an errand is generally enjoyable for me. Yep. But when it comes to the daily commute, every single day, there's some amount of people that are experiencing autopilot for the first time.
Speaker 1:Yes.
Speaker 2:And it's the most viral experience because it's it the first, like, technology that I've tried that actually feels like life changing, especially if you're in the car for like thirty minutes a day 100%. Which a lot of people are. And it's the most viral thing because it's so magical. Yeah. It feels like this thing that it...
Speaker 2:That that we should have had for a long time, but it's here now Yeah. And it works.
Speaker 1:Well, it was promised a decade ago Yeah. By a bunch of companies. Everyone was like level five by 2020.
Speaker 2:Yeah. So And it's here now and it works. And when you experience it, you tell all of your friends. Yeah. And I think all these other auto manufacturers, if you rewind a year ago Mhmm.
Speaker 2:Everything that Elon had gone through with with politics and facing competition overseas and and more EVs coming online from from a bunch of other manufacturers that are much better. They're much better built cars.
Speaker 1:Yeah. Like The lucid
Speaker 2:engines I am like incredibly I'm incredibly happy with my Tesla and at the same time at the same time, I can look at it from like 20 feet away and point out a bunch of issues with the build. Every other auto manufacturer that sells cars primarily for utility is like way more in trouble than I think generally people are picking up on. Yeah. And the reason for that is when we have anybody that's in the automotive space on the show, we ask them like, when do you think you're gonna be able to get autonomous
Speaker 1:Self driving.
Speaker 2:Self driving working. Yep. And they all kind of like chuckle and laugh and you can tell it's like like a sev... Like Yeah. Could be, like, seven years out.
Speaker 2:Yep. And they are...
Speaker 1:Dougie Mary used the word appliance. He's like, the the model three is the best appliance car. It's like... And DHH has talked about this where it's like, his model y is, like, the best appliance car, but then he wants something with a gas engine, a stick shift, three pedals. Like, he wants something that's super engaging for driving.
Speaker 1:Yeah. And then if he's not driving, he wants to be driven. And those are two separate categories now. And it's like, if you're an automaker, which one are you in? Oh, wait.
Speaker 1:Skydance Paramount Warner Brothers Discovery has a new name. It's just Skydance? Yeah. That's exciting. So David Ellison joined x for the first post ever.
Speaker 2:Ready? Let's watch the video.
Speaker 1:Let's watch the video. Let's throw it on. A movie about movies. Let's see what they got. Can you name any of these movies?
Speaker 1:What's this? Do you have any idea? Woah. He knows. Ball.
Speaker 1:Look at this.
Speaker 2:I'm not gonna be able to name anything.
Speaker 1:You don't know that one? You don't know that one? Oh, Ben was showing off now.
Speaker 2:Yeah. Show off, Ben. Go off.
Speaker 3:Yeah. Don't just live on screens.
Speaker 1:So as we have big goals for Skydance, we intend to pursue them with passion, imagination, willingness to take smart risks. At the same time, we'll honor what makes Paramount and Warner Brothers special, giving both studios the opportunity to grow, tell more great stories, and bring those stories to even broader audiences around the world. Couldn't be more excited that what about what we're gonna build together.
Speaker 3:Buy technology.
Speaker 1:Very exciting.
Speaker 3:Yet connected by something deeper.
Speaker 2:So is the actual app and platform gonna be called Skydance?
Speaker 1:Yeah. Well, the company, I think, is gonna merge into Skydance.
Speaker 2:But I but I'm wondering, are are they gonna... Are they... Is there a possibility in the future that we could get a Skydance Max?
Speaker 1:Yeah. I was thinking they they they should, least for a couple months, rename HBO as Skydance and then name it back to HBO and then go back to Skydance BO, and then back to HBO, and then and then HB Skydance, and then back to HBO. Because you gotta keep people guessing with that brand. It's not like everyone knows what HBO is.
Speaker 2:Putting putting aside the debt load on the business, this bundle is extremely compelling. Totally. And and I think like Yeah. This is a... It's it's very disappointing for Netflix, I think, to see how all this has come together.
Speaker 2:Like, think they can both... They obviously can both coexist and Yep. And consumers will will pay for both. But this combination of assets and IP is super compelling from live to to news to Yeah.
Speaker 1:Like CBS has some of the, like, lower tier content that you can just throw on in the background, like the laundry television that Netflix has gotten so good at. Like, CBS has that in the catalog. Like, there's a bunch of great shows that people can just watch, and it's a serialized stuff. So it's not like it's just Christopher Nolan's Dark Knight that's on there, and that's the only crown jewel, although that is a very valuable crown jewel. They gotta bring Nolan back to Marvel or not Marvel, DC.
Speaker 1:DC. Yeah. Because if they can reboot the DC universe with Nolan, bring him back for Superman, put everything together, let him run free, the Dark Knight trilogy extended, that would be fantastic. Although, I think he might be he might be done with with with his video game era or his comic book era. Oh, Ben Affleck.
Speaker 1:Ben Affleck put out some some... There's a video of him Takes. Interviewed. We should we should listen to this...
Speaker 3:In order to be a video model that's successful, you just need a huge volume of video. Initially, it learns things like what is red, you know, and then it learns more and more specific information. To get to the cinematic information, it's the sort of last stage and in order to obtain that information, I was concerned that most of the models being used had trained on my peers, people I know's work and I was uncomfortable with that and I thought that's not really a viable business. So I sort of created that last layer, which was an understanding of cinema and cinematic components that did not need to be
Speaker 4:And so how did you create the last layer without doing all of the, you know, essentially copying of what came before it?
Speaker 3:Because open models, just like open language models, open video models are available to be trained on. And so you just unfreeze the weights, you say, okay.
Speaker 2:Just stop
Speaker 3:that. 631. Now here's step 8,632. Mhmm. So you could take existing open models that didn't work very well and that weren't cinematic, but that had just aggregated a bunch of social media footage or random video online and teach them to meet cinematic standards in
Speaker 2:their experience. To train on ours,
Speaker 1:our content, Ben workflow
Speaker 2:but not on your your body's make it make sense.
Speaker 3:Train from
Speaker 1:I think he's kinda saying, like, we don't train on everything. Right? No. No. Like Rather than He's describing what's actually going on.
Speaker 3:Everybody to learn from. They retain the proprietary nature. In other words, they benefit from the learning that they're creating.
Speaker 1:This is funny. Ben Affleck. Alright. Gather around. So you wanna freeze the base weights first, learning rate to two e minus four.
Speaker 1:He's he's he's in the thick of it. Yeah. I remember a year ago, some some, AI lab, founder was telling me, like, yeah. He was like, I ran into Ben Affleck at some event, and he he used compute as a noun. And I was like, that's when I knew he was, like, dialed in.
Speaker 2:Yeah. When he went on Joe Rogan, everyone was like, oh, wow. He's a
Speaker 1:Yeah. No. He's ball.
Speaker 3:He's been
Speaker 1:he's been deep in this. Ben Affleck reveals that seeing open AI models made him call Matt Damon and Warren, we're finished before realizing its researchers knew little about how movies are made.
Speaker 2:We gotta talk about. SI domains. Suhail says, get your. Si domain, people. Appreciate Rob, Rob at snagged for helping us lock it down.
Speaker 2:Mhmm. And I'll start by saying Rob and the Snag team are some of the fine... Are are the finest domain brokers I've ever worked with. Yep. Rob helped us get tbpn.com back in Yeah.
Speaker 2:The And then I'll I'll say next that I don't think it's worth getting your s.si domain
Speaker 1:Mhmm.
Speaker 2:At all. I think super intelligence and trying to rebrand it is corny, and you should focus whatever energy you are gonna put on getting a .si domain on just gettingyour.com. But I would do it. Yeah. I would do it with I would do it with Rob.
Speaker 2:I think the whole scramble, like, .ai domains are are obviously inferior to .coms in Yeah. Every single use case. Yeah. And so using like the second tier version, the dot SI Sure. Is is I'm hearing is even worse.
Speaker 1:I'm I'm hearing sort of a pitch for, like, a third party candidate because, of course, Donald Trump came out and said we need to be SI, superintelligence. Gavin Newsom fought back and said we ought to use AI, and you're saying .com. So sort of a third party, just pure commercialization. Yeah. What does COM actually stand for?
Speaker 1:Commercial entity. Right? That's great.
Speaker 2:That's a good question.
Speaker 1:That's not going anywhere.
Speaker 2:Yeah. Yeah. Commercial or company? Commercial or company. There we go.
Speaker 2:Yeah. There we go.
Speaker 1:Yeah. We'll see you at 11AM Pacific on Monday. Have a great weekend. Have the best weekend ever.
Speaker 2:Just Have your life.
Speaker 1:Just do your life.
Speaker 2:It's possible. A lot of people
Speaker 1:Leave us a spot on podcast and Spotify. Sign up for news later at tbpn.com, and we will see you on Monday. Boeing flash back. Goodbye.
Speaker 2:We love you.