The CEO Quietly Licensing 2M+ Hours of Content to AI Giants

I'm Rob Kelly and this is Media and the Machine, a show about the biggest technology shift of our lifetime and how to profit from it. Each week I talk with the founders and CEOs closest to AI and content, the ones figuring this out in real time. I'm also building an AI content business myself and chair a lesson to what I learned along the way. You know, life's funny. I began my career lucky enough to interview leaders like Steve Jobs and Bill Gates. Then I went on to be a three time founder and CEO driving $100 million plus in revenue and some failures too. And now I'm back at the table interviewing this new world's current and future leaders. This isn't only a business story, it's a human one. So every episode ends with me asking my guest what AI means for our jobs, our families, and the next generation. We'll figure this out together from the inside. Welcome to Media and the Machine. [Music] My guest today is Clint Stinchcom, CEO of Publicly Tragedy Curiosity Stream. What I love about Clint is that his company might, from the outside, be described simply as the Netflix of documentaries. Not a bad business at all. In 2024, he helped reposition the company so that AI content licensing would become its center of gravity, with a real shot at becoming the majority of the company's revenue. Jerry Ossie's mission is bold and clear, become the largest provider of content for AI training. And with two million hours of content array, they just might be in the pole position. For your content exec, you'll love the nitty-gritty of this conversation. We get into their nine AI licensing partners, 18 deals across them. And why Clint expects those numbers to double or triple in the next year. We talk about what content AI will pay the most for, including why structure matters so much, why categories like wildlife and sports command premium value, and how many hours AI companies want to test or try before they buy. If you're building an AI, you'll love hearing how Clint thinks about giving AI companies what he calls the "FLA." We also get into the numbers, including a run rate that looks to me like $20 million a year in AI content licensing, and how that could exceed subscription revenue by 2027. This is an amazing media transformation story, and I want to give a special thanks to my high schoolmate Danny Hanigan, GoEdgemont High, and Dartmouth, for bringing Clint and I together. Please enjoy my conversation with Clint Stinchcom. When did the light bulb go off on AI content licensing as a revenue source? For me, the light bulb went off in early 2024 when I read about the Google licensing agreement with Reddit. And as I was reading it, obviously I was really interested in how the $200 million was going to transfer and what the commercial exchange was around that. And obviously it was for AI training and for attribution or display rights as it relates to text. And shortly after that deal, which I didn't even completely understand at the time, you saw a spade of publishing deals, not video, but publishing with companies like The Atlantic and the Financial Times and AP and Reuters, and then even newscorp announced $250 million a deal. And so I thought this got to be a video business that follows this. There has to be. And we know that these models need video and we know that developers can't just scrape it all from the internet. They need ethically sourced videos. So we started talking to people, myself, a couple of our salespeople, just to learn as much as we could. And I said, let's just try to do the first deal that we can do. Let's do a deal because there's no better way to learn what's required, what the opportunity is by doing a deal, like just get into the arena. So in summer of last year, we did our first agreement. It was for not a lot of hours. We licensed a thousand hours, but we learned four critical things. First thing we learned was, okay, we can clear the legal hurdles that are required here, because that is something that people spend a lot of time talking about. And so in our case, we're able to do that. Secondly, we learned that raw video is helpful in training the models. And so it's not just finished content that, "Hey, I developers are interested in." And when that came to light, thought, okay, we have a lot of that. And then we know the partners and other contacts who have a lot of that. So that would be a pretty simple way, I thought, to amass content and content that obviously couldn't be scraped from the internet because it's raw video. And then the third thing we learned, which was really interesting, is that the rights that these partners were looking for were not exclusive. And starting to talk and learn more about the handful of deals that had been done, like everything's not exclusive. So, but okay, if we can license similar content to multiple partners, then that's really interesting. And then the fourth thing that I learned was, okay, the rates that we're going to be paid on an hourly basis for the right of an AI developer to train on our provided content. And then there's a fraction of what we would receive from Netflix or HBO Max or any of our other licensing partners. And so if we're going to have a real business here, we need to figure out a way to scale up to a million hours or so. Yeah, I called that paid R&D deal the best where you get paid. Meaning, curiosity is paid to actually learn the new market. Yes. It's a lovely, lovely business model. That's a great way to look at it. I'll say that. The opportunity cost is, if it doesn't work, then you've given up some degree of opportunity somewhere else. But, you know, we thought this was well worth the effort. And then from that, we set about reaching out aggressively to those people, business development people at the hyperscaler level and beyond that to try to get a sense of what their needs were and to try to understand what it would take for us to work toward an agreement. These are a goal, stated goal for the AI licensing. I think that if you look at the CAPEX spending in the AI space, it's almost mind-boggling. And the component of that that's allocated to training data, meaning text, audio, video, etc. That's anywhere from the 2 to 5% range there. And then a percentage of that is video. Our goal is to be the dominant provider there. And is that just of the factual content? That's everything. We want to be the highest quality and the biggest. I mean, you can measure biggest on volume of hours. You can measure biggest on revenue. Obviously, for us, the latter is more important. We think that we can do that because we've built a really substantial library now. We control over rights to well over 2 million hours of content and we anticipate that will double over the next year. Double of next year, amazing. So in this case, that's a combination of content you own or own the rights to. Correct. Correct. Okay, great. It's all factual content. No, it's not. We have a healthy amount of factual content, but we also have sports. We've scripted films. We have scripted TV series. And is the fictional content only for the AI licensing? Correct. Yeah. So fascinating. So in the streaming business, you're still, let's call it non-fiction. Yes. But in the AI licensing business, you're agnostic. You'll grab whatever content is valuable to these AI companies as your customers. Yeah. And, you know, certain people that we have licensed non-fiction content from also have fictional content. So our core category is in factual science, history, nature, technology, crime, lifestyle. And then we have secondary categories of religion, education. This is content. If kids content, biography, outdoor, meaning, you know, fishing, hunting, adventure, we have a lot of combat sports. And now increasingly, we have a lot of stick and ball sports as well. What's among all those categories? What's the most valuable to AI companies? Well, I think one thing that I didn't mention to be the dominant or among the dominant is our ability to structure data today, which most companies just don't have the capability to do. What this enables us to do is clip at scale hundreds of thousands of hours of content into seven to twenty second clips, which is oftentimes what these AI developers want the model to train on. And then as part of that, we'll provide scene description, a shot description, and camera angles and a whole host of other metadata, which is really critical. So in that scenario, you're getting right to the filet. You know, right to the filet of what they want. And so beyond our existing partners today, we announced publicly we've done close to twenty fulfillments with nine different partners. I mean, everybody wants seconds and thirds and force and some sevenths and eighths doesn't mean that we've done that with every partner, but we're kind of on that track. And you know, our view is that whereas our existing partners may be 60 to 80% of our revenue next year, we think there will be an additional level. Well, double or triple the volume of our partners. And that's probably going to deliver 20 to 40% of the overall AI licensing revenue. We have agreements with nine partners and we've done 18 fulfillments. I use the word fulfillments instead of deal, but it's essentially the same thing. Okay. Nine partners and average of two deals each. You know, it's not perfectly that. Okay. So 18 deals among nine partners. Mm-hmm. And the hardest thing are getting the initial agreements done. subscribe. fulfillments typically require one page at 10 or less. So getting back to your question about value, where we have been paid the most on a per unit pricing basis is around some specific wildlife interest in a particular sports interest. And particularly on the sports side, that is not easy. There's a lot of work that has to go into that to work through the whole shane of custody on the right side, which we've been able to do. So we're seeing a couple deals per are they testing you out for the initial deal? Yeah, I think that's another reason why there's a real barrier entry to this is there's typically a pretty extensive evaluation period where you're providing, you might provide 10,000 hours of content for a company to review based on what they're planning to do around their models. And so all of this takes time and just to get like a first deal done typically, I mean, it could take a year. Why sports and wildlife is the most valuable? Well, I think that there will be certainly other categories, but on the sports side, I think for some of the consumer propositions that are being contemplated, they can be pretty meaningful based on some of the apps that they want to be created as an example. We license some soccer content. And I know that part of what they want there is just to train, train, train, train on goal scoring. And so from every angle and every type of weather on every surface, all these types of things, that's interesting. And you can imagine that that's the type of content that people might watch. They might be incorporated into a coaching app. And that's something that many parents would probably be supportive of. The use cases are infinite. But I think with sports, one, it's not readily accessible. You can't scrape division one football from the internet. You can, but be really illegal. So you can't do that. And then that's never happened before. Yeah, right. Yeah. But on the wildlife side, there's certainly footage houses that have lots of animals that they can provide. And there have been a number of deals that have been done, whether it's a cat or a shutter stock or a whomever on the footage side of things. But for longer length video, there's certain animals that have known cameraman in the past who have camped out for eight weeks to give a shot of a snow leopard in the snow. So it's not the easiest thing in the world to get at either. And at the same time, we're able to tell them, somebody might say, "Hey, I want gazelles running through the tall grass at sunset." Volume's still really, really important, I think because you have to, what we know is that scaling laws in AI show that additional video improves accuracy, improves generative capabilities, even with diminishing marginal returns. And so if you're an AI company today and you're not generating continuous performance gains, then you're arguably losing. So one thing I'm hearing is with the two million hours and counting to license, that's of course going to get you in a conversation with just about anyone on the AI side, the quality of the data, if you don't have that, they might turn to someone else pretty quickly. Yeah, well, so the first is, I think if you don't have a critical mass, so 100,000 hours of content, it's going to be really difficult for you to even engage and developers are busy. They don't want to work with 100 partners, 200 partners, they don't work 50 partners. We're going to a very finite number of partners. And so we think by being able to check off the sort of boxes that I've talked through that we can be either the first go to or the second or third. And you kind of touched on this earlier, but is this initial wave of the first nine companies, AI companies? Is it safe to say they're the frontier models, the big LLX in that first wave? Yeah, with a few exceptions, for the most part. And I think again, like to help provide some sense of kind of what we think is possible over the next year, the next 18 months, the next 24 months, I think that are we're not going to create five more hyper scalars. We've got those guys like, yeah, there's not more coming. However, we track 100 companies who have been funded to a certain extent where we believe that they will need to license content, video, audio at some point in the next 12 to 18 months. And so that's why I say that we have 10 partners today. We think that'll either double a triple by the end of next year. And then we see sort of an exponential growth of partners, certain open source models become more readily available to us because we think that there are hundreds or even thousands of companies that would look to fine-tune an open source model with 10, 20, 30, 40,000 hours of content. And they'll pay for? Yeah. Oh, yeah. So these are companies using an open source like Lama for Meta? So open source video models, like Alibaba one, you know, like Genma open source, probably the next version of Meta Movie Gen. You know, there's open source models that exist today, but they're not, I think, as widely used as they might otherwise be for a whole variety of reasons. But as they get better, they absolutely, absolutely will be. And so we're sort of referring to this as the open source and tune market. Was I mean in tune? Fine tune. So they're going to take this open source model. And they're going to, let's say that they envision a tour guide app or a coaching app or a gardening plant identification app or a sports commentary app, whatever it might be, they're going to license video that will help in the creation of that. As the list gets bigger from the 10 to say, I think you talked about. 20, 30, yeah. Yeah. At least funded. Does that change the approach then? In other words, will you need to, and let's say that hundred then, you know, this thing keeps going north. And you know, if it's a huge industry that hundred could go up to a thousand, right? Oh, yes. Absolutely. Five years now. I think faster than that. Like I think in the next three years. Well, that change, you know, instead of doing direct deals and kind of the classic biz dev type deals, will you need to create like a marketplace for the content? I think that's conceivable. I'm not as it relates to video. I'm not a huge fan of the marketplace for a variety of reasons. What's the main one? Well, so we've talked to a handful of people who have tried to create a video marketplace. I just think that in today's world, the thought that if you kind of build this thing, people will come like it's just not true. You know, and I think that any of these developers who need high volumes of content, they're having conversations with people like us and a few others. And so are they going to go to a video marketplace and buy a few hours there possibly? But if it existed today, and I've had people ask me about this and I say like, I plot anybody that wants to build their dream, or be a dream builder, not a dream killer. But I don't see much opportunity in that in the next year or two. However, you know, over time, it's a different deal. And I think as you, especially as you get to, if you get to thousands and thousands of people who need to license this type of thing. Now, we're able to do that in some capacity today by just providing lots of samples and providing viewing rooms and, you know, even giving people access to an S3 cloud. So that's not exactly a marketplace, but it's a way to automate it. Yeah, exactly. And no curassies public, but you've made some comments around this. I feel okay asking it. So subscription revenue, your primary revenue source, 8 million and change the last quarter. Could AI content revenue, licensing revenue, beat that in the foreseeable future? Yeah, our subscription revenue today on annual run rates in the $38, $39 million range. And so you'll a little bit more than nine a quarter now. And that includes retail, wholesale subscriptions. There's some subcategories in the way that we report it. But that's what it rolls up to from pure subscription standpoint. And so what I said at the beginning of the year is I thought that AI licensing could be more than half of our subscription revenue. And in that case, I was referring to something that we thought would probably be in the $35 million range as it related to our subscription revenue. And then later on in the year, we said it's conceivable that can happen a lot faster than that. But we thought that that was a fairly conservative and sober way to communicate it. Yeah, it's fascinating. And you and I haven't chat about this. But one of my first jobs on the internet was I had worked for a publishing company. We had a million article database trade magazines. And this is the same p media. And my job was to license the content out this great archive. And so I just remember this business. And it was just, you know, nothing comparison to how you're doing. It was a couple million a year. But it was like turning on a feed. And it was about 90% profitable. I would say I'm just going to sort of. I know there's a little more to it in the AI stuff with the data labeling, especially. But I mean, this not only could become your largest revenue source, but most profitable, right? It depends on how we continue to a mass content. So we've been able to build a large library through mostly through revshare propositions. And so when we enter into an agreement with somebody, we may be paying out 50% of what we receive to our content partners. But I will say one great thing about that is as you make more partners money, it just increases their interest and ability and willingness to work with you, you know, where it's like, you know, pulling teeth at the beginning because people don't know what the heck is possible and start to do agreements, have a level of success and provide people with a real new revenue line that costs them nothing. That becomes really exciting and then as you start to do well by people, then more people just contact you directly on the content side asking us to represent their content as part of our corpus. Right. So maybe in that case, the 90% margins are what your, what your content partners are getting, you guys can get that maybe 60% on your own content. On our own content, absolutely. Yes. And on the traditional licensing side, you know, where you had 90 to 100% on the margin side. Gotcha. By the way, where's more revenue coming from for the I licensing your own content or your content partners or content partners. We just have a fraction of that two million. So yeah, it's our content partners. Yeah. And that's really at the end of the day, we make money three ways through subscription through licensing and through advertising. That's more of a nascent line for us. But on the licensing side, this is an expansion of our of our licensing work. But we are granting a new right that I didn't even, again, I wouldn't have even known how to define it 12 months ago. But we're granting a training right today that all of the publishing deals that were done around text that taught the models to read. And in the case there, most of those publishing companies had no concerns around being listed as a source for this type of content. In video, granting a display right is a whole host of complications there that will need to be worked out over time. But we saw this as a great opportunity. We're conveying a right to train on our content. And we just thought if we can get in early, if we can really delight our customers, we have an opportunity to be, I think, one of the top, you know, one, two or three provisioners of video for both the hyperscalers and all of the secondary and tertiary companies that need to license content either to train a frontier model or to fine tune a model. What type of company was that first deal? They got the first AI license. Yeah, it was, it was, it was a small company that I don't even know, honestly, if they exist today, but it was a chance to do a deal and to work through everything that we thought would be contemplated. Now, that's not something that I think everybody fully appreciates at the beginning and understands because what we did is August of last year, we just had a working group that will get together and we still do every single day at 11 o'clock. And we kind of go through three things. One is, okay, how are we doing on the content acquisition side? The second thing that we needed to figure out was how in the world, you know, operationally do we handle all of this? How do we store it? How do we deliver it? And then thirdly, we needed to be really persistent and focused on reaching out to the developers, meaning like those people that can write the checks because if nobody sells, we're not eating and none of this stuff is worth anything. So kind of three-pronged approach. How's the kind of the sausage made like if you look internally at curiosity? Yeah. Is it a couple of developers hanging around with your biz dev person who knows the parameters and you're sitting down and slicing and dicing this stuff up? Or is it all in house? It's at least as it relates to the business conversations, those are all in house. Operations is largely in house, you know, content acquisition is in house, but we have three outwardly focused people working with the AI community with me being one of those three. And then we have a pretty strong operational team that just gets better every day that does things like, you know, ingest the content because we have content from, I don't know, 150 partners, a couple hundred partners, something like that. And so there's sort of a constant delivery going on that we need to manage and control and communicate with people. And there's an internal process there that's really overseen by one guy. And then we're often being asked to provide samples. And so somebody that will put together a video showcase and send it out and people just kind of constantly working on summarizing everything that we have. So it's easily accessible. So there's a lot going on behind the scenes. We chose to repurpose existing people that we had as compared to eliminating jobs or going outside to find people who may or may not be perfectly suited for this. So, you know, like a lot of these jobs didn't really exist. So it's helpful to have a background in production when you move to doing content acquisition. I mean, it's helpful to have a background in managing studios when you go to being responsible for organizing high volumes of content. There's no perfect path to many of the jobs that exist at curiosity today. Are the titles any different now? Like is it still just said content acquisition? It's AI content acquisition? Yeah, no, it's it's um, you know, content acquisition is content acquisition. We don't have a single person who has AI in their title. That's the kind of thing that you see really early on. I understand why certain companies do that. And obviously, you know, on the operational side, you want to use whatever productivity tools are available to you because I mentioned the translation component. Like that would be an easy one for us. It's been helpful to us in areas like customer service, been helpful in areas like editing, sequencing, and going faster. So we want to use the tools that are available to us. But we don't have anyone with an AI title. Maybe we should, but I think it's it's almost right now. Now you mentioned last time we chatted that a common question on your earnings calls, analysts are why can't synthetic data replace what you're providing right now? What's your answer? I think that's I think it's a great question. And what I would say is synthetic data, what I mean more people than I would say this, but synthetic data is incomplete. Synthetic video can augment real data, but it doesn't fully replicate real world physics actions or the diversity and context of authentic data, authentic video and data. I know this is real new, but some key people from Facebook have recently left to follow this. What if anything are you doing in this new AI world model? You just brought up physics and 3D and that's part of this new world, which is getting defined kind of as we speak. Is that new set of customers? I think it's absolutely a new set of customers. And I think they're trying to figure out like exactly what they need from a training standpoint right now. Is there a new type of content in your view that they'll need? For instance, say you might not have right now. Something we don't have that I know people have paid a premium for, but there's not a lot of it is something like video around STEM physics as an example. Courses were strongly learning focused. We don't have that. And so I'm sure they would need things like that. But at the same time, we think that what we have will be helpful to them based on how they view this new type of model. What do you think when people ask, hey isn't AI data going to run out? Yeah. We have very few concerns around our ability to continue to amass high quality content, which is too many partners. We know where lots of things are and we have, we're going through it this morning to summarize our foreign language scripted content, whether it's, yeah, it might be Japanese, it might be Mandarin, it might be Hindi, it might be French. There's still a lot of content in the US. And increasingly though, there's content that's interesting that comes from outside the US and I have a good buddy in the adventure space who's built a hundred fifty thousand hour library. Now he's been grinding away on that for 25 years. And that's great. It's kind of a one stop there. But because we work with a lot of partners and even certain distributors, they're able to bring us a lot of content. And even some that can just kind of cycle in and cycle out based on deals that we're doing. So, I heard somebody talk about the way Elon Musk works one time and they said, look, yes, he, everybody understands that he sleeps four or five hours a night and he's really well organized and he's really good with people and he can, super smart, can cut to the core. But what many people don't understand is over 10, 15 years, he's developed 20 people that work like him. And so he's got kind of this 20 person group that he trusts completely and they have shorthand communication. And those people, that's a big part of what makes him able to run five companies or four companies. What it is. I mean, and so in our case, we have some really talented people, aggressive people on the content acquisition side that don't work for us. But they have a kind of a base level librarian. So they're going out and amassing more, you know, that comes to us just based on the existing terms that we have with them. So you got kind of a feeder system. So you're not totally dependent just on your internal - You got it, exactly. - Now, I've heard you on earning calls, talk about the four types, premium, video, audio, scripts, and study guides, like four pieces. You mentioned that the video is, the video is first among equals, but by far, and I don't have a ton of data to support this, but we're working on more in the audio space now, a lot more than we were a year ago. - Yeah, and on the audio side, I mean, I'm familiar with 11 labs, for instance, let's me create content with a huge grateful dead fan, so it allows me to create content with Jerry Garcia's voice. - Yes. - I assume they got that lately. I haven't checked yet. Is that kind of a core use case of audio or whatever? - Yeah, that's absolutely a core use case. And I think that-- - For you guys too, though, like, you know, are people wanting David Addenborough's voice? - Well, they do, so there's certain places you can go today, and it won't say David Addenborough, but it might say British professor. - Yeah. - And if you listen to it, again, it's not my business pretty darn close. (laughs) - Yeah. - And so in Latin America, we have probably, I don't know, 60 to 80 natural history documentaries, which are, you know, voice of God, one voice. Those are synthetically dubbed. There's an AI voice that narrates those. It gets a little harder when you get into multiple speakers to really nail it. And even in the audio book space, there are, that's a big location for AI narration. In part, because, especially when certain authors want to narrate their own books, it can sound awful. If you want a voice that appeals to the broadest group of people, without a doubt, there are, you know, there's people like Matthew McCommey, I think is the number one voiceover guy in the country, I think, pretty close to that. But there are AI voices that are just pleasing to the ear of a much greater number of people than, firstly, any humans. We don't have a lot of code, you know, but I think we had, you know, close to 10 million tokens of code, we've licensed that. It wasn't available on, you know, GitHub or any of these, you know, open places like certain people's code is. - I see. You're actual programming code for being, underlying the engine of a streaming business. - Yep. - Okay. And AI companies want to learn from this, to create their own AI based streaming platforms? - Well, possibly, they're after like massive volume on the code side, right? - Oh, I see. So just for development, for engineering in general, they could be using it for nothing to do with streaming. - Correct. But it's just emblematic of what is possible. I mean, if you'd have asked me two years ago, like if you just said, hey, you're gonna license your code, six figure deal, you know, I would have said, what are you talking about? So that makes it really interesting and exciting. - Yeah. Greenfield. - Yeah. - What most affects the price of the AI licensing data? - Yeah. On the video side, it's what heavily impacts it is, just how structured the data is. How much metadata is there? And it's really unbelievable today. We're able to provide a same description, a shot description, the camera angle, a 20 second clip could have 12 tags. If you structure it in a way where I say you're getting to the filet of what everybody wants, it's an efficient buy. And certainly people like in anything, you'll pay more for efficiency, because the overall result will be equal or better. - Is it more often the case that the customer, you know, the AI developer is the one defining what data categories there are versus you? And then you're going to see what's possible. We have to communicate to people the scope of our data sets. Some of them are obvious, but many of them are not. So we, like any other company, we have like an eight to 10 page deck that helps people understand the broad quality and scope of our corpus. And then it's hard to find hours that aren't helpful to them as it relates to video training, but they're going to be certain categories that are more interesting based on whatever kind of use case they're focused on at the time. - I'm curious who right now, who's reaching out to who first is it? - Like what happens is to do the initial agreement, we're reaching out to everybody because we don't know anyone. Once you have an initial agreement, once you prove that you're a good partner, then one thing we've seen like over the last 12 months in the AI space is you've had a lot of turnover on the kind of the research engineering science leadership. And that content to sort of make their requirements the agreements that they enter into very choppy, 'cause new guy comes in, he has to confirm that this is what they want to do. So we're reaching out obviously, but now more people are coming to us. And as you do, as you do more business with people, you're just in contact with them more. And we want to make it easy. So, as I said, we try to over-deliver in everything that we do. And we are working to be the number one go-to source. We've both been in tech for a long time and it used to be kind of a rule of thumb that folks outside the US trailed the US in terms of trends and models, often by like two to three years, sometimes more to hang on what part of the world. How does that compare in the new world of AI? - There are no hyperscalers other than possibly China that existed Europe, right? I mean, so what you have there are, I say this was a fiction, but just like from a side standpoint, like secondary and tertiary companies who are starting to license content and we know we'll need more content, but we're definitely behind, certainly from a licensing volume standpoint and just from a capacity standpoint. So we think that outside the US will be a component of our partners. And there's some saying that we have a deal with them, but this company's like Synthesia, and Mistral that have been pretty public and then press about what they're looking for and what they want. And certain Israeli companies that are, we think are gonna be pretty strong. It's companies in Germany who need a certain amount of content for frontier models, but also for fine tuning. And so we know that portion of our business will come from there, but the US based companies will dominate. - Yeah, it's interesting. If you put November 3rd, 2022, Chatsy PT launch is kind of the starter pistol going off. And you exclude China. Other than Mistral, I couldn't name you a major frontier model outside the US. I'm sure there are and they're kind of small and maybe it is still a two to three year lag. - This is my kind of tracking sheet. This is like everything you'd ever need to know about your costy business. And we have 87 companies on here who, we think have been funded to a level where they'll be in a position to license content. - What percentage are US? Fast, pretty. - Yeah, fast majority. - And he surprises on countries that are kind of like, you know, the second most common country. - Not yet, but we're certainly prepared for that. And you know, I don't know that we've done a work in Turkey, but again, I'm not an engineer and you know, I've done all the jobs you can do in media with the exception of kind of engineering and tech. But the guys I talked to that are smart and working internationally. They always tell me that Turkey has the best engineers. And so, you know, do we think something will come out of there? Probably, you know, they also have a pretty robust kind of video marketplace or increasingly so. But Germany, the UK, France, that's probably where most of it will come from. I think again, excluding China in the near term. - Being a content guy, I want to get your hot take on the big AI companies first. Just from a content perspective, just like, you know, what word or two comes to mind about the following AI companies. And I'm just gonna round them off, you just give me whatever comes to mind. Open AI. - The first. - Google. - The biggest. - Anthropic. - Code. - This is fun. X AI or AK GROC. - Thank God we have a competitor Wikipedia. - What do you mean by that? - Well, I just think that Wikipedia has a certain point of view and have never completely understood like why certain things go into pages and why others don't. And so I just think it's good to have a counterbalance for anything. - Oh, I see. So kind of an alternative. - Rock and Peanut. - Yeah, yeah, yeah. - Okay. - Yes. - How about meta? - The most tech expertise, is it relates to structuring data among a few other things. - And I know these two are in frontier models yet or maybe they won't be, but just looking at them as AI companies and having a huge impact, first Amazon. Starting to license, did it take a few New York Times? - I just think personally, I'm the last person people should probably go to for investment advice, but I think my largest holding for a long time has been Amazon. And so I think that they just have escape velocity. They're just so dominant. And I know this, like Amazon people a year ago would tell me like we're behind. We're behind, we're behind. But I think it's inaccurate to believe that they won't catch up. - Yeah, kind of like, you know, people you say this about Microsoft, not the first word processor, not the first spreadsheet, and then they not only catch up, but they beat you. And then people seem to say the same thing about Google now in terms of they weren't the first with a-- - You just look at the Gemini results last, you know, from a week ago. It's incredible. Yeah. - Unbelievable. - They are all sleeping giants. [BLANK_AUDIO] capacity has to relate to this stuff. Yeah. Don't poke the giant. Yeah. Microsoft similarly. And then Apple. But just Apple is an AI company first. I'm going to ask you about some streaming companies too to get your same hot take on that. But just Apple in terms of a potential LLM or this world that we've been talking about. So they are not doing an LLM. At least they haven't communicated that they are publicly. And nothing I've seen. And so I mean, I like many people. I'm an Apple user. I love Apple products. But they seem to be in search of a solution. Yeah. Could come from M&A easily enough. I mean, they have a ton of money on the balance sheet. Are there any streamers other than curiosity that you know about training AI models? I might be inaccurate here, but I don't know any that are. You know, if you're running one of the studios, like there are a lot of issues to work through. And probably they would say that the juice doesn't necessarily justify the squeeze at this point. Hey, couple of selfish questions. You, I think you know, I love documentaries. I've got a newsletter where I recommend documentaries. What are some examples documentaries that do surprisingly well just in your world in the business of documentary streaming? Beyond just AI, which ones do well that you wouldn't expect? Well, I think that, you know, one great thing about natural history content or nature content is that something that can be enjoyed kind of multi-generationaly. And so there, what tends to do well are programs about animals that can eat you. Like that's, that always works. I've got a tenure all the night in the test today. Yeah. Yeah. We do it with jaws and claws promotion that we do every year at Curiosity. So, you know, our not as big, but you know, along the lines of Shark Week, that discovery. So, we do jaws and claws. So, that stuff does incredibly well. We tend to over index, I think, on people who love science content. And then, you know, history is very accessible for a lot of people. And so that's sort of my go-to is the history side. John Hendrix, you got to work closely with John Hendrix. The Godfather of factual content is what Wikipedia said. I haven't looked on Grock to see what they said. You should say that. Yeah. I'll take credit for that. Can you just give a little background on what John founded? Yeah. So John grew up in the, so shadow of the space station in Huntsville, Alabama. And so he has real passion for factual content, for space, you know, for science, for history. So his move into this world is what he is, you know, just loves it. And so he founded Discovery Channel, you know, what became Discovery Communications. So, they built that from, I spent, you would say, four tortures years raising money. And then ultimately, you know, John Malone at TCI and the Cox family and the New House family came in and invested in his idea. And, you know, it was actual network at the time, like right at the time where he was out of money. The thing was going to go under. And so they built, they took advantage of a cable business that increased. But they grew that from zero to $24 billion at his peak, something like that. One of the great highlights of my career in working with John is he's just somebody who's always thinking about the future. Trying to think about what's around the corner. And, you know, he's confident and he has the courage of his convictions. But at the same time, he's a gentleman, which you don't, he's a true Southern gentleman, which you don't see across media a lot. What does he think of AI? I think he's really intrigued. He has a great eye for creative and he loves elegance in creative. And we've shared recently a few of the spots that we've developed solely from AI. And so when you show somebody something like that who is a big thinker who is, you know, a futurist like from that, you know, you might get 10 or 20 ideas from him. Well, you know, what was it? They showed him. Can you share? We showed him a couple of spots that were promoting, you know, some new fast channels that were rolling out. And we showed him a 30 second spot that was a direct consumer acquisition spot, you know, for curiosity string. And, you know, just to watch it and think like, okay, this is great. And it's 100% AI. Because you can spend a lot of money in a 30 second spot if you want, you know, a lot. And so the things that, you know, may have cost us, you know, 15, $20,000 to create, you know, 25,000, 30,000. Again, you can spend hundreds of thousand dollars on a spot, but cost us just time of existing interested in employees. So I got some humanitarian questions to ask you and it could be like a speed round. Just give me a short answer as you'd like. Do you consider yourself an AI optimist, pessimist or some other descriptor? I would say an optimist with a clear view of what can go wrong, but, you know, without a doubt an optimist. When do you think AI might replace your job? As our business becomes more predictable, more, you know, reliable in certain categories, I think that would be the point in time what I would be replaced. And so I don't know what the time table is about that. What are your thoughts on universal basic income? Yeah, I mean, I think we should do whatever we can to take care of the least of those among us, but I'm not a fan. I think whenever you just provide that to people, you know, the results historically have shown that's not the results are not net positive. Not at all. What do you tell the young kids in your life about what changes to make in this new world of AI? Well, you know, I've told them one, like learn everything you can't become an expert. Like one, you don't have to tell them necessarily to use the products because they probably do already. And it is interesting that they tend to look at them as kind of a companion or pal as compared to older people like myself still sort of probably think of them as glorified search. So I just tell them to learn everything that they can there. And so tell them that especially nephew just is just graduated from high school and getting ready to take a gap year. And I said, look, you know, my dad was university administrator. His whole, what he was solely focused on was bringing older people, veterans and others, you know, back to college either back or for the first time. And today I would say one, if you have an idea, like I'll be really supportive, but as you go to college and even for those that are in college, like just understand what you're there for. If you're there, I mean, if you're playing a sport, obviously that's, that's a, you know, that might be a little bit different. But one gets to know people, lift you head up and get to know as many people as you can. And you know, realize that the courses that you're taking there are not necessarily going to, unless it's maybe engineering or medical or something like that, not necessarily going to prepare you for what you end up doing in life. And I've also said, you know, that as much as I appreciate college and paid for six college to wishes, there are many paths to success and fulfillment. And, you know, college doesn't have to be one of those and there's a lot of ways to learn. And so I'd be supportive of anybody that, you know, my extended family that didn't want to go, but had something else they wanted to pursue. If AI ends up doing a lot of work in life, like your work and you had it in less time, what would you do with all your newfound time? That would be an unbelievably great problem to solve for. I'd like to say I would spend more time with the people that are important to me, so I'd focus a little bit more on relationships. I would focus a little bit more on my, on my health. And I would like to think that I would focus on at this stage of my life, like, I have lots of dreams, but more in helping people work toward and realize what it is they're most excited about. I'd also go to a lot of SEC football games, but there you go. And finally, will you create an AI avatar of yourself for family and friends, including maybe while you're alive, but also for them to use when you're no longer physically around? That's a fascinating question. And I will say my mom passed away a little over a year ago and right after 80th birthday, and my sister and I had dinner with her on her 80th birthday, and I just asked her a ton of questions, you know, stuff that, you know, what was the best year of your life, you know, who was love of your life? Like, what was the best advice your dad gave you, you know, what was your relationship like with your grandfather, all these things that probably should have talked about, but never had the chance to ask her like, what's the best and worst part about getting old her answer there? My mom would not, she had British boarding school education, but she, so she had a pretty extensive vocabulary, but, you know, unless she was yelling at one of us, she never swore. And so she just said, yeah, like the worst part are your infirmities and the best part is you don't give a flip about what anybody says or thinks about you. And so I regret not recording that in some way. And so I've encouraged friends that I have whose parents are older and getting near the end to really sit down with them and capture them on video, you know, just have something like that that you'll be able to show sort of generationally and that will be a nice way to help remember them. I hadn't even considered the the avatar idea, but it's that is that's it's fascinating. And is an interesting way to kind of ideally contribute some capacity that you're gone. That's a cool answer because I had the assumption that it would be us. deciding about it for ourselves. But I really like your answer that let the others who want it, make that decision and do the work and let it be up to them, which is kind of how it works now. Yeah. Well, my sister still says to me, she's, I wish we, you know, and she's super close to her and we, you know, moved my mom and her husband out to live with my sister in the, you know, four, five years before they both died. And she said, wow, I, you know, like, I've spent tons and tons of time with her and I never knew those things. And so she's like, I wish we would have taped that. Like, yeah, well, I said, if you get to your end of life, I'll tape it for you. But hopefully I get there first. She's much younger than I am. Hey, Clint, thanks so much for investing the time. Thank you Rob. Much appreciate it. Well, this is Media in the Machine. A few things about you and me. If you want to hear about the next new episode, make sure you hit follow on the show and your podcast app. If you want to go a little deeper, head to media and the machine.com and subscribe. When you share your email with me, you can see handcrafted transcripts, read the essays in my newsletter and be the first to hear about who the guest is on the next show. You can also email me directly from there. Maybe you want to recommend a guest. I'll give you a shout out if you do. I love paying it forward. In time to time, I also open up office hours and host small meetups for subscribers. Just to meet, talk, and build things together. If you're creating something to your own or thinking about it, I'd love to help. Maybe you've got a podcast in you. Finally, I don't have a marketing budget for this show. So if it's finding you and others, it's because someone like you passed it along. I'm genuinely grateful. If you have a moment, an honest rating helps you make this better for you. You just go to the show page, you click one of the stars. And I'd rather you give me a low rating than no rating at all. I mean it. It pushes me to get better. Thanks again. And see you next time. [Music]