Hello and welcome back to Lion Trees podcast. I'm Alex Michael and today we've got a guest who sits at the intersection of creativity, business, and technology, which is why we're having him for a record third time on the cast. His name is Scott Belsky. Scott is here and he's now a partner at A24, the popular independent film and TV company, and the founder of A24 Labs, the studio's new AI and innovation arm. In his words, the mission is simple: help creatives experiment more. And I'm sure Scott is doing that every day. And of course, many of you know Scott from the years leading product strategy and design at Adobe. Before that he founded Behance which he successfully exited and he invested in companies like Pinterest, Uber, Airtable, Ramp and a lot more. He writes a newsletter that everyone should check out called Implications. He's also the author of a great book called The Messy Middle, which you should all check out. And we're so excited to have Scott here because he's going to walk us through his predictions for 2026. And we'll do a look back at his 25 predictions. It's something not to miss. Scott, thank you for being here. You're part of the very rare three-timers club here at the Lion Tree Podcast.
It is a great sign. You're a very important, thoughtful person. So, thank you for coming back.
Keep going. Thank you. It is truly a thrill to have you here. We have a lot to discuss. In fact, when we look back at I don't know, we've done 120 plus episodes. One of our best rated episodes, let's say, is the one where you did predictions. You gave us your predictions for the year ahead. And I'm very excited to tell our audience that you are back with predictions for 2026. In fact, you have not released these predictions yet. Maybe when this comes out, they'll be released. But we are going to step through those predictions. We are going to take a quick look back at the predictions from last year. To be a little honest here, you can't just throw the stuff into the world. The multi-year predictions from last year.
The multi-year you did a plus on it which I thought was cheating and we're going to talk a little bit about media and other things given you know as we said in the beginning you're now in the media industry maybe let me just start there and then we'll get into the predictions for 26. So you made this big career pivot it felt like a big career pivot going from tech Adobe to now being in entertainment. Can you let us into why you did that?
Yeah sure. Well, first of all, my passion's always been at the intersection of creativity and technology. You, as you know, I founded a company called Behance back in 2006 that was about organizing the creative world at Adobe. I was always working with creative professionals across industries and in the last few years spent a lot of time in the world of Hollywood, which I became fascinated by. This is the stack of technology and people, more people than technology, that creates stories that change our lives, that inspire us. I feel like every great technologist I've ever met says that science fiction is a prototype for the future. I know synthetic biologists that were inspired to go into that field from watching Jurassic Park. This is just such an interesting space and at the same time it's disorganized. It's antiquated. And I became fascinated with how technology can be used to improve the way a studio operates, to improve the way marketing is done, to help solve storyteller and filmmaker problems, and ultimately to help creatives take more creative risk, which I think has always been an interest of mine. And then to go to a brand like A24 that is known for helping people take creative risk and doing new things was just a privilege. So early days but I'm having a lot of fun.
I think it's fascinating. You really are living the forefront of both the opportunity but also really the concern. What is your take today on where Hollywood? There's a lot of hand-wringing. Obviously there's the Warner Brothers, Paramount, Netflix. I won't ask you to go too deeply into that. I know you won't comment, but where are we in your sort of feeling of the future for Hollywood?
Well, a couple thoughts. First of all, I think that technologists tend to run ahead of where the customers actually want to be. A few examples of this that I see in Hollywood: I know technologists that are obsessed with personalized entertainment and a film starring the audience as cameos and people having different endings. And I don't see evidence that audiences want that. I think that audiences want a shared experience that they can all recollect and talk about at a dinner table together. As soon as you saw something different than I saw, we just stopped talking about it. So I think there's examples where technology will be reined back to audience preferences as opposed to technologists' dreams. I think similarly you're seeing the same thing happen with a lot of the new emerging technology for filmmaking. There are these AI-native film studios and this effort underway to just use this new technology rampantly. And I think once again there's a very big misunderstanding. I feel like there are two types of creators in the world. There are content creators who are willing to trade control for speed all day every day. These are the people that make stuff for YouTube and TikTok. They're making ads. They're making short form content for brands. These are folks who are playing the algorithms. They're always up against time. They don't know what will perform and what won't. So they just try everything. And in those instances, content creators who are willing to trade control for speed, a lot of these AI tools are actually great for them because they're literally trading control for speed. Then you juxtapose that with the needs of artists. I work with artists. I work with filmmakers who are never willing to take that trade. They're not going to trade control for speed. They're not going to have the doorknob change colors and the facial structure of actors change. They want to have obsessive control over achieving what's in their mind's eye in a story, and that's what elevates the craft. And it's really interesting. We're starting to see evidence of stories, films, even ads coming out with a behind-the-scenes making of in parallel. When Apple came out with their new TV logo, they at the same time launched this episode on the making of this logo that demonstrated the craftsmanship and the humanity behind it. And they came out with their Christmas ad with these gopher things running around. They again, even Tim Cook tweeted about the fact that there was this incredible set of craftsmanship and crafts people behind it. And why? I think it's because they recognize that audiences, just like they crave stories that they can share, shared experiences, I think audiences also crave stories with meaning and craft. And so I think that the tools for artists have actually yet to be developed. And I'm fascinated by that. And I think that as we think about Hollywood, I think a lot of these sort of sound bites are going to start to be reduced to realities of the differences that I'm talking about.
So you're actually getting into some of your predictions. So you've nicely bridged where we're trying to go here. These are you have such an interesting vantage point that your predictions are very thoughtful whether they're right or wrong. They're very thoughtful and they're very helpful I think in crafting thematics which we like to do here at Lion Tree. So you just talked about the proof of craft content. Which essentially was this idea that people want to see behind the scenes. And essentially touch on really verify the humanism of the content by creating the shoulder cut. Like it wasn't a machine that made this ad. It was people doing something. And so tell us a little bit about that and how that will become mainstream for advertising and entertainment. The analogy here is Canal Street and Louis Vuitton purses versus the real thing or Birkin bags. Whenever anything is commoditized, clothes, alcohol, cars, people end up seeking the more scarce meaning-infused version of that thing signaled by a brand that LVMH is essentially in a business of commodities and yet they're commanding a pricing premium because they have brand designating quality. So now that content has become commoditized to some extent, anyone can make something. I think then there's a question of will people have a higher bar for what they're looking for as audiences which will then raise the bar of what's required on the production side which requires more craftsmanship and will people actually seek content signified by a brand. When you look at streaming services like HBO and Apple that are clearly aspiring for a brand signaled degree of quality, is that a good bet or a bad bet? I think it's a good bet.
And I think that we just have it's like a natural human tendency for us when something is readily available to crave the more scarce and meaning-infused version of it. So, when I talk about proof of craft as an idea, going into 2026 and beyond, I think we're going to see a lot more flex from marketers and storytellers of what went in behind it so that the audience appreciates the craft and truly understands that this was real. The equivalent of the stitcher in Italy in that beautiful little factory making your purse.
And also, I mean, the power of editorial, too, right, in the sea of content, the pacemakers need more. All right. So, that was one of your predictions. Another one is that, and this is a big one, is that there are massive amounts of talent arbitrage. Yes. And you talk about for those who are AI native essentially younger generations at this point that is a big advantage and not oh watch out your job is being taken. Tell us about that one.
Well in some ways this is a contrarian and optimistic view of which I aspire to be both right. In this instance everyone's saying oh my god entry level jobs are gone. Why would someone need to hire someone right out of college when they can have the work that they would do, which is typical menial work that doesn't require a lot of skill because they haven't been on the job for so long. Why would we want them? And I go back in my head to the early 2000s when social media was clearly the new way that advertising was going. And these days companies probably spend more than 50% of their budget on social ads, right? But back then CMOs of companies like GE, I remember Beth Comstock at GE and others were just trying to grapple like what is this stuff? How would they learn? They wouldn't go to their vice presidents of marketing. They would actually hire these 20-somethings who in many cases created these digital agencies that became great companies. And these folks would go in with a degree of knowledge arbitrage because they were native. This is what they had in college. They knew more than any of the superiors in the marketing world. They would run circles around everyone else. Here we are again. AI is a big thing, but you go into big companies and they're still trying to hire SEO people. It's not about SEO. It's about AEO, answer engine optimization, right? All the aspects of crunching large amounts of data, optimizing for discovery on ChatGPT, LLM type experiences. This is all a new craft that people in their 20s are native with. And the folks who've been in the careers for and are the experts don't even understand. So my encouragement to anyone listening who is either in this demographic or has nephews or kids or whatever is like encourage them to leverage knowledge arbitrage in this window where they know more than the people that have more experience.
And you're right there is a precious window for young or overlooked talent to gain an advantage as early adopters.
Yeah and it's optimism among younger people. Yes. Okay. All right. Number three, hardware becomes a more popular moat. Hardware becomes a more popular moat along with a surge of hardware startups. Now in our world, from the banking perspective, hardware has definitely not been something people and the fact is is a discount if you have hardware involved. You're saying, uh-oh, actually not uh-oh, look out.
Something's about to change. Well, let me make the case, right? Also, as an angel investor, I was always taught to avoid hardware startups at all costs. Right. For sure. Hardware is hard and again like the discount as you're describing. Why though? Because the stack of designing, making, distributing hardware has actually always been extraordinarily hard. Fast forward, what do we see now? There are startups like Flux that make chip design more scalable. It's almost like a Figma for chip design. And they also do the automated pricing and quoting and manufacturing orders for the user. And then you have fulfillment as a service. You have hardware 3D printing for prototypes as a service. All these parts of the stack that were extraordinarily hard to build much like even having payment providing was before the age of Stripe. I mean you used to have to actually build your own store to be an e-commerce store and now it's like everything as a service. So that stack of services to enable hardware design, development, prototyping and distribution, manufacturing, fulfillment, those are now becoming automated and therefore the friction's gone down. Now couple that with the idea that oh my goodness, everything is being replicable through this age of vibe coding software. If you are a pure software play, how many years are we away from you being able to just say to an LLM, 'Make me that software' and it just boom makes it. I think we're getting closer. And as a result, having that tightly coupled hardware software solution that does something unique for you. I just bought a product called The Board, which is this new game for your family that is actually a physical board with a screen for multiplayer hands, but it's actually like a board game. So, it's using a screen to get your kids off a screen, which I'm still grappling with, but it's amazing. And it downloads the software. It's tightly coupled with the hardware. You and I are both wearing Whoops. And there's Auras and all these other these are all hardware software products and that's going to become a moat that I think more and more startups are going to chase and now actually can based on the stack that's changed underneath them.
So, I'm with you and we were fortunate to invest in Whoop fairly early. The mirror some people early. And obviously Apple's the OG in many ways of tightly coupling and other things. So, it's interesting that that comes back. But at the same time, your number four here is data becomes a less valuable moat.
Data becomes a less valuable moat. And I think that's almost in contrast to the hardware because I would argue that the Whoop and the Aura and these things are basically data capture devices and yet you're saying data is less valuable. Tell us about that. First of all, I was trying to just suppose a better moat of hardware and a declining moat of data. Okay.
And I do actually think that data is becoming less of a moat and hardware is going to kind of become more of a moat. Now, why is data becoming less of a moat? You have the advent of the MCP protocol where anyone can basically link different data sources and different underlying APIs that go to models with other models. You have this very quick degree of connection between different services. Obviously you have all these connectors that every company is making. I'm on the board of Atlassian and we have probably 30 to 50 connectors for all of your different sources of work and so does Slack and so does Glean and so does every Microsoft and everyone's making these connectors and at some point everyone's going to have everyone's data. Now, that's driving one thing that I would call the data wars, which is some sort of negotiation of wait, if you're taking mine, I should take yours. We're going to see more of that. But I'm fast forwarding and also recognizing that eventually we're going to have vision models on our computer that are going to be able to grab the data anyways. So even if you say, 'No, I'm not going to let you sync my data.' Companies are going to be able to grab it anyways. So at some point, if you go to the computer vision level, everyone's going to have everyone's data. So then the question is if data is no longer a moat because it's everywhere what does the moat become? Now I think on the data side it becomes the graph that explains the connection between sets of data that is more proprietary. For example, at Atlassian, their team is building the teamwork graph. Their view is that even if you have all the data of Jira tickets and stuff, what's really useful is knowing who works with whom in the organization on what project and who's allowed to see what permissioning data. Google, Microsoft, these are graphs that are proprietary to organizations that you can't just download. Social graphs, friend graphs, these are the graphs that are going to become more important as everyone has the raw data. So hardware up, data down.
And also related to your comment on data essentially that anyone could have the data your number five is ambient listening and summarization will go mainstream.
So essentially you're just saying there's more inputs of data we're getting data everywhere. Any place you look, there's data coming in. We better capture that. Talk about that one.
Yeah, I like this one because this is one of those that still sounds weird and scary to most. The idea that you would be recording yourself and everyone you encounter all day, every day feels like a blatant violation of others' privacy. It feels weird. It feels big brotherish. It feels risky. And yet, when you actually start to play with some of these tools, these little devices, and this is back to the hardware as a piece, dependence, the companies are coming out and whatever else. These are tightly coupled systems that are coming out where they're recording, but then they're actually not keeping the recording. They're just summarizing and keeping the main points and the takeaways from your daily actions. Increasingly, these are going to be happening on very private locally run models on our device. And at some point, there will be zero risk in the sense that they don't live in the cloud. They live on your phone just like everything else in your life. It's encrypted the same way everything else is on your phone and it's protected by a password and at that point we're going to feel disadvantaged without that experience. It's interesting. I don't know if you've ever served in a jury before. Have you?
I have. Yes, I'm a citizen.
So, I'm glad you didn't get your get out of it. I found it fascinating.
And one thing I really was humbled by in that experience was the jury room. We would be debriefing what happened that day and someone would say, 'I don't remember that.' And then you'd have the court reporter come in and we were all like, 'Oh my gosh.' Like we all kind of invented that part of the narrative in our head. And I just was like, 'Wow, the biases at work every day and the things we kind of make up and fill in the blanks in our brains. What if we actually had this record? How much more accurate and insightful and self-aware would we be?'
It's just crazy. I'm laughing because the implications are just crazy. Like it's essentially a wiretap 24/7.
It is. But it's also civil liberties. So let's talk about how to make this work, right? You got to first of all, you got to discard the recordings. You just have to keep the summarization, right? You have to keep the summarization. The summarization should be done in a way that is in a discovery process. It's all conjecture. There's hallucination involved, right? So I'm just saying people have to feel comfortable with this, right? Technology and the document retention policy, if you will, has to be there to make people feel comfortable using things without being worried about being taken out of context.
But that's sorry, Scott, but that's for you to delete. What about me having this conversation with you? You've captured it, you've edited it, whatever. But you've captured me. Do I need to opt into allowing this happening?
And this is a design question, right? First of all, it wouldn't say Alex, it would say speaker one or speaker two. Sure. And essentially if you look at if you play with some of these products which I have it's like it's as if I were taking copious notes right so I'm in our meeting I'm not word for wording you I'm just sort of saying the general summaries and takeaways but my macro prediction here is that as these models become local and private and summarized only and recordings are discarded and the devices get perfected I think we're all going to need this and we're all going to live on it.
Does this make the iPhone let's do the iPhone for this purpose. More valuable or less valuable in the face of these other devices?
It's a great question because it's theirs to lose, right? To me, there is a major shift happening which I also chronicle here of in one of my predictions here of the shift of power going to these tightly coupled hardware.
That's the next one. In fact, so the answer to your question, you know, is we should get you in the next one.
Yes. The power in consumer AI will shift to tightly coupled hardware and operating system providers.
Yeah. So this is I got another maybe controversial suggestion here and I know there's but not necessarily contrarian. I think that a lot of folks are saying, 'Wow, so the small language models and the ability for AI to operate locally, these and open source models you can download and operate locally are usually a six-month lag in terms of their capabilities to the frontier models in certain instances. As more of our use cases fall below the frontier of what they're capable of, we're going to be able to go and do more things locally.' Now, that's great. Just so I understand, what do you mean by locally versus not locally?
Locally means it literally is not happening in the cloud. It's happening on your device with a model that's running on chips on your device from probably, it's less Nvidia. It's more micron. It's a different stack of chips and the way AI is managed that allows it all to happen on device without any cloud call. And as a result of that, it's fast, it's free, and it's private.
And how could that not be fielding 80 to 90% of the things we need AI for in the next, within 5 years. And it's really what companies are best positioned for that. Apple, Google, is OpenAI coming out with a device? We'll see. But certainly if they do, them as well. As well as all these startups with hardware that we're talking about.
This is non sequitur, but as I was thinking of the local versus into the cloud and sharing the copyright elements of all that, there was the announcement of Disney and OpenAI. I just I'd love your quick take on what you felt about that deal. Disney invested a billion dollars in OpenAI. They essentially allowed their IP to now be scanned sourced by OpenAI. Do you think that was a prudent move for either both? None. Yeah, I I'm like trying to wrap my head around it because I think it's a limited amount of their IP that's older and I think that if Sora is trying to be a consumer product, I do think that people love playing with IP and I think that in some ways, there's a good argument these days that in the world of AI, IP becomes more valuable because everyone wants to play with it. And despite the technology, there's still copyright law and so you have to be allowed to. And in that world, IP becomes more valuable. So, I think this is one of those first deals that represents the likelihood of IP being valuable for consumer AI. And we'll see if people want to play with Mickey Mouse.
I probably and what happens, what Mickey Mouse ends up doing, right? Well, that Disney has been litigious over the years around its IP as it should be. And it'll be interesting what they allow. Let's do your last two here and keep moving. But these are really good. And I like these. I may have liked these the most actually. So number seven, the end of waste will become a new benefit of AI as adoption transforms resource utilization and predictive analytics in ways that will materially reduce wastage across industries, curb climate impact, and improve margins. Basically, AI reduces waste.
Yeah. And I think that we're in this phase of a slightly mature phase of AI being used for tools for different industries. And I'm seeing more startups now that are either focused on restaurant procurement or helping big waste management companies optimize their routes leveraging the data of pickups of waste to determine when to actually send cars and when not to. We were getting into this phase where all this data which is more ubiquitous now to our earlier point can then be leveraged with AI to really optimize in surprising ways across industries. And so one of the outlooks I have here is just the amount of waste in excess inventory sitting in warehouses, extra stuff being shipped to warehouses that don't even need it, extra food being ordered by restaurants that's discarded at the end of the night. I mean, if you go through every
In industry, there's always this margin of stuff that is still human-managed or in Excel spreadsheets. Now, as AI tackles it, it's going to make a material dent. What does that do? It increases margins with less wastage, excess inventory, excess shipping, and everything else. It's certainly better for the environment because of all the transportation, throwaway, methane, and whatever else. I think humanity is going to become more lean. It's a cool concept to think about, and my optimistic tendencies creep in.
This one is perfectly intuitive. It's just how far and how wide this gets adopted. It clearly makes sense, especially if there's motivation behind it. If a restaurant can save money by reducing its waste and getting the supply-demand right, it would behoove them to tap into AI and do these things. It's just a question of the adoption curve.
And who is supplying that ultimately?
That's right. I think the adoption is obvious because everyone's incentivized. You're saving money, it's good for the environment, it's good for everyone. It also goes to the point about margin expansion across industries over the coming years.
We are worried about jobs.
We are worried about jobs because people need income to buy things from these companies. But at the same time, these companies are going to increase margins in two ways. One is through reduced wastage and better optimization of everything. The second, which is unfortunate in some cases, is there's just going to be more talent density. I think we're probably several quarters away from hearing so many companies reporting their earnings and flexing the fact that they're decreasing their workforce, which is going to be a very strange world to live in. I always worked under a CEO who said it's a failure of a CEO to have to do layoffs; it's bad planning. I think we're going to get into an era where a company flexes its resourcefulness, which is going to be a strange conundrum for the economy.
Right, as we see, everyone's going to use AI. So finally, number eight: hospitality becomes a differentiating factor of commerce. Tell us about that one.
This is an evolution of one we discussed last year, and I've been thinking about for years: this wave of personalization. The cardinal rule of technology that's great for consumers is that it takes us back to the way things once were, with more scale and efficiency. For almost 300,000 years, we lived in villages and small groups where everyone knew each other. There was a comfort in being known and recognized for what you were good at. Only in the past few hundred years, with the industrial revolution, did we lose that. Everyone became anonymous. The internet was born with the same idea: it doesn't know who you are. Nike.com treats you as a random person, restaurants don't know your allergies. We long to be known and treated special. I've been thinking about how AI and personalization through technology can make this happen again. I believe one word for that is hospitality, beyond just hotels and restaurants, but also in commerce. What does it feel like to be greeted by name in a store or digital experience? How do humans get redeployed for that? If the back of house becomes automated, maybe those humans are deployed to the front of the house, giving us better service, more familiarity, and more human experiences. That could differentiate brand experiences even more.
We get better service and more familiarity. That ends up differentiating brand experiences even more than before.
That's part of why it works. We love the warmth. Danny Meyer talks about hospitality for everyone. You're going to need to log in for that environment digitally, or do you think it's just ambient?
What is the technology that enables this?
There are companies building what I call the memory layer, allowing memory to be portable across different agent experiences.
Essentially. Imagine going to a digital experience, giving it permission to know you, and everything is synced.
You have this experience.
Can we get rid of passwords and all that stuff?
We'll see where it goes, but it would be nice. We want to be known. I believe that.
I believe that as well. Everyone wants to be special.
Everyone wants to be special. Those were your eight predictions for this year. We'll check back in a year. But speaking of checking back, I'm going to quickly run down your six predictions for 2025. I don't even remember what they are.
I'm going to run them down very quickly, and you're going to tell us what you got right and what you maybe got wrong.
Not that you get so much wrong, but everyone's human. So 2025 predictions: one, betting markets as truth engines. Betting markets got hot this year. There's something there.
They are being cited more for what people actually believe and what actually happens. But it hasn't entered mainstream news yet. I think it's still going.
You don't think it has? I feel like it has. I'm a big New York Giants fan. They talk about this coach being the front runner based on the predictor. It's cheap, but it'd be cool to see a story someday that allows people to bet on whether it's true with some sort of verification.
The fake news betting market. Okay. Number two: do-it-yourself agent-built software everywhere.
There are startups now allowing you to create agents that go through your email and do things for you. But none of them are really there yet.
You're on the right track, just early. Number three: AR on a slow boil then a step change.
I feel like augmented reality has come back to earth a little bit. People aren't talking about it.
