Free with a sign-in.
The best 16 min of 43
Runway’s Cris Valenzuela: Building The Hollywood-Friendly Video AI That OpenAI Couldn’t
Cristóbal Valenzuela argues that video-based world models can support creative tools and applications beyond filmmaking, including robotics.
Upstarts · Alex Konrad
- Cristóbal Valenzuela is co-founder and CEO of Runway.
The best minutes
- 20:02Runway kept going when investors said generative AI was not a market5 min
- 25:19Build for changing models, not today’s limitations5 min
- 6:11Runway’s founders treated early AI as a creative medium7 min
Also worth watching
- 15:29A single editor showed how Runway could fit into TV production2 min
- 38:56Video models could train robots and power real-time characters4 min
- 13:00Creative satisfaction can make AI tools worth paying for3 min
- 30:13Runway gives amateur filmmakers access to the same tools as top directors3 min
- 32:56AI misuse is the creator's responsibility, not the tool's3 min
Skip the opening company and world-model setup, the audience-reaction discussion, the sponsor read, and the later access-to-creation discussion.
Full analysis · 6 min read
Real Alpha · Nov 8
Cristóbal Valenzuela says Runway's festival drew 6,000 submissions
- Runway's AI film festival drew 6,000 submissions last year, Valenzuela said.
- That signals creator participation, but the episode offers no revenue, retention or unit economics.
- For buyers, test workflow gains now and treat robotics as an option, not proven returns.
Free with an account
Read the rest of the analysis
6 sections, the 6 other voices it tests Valenzuela against, and what it means for you. Every episode we cut gets one.
Check your inbox
We sent a link to your address. Click it to finish signing up. Not there in a minute? Look in spam.
or
Signed up before? The same buttons sign you in. The episode, its moments and the transcript stay free.
Runway's festival shows reach, not a paying market
Valenzuela said Runway's festival received 6,000 submissions last year, from around the world. He described them as AI-generated or AI-made films.
The count measures submissions to Runway's event, not necessarily Runway product users. That distinction matters because the number signals reach, not customers.
He said a musician won last year's festival and then got a deal to make more. That is a plausible route from access to paid work, but one winner cannot establish typical outcomes.
The episode gives no follow-up on the deal's scale or whether the musician used Runway commercially. So the example illustrates possibility, not conversion rate.
The count supports Runway's visibility among people making AI-assisted films. It does not say how many submissions came from unique creators, how many paid, or whether they returned.
Without that denominator, 6,000 cannot establish demand's size.
Our read
The figure is evidence of creative participation, not economic validation. Treat it as an adoption lead to verify, not a market-size estimate.
Creative rewards can broaden demand beyond commercial work
Valenzuela's consumer case is emotional utility, not only production savings. He compared making with AI to a gym for the mind, where the reward can be creative practice itself.
That argument matters because willingness to pay may precede commercial use. Yet his metaphor is a product thesis, not retention data.
He said engineers and finance workers can have ideas but lack technical experience or resources to execute them. Runway's pitch is that its tools lower that execution barrier.
The buyer still needs to test whether finished work earns attention or saves time. For example, a late-night editor used Runway to edit scenes in minutes, work that otherwise could take hours or days.
At first he hid the tool from colleagues, then disclosed it, and the show became an early enterprise customer. That links time saved with a company buyer, but the episode offers one anecdote, not repeatable savings across a team.
Measure the result against review time and recurring use.
Our read
Personal satisfaction may widen the market, while measured production gains offer a clearer budget case. Runway needs separate evidence for each.
Control may distinguish Runway, but evidence remains thin
The host described OpenAI's decision to sunset Sora as a miss. Valenzuela responded that video products are hard to build and consumers are hard to serve.
The exchange does not establish that Runway beat Sora on quality, adoption or costs. Valenzuela said the level of craft and control needed in the tools is critical.
That matters in creative work, where users may need repeated direction rather than a single generated clip. But the episode offers no comparative user study to show this control improves results.
Neal Stephenson described a Fortnite project on Core Memory in October 2026. AI helps with scripts and characters, while Unreal Editor for Fortnite keeps visuals on players' devices.
That is a loose analogy for keeping creators involved, not evidence about Runway or Hollywood.
Our read
Control is a credible product distinction, but not a demonstrated advantage over competitors. Buyers should test whether it improves output or approval time.
Model churn makes adaptation part of the product
Valenzuela said it feels like new model standards arrive every couple of weeks now, after months or years between shifts. Products built around today's model weaknesses can become irrelevant after a leap forward.
That's his warning, not a measured estimate of how often Runway itself rewrites products. His proposed response is organizational as well as technical.
Teams need to drop sunk costs, learn from discarded work and keep a long-term direction. But repeated replacement can frustrate engineers and consume time, so adaptability needs clear goals.
Harrison Chase, on 20VC in October 2026, described LangChain deleting features as models absorb them. He distinguished those temporary features from APIs and proprietary data.
Guillermo Rauch, on Sourcery that month, described the challenge of explaining a changed product vision to employees. Together, their examples make Valenzuela's principle operational: decide what is temporary, then explain changes clearly.
“they spend too much time on building the scaffolding and the processes and the orchestration for the limitations of the models”
Harrison Chase · 20VC · Oct 17Models make product layers obsolete, so build around durable needs
Guillermo Rauch · Sourcery · Oct 16Leaders must explain when their product vision has changed
Our read
Runway's lasting advantage may depend less on any one interface than on how quickly teams can replace it. The episode offers a principle, not proof that Runway executes it better.
World models widen the bet and raise proof burden
Valenzuela sees video models as more than entertainment tools. He said scaled models may learn patterns of physics, gravity, occlusion and movement by observing video.
He sees that as a path toward robotics, autonomous systems and real-time interactive characters. That is a larger ambition than editing, but it raises a heavier proof burden.
The episode names no robot task, benchmark or result showing Runway's models improve a deployed system. So robotics remains a proposed extension, not a demonstrated product.
Dave Blundin discussed visual planning beyond games on Moonshots in October 2026. His panel's examples include training, design and management, supporting the broader possibility without validating Runway.
Josh Gao, on Training Data in October 2026, expects failures and corrections from deployed robots to improve later systems. That is a useful test for any robotics plan, because claims need a connection to operating feedback.
For Runway, that still means defining a specific task and measuring transfer. Valenzuela also said Runway had released a system for real-time, open-ended avatar conversations, but gave no usage or retention figures.
Dave Blundin · Moonshots · Oct 31World models could make visual planning useful across work
Josh Gao · Training Data · Oct 20Robot integrators can improve installed bots with data
Our read
World models add meaningful upside only if performance transfers into specific tasks. Treat robotics and interactive characters as separate bets, each requiring its own evidence.
Runway's business case remains unpriced in this conversation
Valenzuela's CEO role matters when weighing these examples. He is making the case for Runway's own creative products and model roadmap, which is relevant context rather than a reason to dismiss them.
The host asked whether Runway could be sustainable amid AI pricing concerns. Valenzuela said serving tens of millions or hundreds of millions of users creates engineering challenges, but called them solvable.
He supplied no revenue, margins, inference costs, customer retention or payback periods. Without those figures, the episode cannot show whether creative interest supports durable economics.
Cal Newport, on Prof G Markets in October 2026, said enterprise AI investment was heavy while returns remained mixed. His comparison reinforces the gap between experimentation and measured operating value.
For teams considering a similar tool, compare time saved with review effort, repeat usage and total cost. For example, track personal paid use separately from festival entries, novelty trials and enthusiastic feedback.
Our read
The episode makes a case for user value, but not for unit economics. Budget decisions should rest on measured workflow results, not the festival figure alone.
What it means for you
- Measure editor hours. Run a controlled trial on one editing workflow. Track hours saved, review time and repeat use against the current process.
- Track creator returns. Separate first-time generations from repeat sessions and paid renewals. A festival submission count cannot answer whether users return.
- Test creative control. Ask creators to complete the same brief with one-shot generation and guided tools. Compare revisions, approval rates and time to final output.
- Gate robotics experiments. Require a named task and deployment benchmark before allocating a robotics budget. Track whether real-world corrections improve later performance.
What to watch
- Whether Runway reports repeat creator use or commercial workflow results.
- Whether its world models show measurable gains on a named robotics task.
- Whether rapid model changes require repeated product rewrites or improve shipping speed.
Send this to product leaders evaluating generative media workflows.
Watch the minute where Valenzuela gives Runway's 6,000-submission figure
Written by Real Alpha from the episode's transcript and the other moments it links, then checked against the transcript. Quotes are the speaker's exact words.
Minute by minute
- 0:49Runway began at NYU in New York
Valenzuela says the three founders met at NYU and began Runway while studying in New York. He describes it as an applied research company that trains video and world models, then builds products on top for entertainment, gaming, robotics and other industries.
- 2:23Video models simulate the world
Valenzuela contrasts language models, which he says describe reality through language, with world models trained on video that can learn by simulating how things behave. He says Runway sources video data through industry partnerships, purchases and collection, depending on the task.
- 4:35Video products need craft and control
Valenzuela says OpenAI’s Sora was the first direct competition Runway felt it had, and that building video products to meet user demands is difficult, especially for consumers. He says Runway focuses on craft and user control so people can shape the output through their process.
- 6:22Art and engineering shaped Runway
Valenzuela says the founders met at NYU’s ITP, which he describes as a mix of art and engineering. Early experiments included a visual search tool that matched a street recording to similar shots in films; those prototypes helped lead to a product and company.
- 10:33Creative voices should shape AI
Valenzuela says the early reaction to AI tools was excitement about making things that had previously been difficult or impossible. He argues that creative people should help shape these systems, and recalls building a browser-based writing tool that completed sentences in the style of authors he liked.
- 13:00Creative work rewards users
Valenzuela compares creative AI tools to a gym for the mind: people may use them because making something feels rewarding, not because they plan to sell it. He says engineers and people in finance use Runway to realize ideas they could not execute with their existing skills or resources.
- 15:55Faster editing found an enterprise use
Valenzuela recalls someone at a late-night show using Runway to speed up work without telling colleagues at first. The tool automated editing tasks, including removing things from scenes, in minutes rather than the hours or days he says they could otherwise take.
- 17:55Audiences judge the story
Valenzuela argues that viewers care more about whether a film’s story moves them than which camera or editing software produced it. He says AI will become part of storytelling and that people are already engaging with content made or mixed with AI.
- 20:27Startups need non-consensus conviction
Valenzuela recalls showing his team investor emails rejecting Runway because generative AI was not yet seen as a market or opportunity. He says repeated rejections created doubt, but founders need to stay willing to pursue ideas that are not yet consensus.
- 23:09Teams must adapt as models change
Valenzuela says model capabilities now set new standards every few weeks, so products built around current model limitations can quickly become obsolete. He says teams need to drop sunk-cost projects, stay adaptable and work toward long-term goals; he gives elaborate prompt engineering as an approach that became less important as prompting grew more natural.
- 30:20Video tools widen access to filmmaking
Valenzuela says Runway’s tools can help established filmmakers push their craft and give amateurs access to similar technology, leaving ideas and storytelling as the greater distinction. He says the company’s film festival received 6,000 AI-generated or AI-assisted submissions last year and has helped new creators get opportunities.
- 33:05Creators remain responsible for AI use
Valenzuela predicts that AI will become as ordinary as cameras or the internet. He argues that responsibility for harmful or misleading uses belongs to the people who create them, not to the technology itself.
- 36:12Ideas matter more when tools get easier
Valenzuela says easier tools lower the technical barrier to making things, but leave people with the harder question of what they want to create. He argues that artists should use advanced technology, while their ideas should come from their experiences and view of the world.
- 38:56World models could support robots and characters
Valenzuela says video models may learn patterns of the world, including movement, gravity and how objects interact, which could make them useful for robotics and autonomous systems. He also describes real-time, lifelike characters that can hold open-ended conversations, including one he uses to practice French.
Published May 1, 2026.
Transcript · 39 min read
Upstarts, hosted by Alex Konrad, with Cristóbal Valenzuela.
Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.
It's a dream. It's like you have this thing that basically conjures whatever you want. And the first thing people struggle to do is like, it's not to use the tool. It's like, what do you want to make? You know, what is in your head you want to make? Welcome to the Upstarts podcast, our weekly show where we talk to emerging startup founders about their upstart moment. Upstarts are challengers who punch above their weight and take on the status quo to improve the world, all while building a big business too. I'm your host, Alex Konrad, founder and editor of Upstarts Media. I'm excited to be joined by the co-founder and CEO of Runway, Chris Valenzuela. Thanks for joining us, Chris. Yeah, of course. Thank you for having me. This podcast is brought to you by Mercury, banking redesigned from the ground up. You've historically made these pretty technical video models using AI. You're also based here in New York. We're at the New York Stock Exchange this morning. And you have the distinction of being the only guest that I think in the run-up to the show I've ever run into on the street, because we both live in Brooklyn.
Yeah, we're neighbors. Why are you building this company in New York? We started, the three founders met in NYU and we started the kind of initial beginnings of the company while we were at school in New York. And so it was very natural for us. It wasn't typical back then, to be honest. It was just like eight years, nine years ago. But I think we just, I love New York. I love the city itself, the people, the culture. It just feels very special. And it's becoming very special for the company as well to be able to be based here in New York. So for people who are not familiar, what exactly is Runway and how do you describe it to people today?
Runway is an apply research company that builds video and world models. And so what a world model is, is kind of an extra tier of AI research and AI capabilities. These are basically models that can understand and take actions within the world, maybe in a similar way that we do. And then the way you get there is by training large video models. And so we've been working on this for quite some time, and now our models are getting deployed in entertainment, in movie making, in robotics, in many other industries, in gaming. And so Runway builds these world models, pre-trains them, develops them internally, and then we build products on top of them.
Read the whole transcript
And so those are kind of the best way to think about it. It's like an apply research company that builds the products, build the infrastructure, and of course, builds the models to make these world models work out in the world. A lot of people are talking about world models in the tech world. For people who are not deeply in that online discourse, why is the world model... model new and so exciting in this moment? A lot of the paradigm of AI that we've gotten into until now is based on this idea of training large language models, if you've probably heard about LLMs, right? LLMs are just large AI models that are trained on language alone.
And so they're trying to get a sense of reality by describing reality. The thing about language is that language is not reality. Language is an obstruction of reality that we humans have created to try to approximate to what things are happening, right? So a glass of water is not really a glass of water. It's like we made up the symbols and the representation of that. And we train models on that language alone. What we do with world models is slightly different. We train them on video data. And if you show video data and models enough kind of information in pixels, there's enough information about what water is and how it should behave in containers, in other spaces, that the model starts to understand slowly.
Not by describing it, but by simulating it. So that's a good way of thinking about it. Language models describe reality. World models simulate reality. Where do you get the data, the video data to train off of? It depends. We need to partner with partners, with industry folks. We have people, we buy data, we collect data. Depending on the task at hand, you might need to source from different points. of different points. It's a very technical challenge to do. What made you guys confident that you could do it better or have a different approach than maybe some of these big famous AI labs that our audience is familiar with?
We started the company in 2018. And I think back then there wasn't a lot of intuition into how to build the models and the kind of models that we're building these days. And I think over time you build some sort of experience and intuition into what are the best ideas and approaches you can take to scale deep learning models for them to work at this kind of pace and level. And I would say that most of the best research labs these days are not, have been working on this for at least a couple of years. And I think that separates the kind of, I would say, the best from like the new, the kind of like the ones that are trying to build the best models.
And I think for us, that's been the lesson. We've been working on this for quite some time and we have a natural sort of research ideas and approaches and experiences that allow us to differentiate more than others. Now, people are probably aware that OpenAI had this consumer video app that they launched a great fanfare and then they recently sunset it. It was kind of actually a miss that OpenAI sunset Sora. When we saw it first coming up, it was the first ever real competition we felt we had. And now, like, two years after they announced that they shut it down, a few things are true. One is, like, it's very hard to build the kind of products and models that video requires and the users kind of demand. And specifically, consumers are just very, yeah, very hard as well.
For us, it's always been around, well, we're building towards a world where we think this might become the foundational base on which you make every piece of content out there, from ads to media to films to things you share with your family, with friends, or just things you make for you. And so the level of craft and the level of control that you need to have in the tools needs to be critical. When people think of Runway and people use Runway, one of the things that keeps coming up as a core differentiation is that it's not just like one prompt and you kind of make a thing, it's you're in control.
This is a tool and you kind of like wield it in the way that you want and the outputs are a reflection of your process. And I think the way you think, we think about it might be different from how others might have thought about it because you can see how different like building of products and models is. And I think from, at least from us, from our professional enterprise consumer side, which is a lot of what we're focusing on, that really reflects really well. And people come back to us because of those kind of things. So the initial idea, it comes out of this kind of weird lab at NYU that is a mix of
“it's really important to have an open mind and be kind of less obsessed with what you know and more of what you can discover.”
I guess I'm trying to picture you create these visually impressive videos that maybe are in the manner of a movie or something.
And researchers are playing with this. How do you start making money? Like who ends up being willing to actually pay for what otherwise is like a really cool thing to post on social media or something?
“i like to think of sometimes these models and the things that we're building as kind of a gym for the mind where you are exercising your creative muscles”
Was it sort of like, hey, you can take this beyond that fun side project and actually use this in commercial ways or ways that, you know, advance your business?
“he was using runway that was his weapon and then at some point he's like okay i'm going to tell you what i'm using i'm using this thing called runway”
whether it's a tv show or a movie if they told their audiences hey we used ai to help make this scene or these two scenes do you think the audiences would see that as cheapening the process or they wouldn't like it or you think people are kind of proud to say when they're using it and has that changed at all i really want to get and i think we're approaching that time to be honest we're like you don't have to say how you make things right like and if you think about how you consume anything um you never care about what cameras were used what editing software was used to make a movie what was the last time what was the last movie you you you saw i liked project tail mary for the story not necessarily how it came together right and and And people might be, okay, I want to understand the behind the scenes or whatever, but like 99% of the world goes and watches a movie because of the story, right?
It makes you feel something. And if you think of AI as a technology within the evolution of storytelling, dating back to filmmaking even before that, we're reaching a point where I can tell a really good story. I don't have to tell you how I made it. Who cares how it was made? Did it work as a story or not? You're going to judge it not based on the tactical merits of it and the execution of it. You're going to judge it on like, did it make you move and feel something in some way? And so that's how I feel about it. And I think we're now at a point where that's beginning to happen, where people are really engaging with really good content that can be generated, mixed or created fully that really doesn't matter.
But now you're seeing stuff and you're just engaging with it and you like it. And that's, I think, where we're going to go very soon. And to be to be fair, this conversation we had had it many times before in the history of like storytelling and creative, where like we we tend to emphasize for technical reasons the way things are made until eventually you realize that no one cares. Just move on. You know, as a year old startup, we experienced a lot of firsts at Upstarts Media. So when it was time to process our first international wire following a London event, I braced myself. This is going to be a painful lesson.
instead Mercury made the process easy issuing an invoice sharing routing numbers and processing the wire with just a couple clicks no phone calls no paperwork no learning curve we're still figuring plenty of stuff out when it comes to our banking Mercury's already got it all figured out for us Visit mercury.com to learn more and apply online in minutes. Mercury is a fintech company, not an FDIC-insured bank. Banking services provided through Choice Financial Group and Column NA members FDIC.
“And there's always doubt because, I mean, you get 50 rejections, and you're like, okay, there's definitely something odd here.”
Do you think you've already proven this can be a long term sustainable business? Because, you know, there are those macro doubts of, you know, Sora maybe was burning too much money for OpenAI. A lot of these AI companies are facing pricing challenges right now. You know, there's been a lot of brouhaha about, you know, Isanthropic having to change how it's pricing its claw. code because maybe it's too popular is there a scenario where runway is too popular and that's a problem for you as well i mean there's always problems as you scale of course like serving tens of millions and hundreds of millions of users it's just requires different engineering sort of problems but i think most of the things are solvable to be honest with the right kind of process and people the biggest thing for me is less of like how do you solve for those and more How do you not get trapped in like local maximums?
“they spend too much time on building the scaffolding and the processes and the orchestration for the limitations of the models”
I met or at least had one of our first long conversations at a dinner with a famous filmmaker, Darren Aronofsky in L.A. Darren has since done a lot of stuff in A.I., including a series around George Washington where he was using A.I. Do you see Runway and sort of tools like it as most exciting for folks like Darren, these established filmmakers to be pushing the limits of their craft?
“because if you think about it, you have access to the same level of technology that someone like Darren has. So you're basically both starting at the same level.”
kind of go away or what's your prediction for sort of what the state of play will become yeah so it's hard to predict well i guess i guess one that i was seeing on the way here thinking about was controversy around a famous actor being used after their death in a you know
“You're going to blame the intention and the creator and the person who did that. And I think AI is the same.”
make things, you know, globally and not just in a place like Hollywood. But what would be your message to people who do feel like whether it's, you know, that photographer or that, you know, wannabe filmmaker that now it's more about how good you are with these tools versus, you know, do you have the right idea? that basically people would get left behind in that transition. In a way, if you think about creation as a whole, creation is a representation of how accessible the technologies that we have at large are able to allow us to express those ideas, right?
So if you want to code a website or, I don't know, create an application, you had to go through a painful process of learning how to do and code and maybe even spend months and many, many resources on learning. or now you can get to a point where like you can just five code something and get really close to what you want that that feels for me like insane it's like we've normalized the fact that only a couple months ago that would be magic and that allows way more people to express themselves creatively to do more work you can think about that same kind of way of thinking across every kind of human modality in any shape way or form there's less of a barrier technically for you to execute something it's more of like well what's the idea and i see this consistently over and over time where i sometimes get people with like new models and new tools and we give them a preview with something And it's like, it's a dream.
And so you have this thing that basically conjures whatever you want. And the first thing people struggle to do is like, it's not to use the tool. It's like, what do you want to make? You know, what is in your head you want to make? And that's when the conversation around some of these tools, like lovable, the app generators is like, okay, it's easy maybe to vibe code an app now. Do you actually know what app to make? Correct, correct. Correct. And I think that for me is like the most interesting thing where like that goes back to like, it's not about the medium. It's about like the idea that you want to use.
And I really like this. Luke Dubois, who's an artist and a professor at NYU, has this beautiful quote about every artist should use the maximum enough technology in their civilization to make art and to question what it means. I think we're at a point where like everyone who was an artist or who wants to make something should use the maximum level of like technology which is like this frontier models to make anything they want now what do you want to make well that's that's on you and that's a that's the hard part I'm not gonna help you in that because they you need to have live experiences you need to like think about the world there's something you want to see and you want to convey now you can but what is it would it make sense and why would it make sense and so that I think doesn't change and so as you expand beyond video um at least filmmaking use cases into robotics gaming and obviously robotics is super exciting right now in the tech world we see you know vcs lining up we see lots of labs creating um their own models just for robots where does runway naturally fit there and what makes you excited about these use cases beyond sort of the original idea so the
“if you scale these video models to a significant point, these models not only encode storytelling, they encode and they figure out the patterns of the world at large.”
So thanks for joining us, Chris. Yeah, of course. Thank you for having me.
