Invisible Universe and the Future State of Video
What Business Model Emerges from the Video Wreckage?
The future is already here.
-William Gibson (maybe)
Let’s say the quiet part out loud. The video business model is in trouble.
The video ecosystem was disrupted over the last 15 years as the cost to distribute content plummeted, which reduced barriers to entry, enabling new competitors and compressing aggregate profits. Most of the current problems in the industry today can be traced back to the fact that there’s now less money to go around. The advent of GenAI will reduce the costs of content creation, threatening to erode another critical moat and disrupt the industry again. Things will likely get harder, not easier, especially for pure-play video companies.
Over the last year or so, I’ve had umpteen conversations with people who are trying to figure out what emerges on the other side of these disruptions. As entry barriers fall, what will be the business model of professional video creation in, say, five years?
In many of those conversations, I’ve pointed to a small animation studio called Invisible Universe. There are dozens or hundreds of independent animation studios, so many have not heard of it. What makes it interesting is that its business model is built around the likely future state of the video business.
Tl;dr:
Invisible Universe (IU) is an internet-first animation studio that develops animation franchises and distributes chiefly on YouTube and TikTok.
Instead of the traditional model of creating content and then distributing it, IU employs a highly iterative content development process, continuously engaging with its fan base and quickly turning new content. Fans play an active role in plot and character development. This shift—from creating content and then building community to building community and then creating content—is much less risky.
Recognizing that falling entry barriers generally mean lower margins, IU doesn’t view video as a profit center. Video is a mechanism for creating franchises that are monetized through other channels.
Contrary to the perception that GenAI is currently a hobbyist tool, IU has fully transitioned to AI-enabled content generation. Over the last year, it has created its own fine-tuned models, trained on its IP, and developed a custom workflow. As a result, it has reduced development time on its videos from weeks to hours and reduced content development costs by over 90%.
It intends to productize its workflow as a IU Studio that enables fans to create content using its IP.
Will IU eventually become the “Pixar of the Internet,” its stated ambition? Maybe, maybe not. But if you’re in or around the TV business, you should be paying attention. Its business model may foreshadow where video is headed.
What’s Invisible Universe?
Invisible Universe (IU) was founded in 2020 by Tricia Biggio and John Brennan, former Snap and MGM executives, to create a next generation animation studio, with backing from Initialized Capital, Seven Seven Six (Alexis Ohanian’s fund), and Dreamers VC (founded by Will Smith), among others.
Figure 1. Invisible Universe Animated Franchises
Source: Invisible Universe.
It publishes on social, primarily TikTok, YouTube and Instagram, and currently has seven animation franchises. Most of these are associated with a celebrity. For instance, Qai Qai is its most popular franchise, with 5 million social followers and over 1 billion video views. She originated as the doll of Ohanian and wife Serena Williams’ daughter, Olympia. Other celebrity affiliations include Jennifer Aniston, the D’Amelio family and Karlie Kloss.
What makes IU especially—and broadly—interesting is its business model.
An Iterative Content Development Process
The chief insight underlying the founding of IU was that social made it possible to develop content in a far more iterative and less risky way.
Traditionally, production and distribution of content have been two distinct steps. Studios spend a lot of time and money developing a series or movie—for a AAA project, tens or hundreds of millions of dollars—then they spend a lot more money to market it and persuade people to show up. Social is used as one of several marketing channels. When used well, it is both a paid and earned channel; the best social marketers can create buzz that propagates organically.
IU uses social as a focus group and co-creator, not just a marketing outlet.
IU does something different. It turns animation very quickly, incorporating current events and social trends (like Qai Qai dancing to Megan Thee Stallion’s “Mamushi” within days of the song blowing up). Its community managers engage closely with the fan base, monitoring what’s working and responding to comments (as the characters). New plotlines, characters and character development often originate from fan comments—for instance, Qai Qai’s relationship with her dog and the existence of her sister both emerged out of the comments. IU also teases new storylines in the comments to get fans excited. They also incubate new IP and, through this feedback mechanism, develop what works and quickly kill what doesn’t.
Building a community and then creating IP is fundamentally less risky than creating IP and then trying to build a community.
In other words, IU uses social as a focus group and co-creator, not just a marketing channel.
This approach—creating the IP around the community, as opposed to trying to build the community after creating the IP—is far lower risk. It also creates fans who feel a deeper connection with the characters and story.
Video is Not a Profit Center
Over the past year or so I’ve connected with a lot of people who are either launching or contemplating launching a next generation studio. The idea is always to leverage GenAI tools to create content at a fraction of the cost of traditional production.
Whenever I talk to anyone contemplating launching an AI-first studio, my question is always the same: how do you make money long term?
Eventually, I come around to the same set of questions: If GenAI is lowering the barriers to entry for you, isn’t it lowering them for everyone else too? Why isn’t video content creation going to be a red ocean? Plus, lower cost production doesn’t pay off in a cost-plus licensing model. Netflix line-item audits every production budget, so if you can make a $100 million series for $10 million or $1 million, they’ll know it, and a 20% cost-plus premium on a $10 million project is a lot less than a 20% premium on a $100 million project (er, it’s 1/10). How does this make money long term?
I’ve heard a lot of thoughtful responses, such as those who believe their IP, proprietary data or technology will provide a moat. IU’s answer, however, is that it doesn’t need to make money in video. It doesn’t view video as a profit center, but rather as a mechanism to create franchises that it can monetize in other ways. For instance, Qai Qai has a doll, a children’s book, just signed a record deal with Universal’s Republic Records and has a series in development with an unnamed streamer.
The concept of 360-degree (or omnichannel or multi-platform, whatever you call it) monetization obviously isn’t new; Walt Disney drew his famous diagram in 1957 (Figure 2). What is new is the explicit acknowledgement that, in the future, video may not be a chief source of profits.
Figure 2. Walt Disney Had it Figured Out 70 Years Ago
Source: Walt Disney.
Commercializing GenAI Now
Perhaps the most interesting aspect of IU’s model is that all its content today is GenAI enabled. (By “GenAI-enabled,” I mean humans working with AI tools.)
The common perception right now is that AI video is currently a hobbyist’s tool and isn’t ready for primetime. There is a large class of creators who are using GenAI to make fake movie trailers and experimental videos and posting them on Twitter and Reddit. These are cool and—relative to the state of the technology just a year ago—pretty amazing. But, putting aside the novelty, these videos aren’t generally something you would sit down with a bowl of popcorn and watch (something Mike Gioia explores here).
At the other end of the spectrum, the major studios are proceeding cautiously. Many are experimenting with AI in some parts of their production processes—such as using AI image generation for rapid, inexpensive creation of concept art or exploring using it for localization services (“subbing and dubbing”)—but they’re far from using it to replace or augment principal photography. As I explained in a recent piece (see Fear and Loathing (and Hype and Reality) in Los Angeles), there are three big hurdles: labor relations, legal questions and technical standards. They are very leery of the perception that they intend to use AI to eliminate jobs (especially on the heels of the writers and actors strikes last year); they’re stymied by unresolved legal issues around GenAI, especially questions of copyright infringement and IP rights; and AI video generators can’t yet deliver the kind of precise control that top directors/showrunners/cinematographers expect.
The common perception is that AI video is a hobbyist tool and isn’t ready for primetime.
IU’s use of GenAI, however, shows that there is a middle ground between these two extremes: small producers without the same constraints as the major studios, who are already putting it to work. Another example is Toonstar, another independent digital animation studio that is using GenAI extensively in its productions.
Not surprisingly, GenAI is being commercialized first in animation, particularly kids’ animation. With animation, there is no uncanny valley to worry about and kids have lower standards for production quality. (No offense to all the kids who are avid readers of my posts.) See: Teletubbies and Power Rangers.
How? IU spent much of the last year fine-tuning open source video models (first, Stable Diffusion, now Flux) and LLMs on its own IP (characters, movements, environments, backstories, mythology, etc.). They also developed their own custom workflow (called IU Studio), which enables them to rapidly go from ideation to completed video: LLMs that act as creative co-pilots; auto-generated storyboards; image generation; image-to-video generation; and an editing platform where they pull it all together, including sound.
Videos that formerly took weeks can now be completed in hours and production costs have declined more than 90%.
Through the use of these tools/workflow, IU has transitioned to entirely AI-enabled content. Videos that previously took 10-12 business days now take several hours and production costs have declined by over 90%. (Toonstar has also stated that its production costs are 90% less than traditional production processes.) Both data points suggest that Jeffrey Katzenberg’s prediction that AI will cut the cost of animated production by 90% was spot on.1
The Next Step: Enabling Fan Creation
IU intends to productize IU Studio to enable fans to create content using its IP. The idea is to create specific games and challenges (like “Ember is going on a date, show us what happens!”).
As I’ve written about before (see IP as Platform), the evidence is clear that fans want to create content using their favorite IP. The volume of fan creation in different media is gated by the accessibility of the medium, not the desire. Since most people can write, there is a massive amount of literary fanfic on platforms like Archive of Our Own and FanFiction; song covers are less accessible because you need some musical ability, but there are countless covers on YouTube and TikTok; there is a little less game modding, since it often requires some level of coding knowledge; but there’s very little video fan fiction because it’s so hard and time consuming to do. GenAI has the potential to change that.
Over time, as GenAI makes content creation more accessible, the line between consumption and creation will likely increasingly blur.
Hollywood takes a very precious view of its IP and the concept of enabling fan creation may seem radical or worse. However, over time, I believe that the line between consumption and creation will continue to blur, particularly as GenAI makes creation more accessible. Progressive media companies will figure out how to embrace this and turn it to their advantage. Doing so, with the right guardrails in place, should create more ardent fans and attract new ones; provide free marketing; and possibly source new stories and talent.
The Future is Already Here
Will IU become the “Pixar of the Internet,” its stated ambition? Maybe, maybe not, but it’s worth paying attention either way.
It’s hard to see inflection points when you’re in the middle of one, but we’re in one now. It’s a pretty good bet that media and entertainment will remain an inextricable part of our lives. But as entry barriers to creation fall, over the next 5-10 years the business model of creating and delivering this media will probably change dramatically. No one knows with precision how these new models will work, but IU is one of many efforts around the periphery of established media, trying to figure it out.
For those keeping score, this is the second post in a row in which I’ve written that Katzenberg was right about something.
Thanks! I think it's an interesting question - but as far as I understand, these tools still require a lot of human oversight and intention, it isn't just an algorithm on autopilot.
Interesting article, though distressing for those of us still foolishly invested in narrative art forms. Do consumers really want to tell themselves stories, or create stories by committee, vs being told a story by a skilled storyteller with a unique and surprising vision? Production barriers might be lowered, sure, but the craft of telling quality stories still takes years of hard work to master. A cultural landscape of endless fan service and fan fiction does not sound fun or tenable to me.