Snap is spinning off an internal generative AI video team into a new company called Dotmo, a move that shows how expensive advanced AI video development has become even for large consumer technology companies.
Dotmo will focus on building AI models for interactive gaming experiences, rather than only generating passive videos. The company is being formed from a team that previously worked inside Snap, but it will operate as a separate business while remaining closely connected to the Snapchat parent company.
Snap said one reason for the move is the high cost of developing this kind of AI video technology internally. The decision reflects a larger shift in the AI market. Video models are becoming more capable, but they are also becoming more expensive to train, run, and commercialize. For companies trying to manage costs, spinning out specialized AI work can be a way to keep innovation alive without carrying the full financial burden on the balance sheet.
The creation of Dotmo also fits Snap’s broader pattern of separating experimental or capital-intensive projects into more focused entities. Earlier this year, the company spun out its AR glasses work into a standalone subsidiary called Specs Inc., as it prepared to bring its next-generation glasses to market.
Generative AI video is one of the most technically demanding areas in artificial intelligence.
Image generation is already compute-heavy. Video is harder because models must understand motion, continuity, physics, lighting, camera movement, character consistency, scene structure, and time. A video model cannot simply create one convincing frame. It has to create a sequence of frames that feel connected.
That requires more training data, more compute, more memory, and more inference power. It also creates higher user expectations. A short AI video may look impressive at first glance, but users quickly notice broken hands, flickering faces, strange object movement, inconsistent backgrounds, and physics errors.
For a company such as Snap, which has long experimented with cameras, AR, filters, lenses, and creator tools, AI video is strategically relevant. But it is not cheap. Building advanced models internally can require major infrastructure spending before the product becomes a meaningful business.
Dotmo gives Snap a way to keep the work moving while shifting the project into a structure that may be easier to fund, focus, and scale.
The most interesting part of Dotmo’s plan is its focus on interactive gaming.
Many AI video tools are built around passive generation. A user enters a prompt and receives a short clip. That is useful for creators, advertisers, social posts, and concept art, but it still resembles traditional video production. The user asks, waits, and watches.
Interactive gaming requires something different. The model must respond to player actions, create or modify environments, maintain consistency, and support real-time or near-real-time experiences. The result is closer to a dynamic world model than a simple video generator.
This could make Dotmo more ambitious than a normal AI video startup. If it succeeds, the company could help create games where scenes, characters, settings, or interactions are generated on demand. That would be valuable for game studios, social gaming platforms, virtual worlds, AR experiences, and new forms of interactive storytelling.
It also fits Snap’s history. Snapchat has always been more interactive than traditional media. Lenses, filters, Bitmoji, AR effects, games, and camera-first communication all point toward playful, real-time experiences. Dotmo appears to extend that logic into AI-generated worlds.
The spinout gives Snap several advantages.
First, it reduces internal cost pressure. AI video development can consume enormous resources, and Snap is not in the same financial position as the largest AI infrastructure companies. Alphabet, Microsoft, Amazon, Meta, and OpenAI can spend tens of billions of dollars on data centers, chips, and model training. Snap has to be more selective.
Second, a separate company can raise outside capital. Investors interested in AI video, gaming, and world models may be more willing to fund a focused startup than invest indirectly through Snap’s broader public company story.
Third, Dotmo can recruit and operate with a startup identity. Specialized AI researchers and engineers may prefer the pace, equity structure, and technical focus of a standalone company.
Fourth, Snap can retain strategic access. By staying close to Dotmo, Snap may benefit from the technology if it becomes useful for Snapchat, AR glasses, creator tools, or gaming features.
This is a familiar strategy in tech. A company can spin out a risky but promising project while maintaining a relationship that gives it future upside.
The Dotmo spinout comes during a busy period for Snap.
The company is still trying to strengthen its advertising business, grow user engagement, compete with TikTok and Instagram, and prove that its AR strategy can become a meaningful platform. It has also been investing in AI tools for creators, messaging, lenses, and content experiences.
At the same time, Snap has had to watch costs carefully. The company has gone through restructuring, layoffs, and efforts to improve operational discipline. AI adds another financial challenge because the most advanced models require expensive compute at a time when investors want public companies to show efficiency.
That makes Dotmo’s structure useful. Snap does not have to abandon advanced AI video, but it also does not have to keep every experimental model-building effort fully inside the company.
This approach may become more common. As AI projects become more expensive, companies may split them into separate entities, joint ventures, or outside-funded labs rather than funding everything internally.
Dotmo enters a competitive market.
OpenAI has pushed forward with Sora. Google has advanced Veo. Runway, Pika, Luma, Kling, and other companies are fighting for creators, studios, advertisers, and developers. Meta, Adobe, ByteDance, and several Chinese AI companies are also building video generation systems.
That makes differentiation important. A new AI video company cannot rely only on generating short clips from text prompts. That market is already crowded and improving quickly.
Dotmo’s focus on interactive gaming may help it stand apart. Gaming creates different technical requirements and different business opportunities. If a model can support playable, responsive, generated environments, it may become useful to studios and platforms looking for new kinds of content.
But that is also a harder problem. Games require consistency, control, low latency, user agency, and reliability. A visually impressive video demo is not enough. Players expect interaction to feel coherent.
Dotmo will need to prove that its models can move beyond demo clips and into usable experiences.
The spinout also reveals a deeper issue in AI video: the technology is exciting, but the economics are difficult.
Training models is expensive. Serving generated video to users is expensive. Building safety systems is expensive. Licensing or filtering training data can be expensive. And many users are not yet willing to pay enough to cover those costs at scale.
Consumer AI video apps can attract attention quickly, but turning that attention into profitable revenue is harder. If every prompt costs real compute money, companies need strong pricing, enterprise customers, advertising value, or high-volume paid use cases.
Gaming may offer a stronger commercial path than casual video generation. Game studios and interactive platforms may pay for tools that reduce production costs, create dynamic content, or support new gameplay formats.
That may be one reason Dotmo is not being framed as another general AI video app. It is targeting a category where interactivity could create clearer value.
Dotmo may also support Snap’s long-term AR ambitions.
Snap has spent years working on augmented reality and smart glasses. Its Specs project is now separated into its own company, but still tied closely to Snap’s broader ecosystem. If AR glasses become more useful, they will need interactive content, spatial experiences, responsive characters, and lightweight creation tools.
AI-generated interactive environments could become important for that future.
Imagine AR lenses or games that change based on the user’s surroundings, generate scenes dynamically, or create interactive characters in real time. Those experiences would require more than simple filters. They would need AI systems that understand context and respond visually.
Dotmo’s focus on interactive gaming could therefore serve more than one market. It could help Snap build future AR experiences while also selling technology to outside developers or studios.
That connection may be part of the strategic reason Snap wants to keep the new company close.
For Snap investors, the spinout may also signal cost discipline.
AI has become a pressure point for public companies. Investors want companies to show they are using AI to improve products, but they also worry about uncontrolled spending. The largest tech firms can justify massive AI budgets because they have cloud businesses, dominant platforms, and large profit pools. Smaller public tech companies have less room for open-ended research spending.
By spinning out Dotmo, Snap can show that it is not ignoring AI video, but also not absorbing all of its development cost internally. That may be a more realistic path for a company of Snap’s size.
It also reflects a broader lesson of the AI boom. Not every company can behave like OpenAI, Google, or Meta. Some will need partnerships, spinouts, licensing deals, or specialized structures to stay in the race.
Dotmo’s launch points toward where AI video may go next.
The first wave of AI video focused on generating clips. The next wave may focus on controllable and interactive experiences. Instead of asking a model to produce a finished video, users may ask it to create a world, character, scene, game level, or story that changes in response to them.
That would blur the line between video, gaming, simulation, AR, and social media.
Snap is a logical place for that idea to emerge because the company has always treated the camera as an interactive surface. Its products are built around play, filters, expression, messaging, and lightweight creation. Dotmo takes that creative DNA and applies it to more advanced AI video models.
The question is whether the technology can become practical and affordable.
AI video is moving quickly, but the business model is still unsettled. Dotmo gives Snap a way to explore the next layer of the market without carrying the full burden alone.
For the broader industry, the message is clear. AI video is no longer only about making clips look realistic. It is moving toward systems that can create experiences people can actually interact with.
Share your thoughts about this article.
Be the first to post a comment!
