- The chronology: from internal team to autonomous company
- Who wins and who loses in this scenario
- Video AI is so expensive to develop because it requires significant investment in several key areas: * **Computational Power:** Training AI models, especially those for video, demands massive amounts of processing power. This requires specialized hardware like GPUs (Graphics Processing Units) and TPUs (Tensor Processing Units), which are costly to purchase and maintain. The sheer volume of data processed for video analytics also contributes to high computational costs. * **Data Collection and Annotation:** High-quality video datasets are essential for training effective AI models. Collecting and curating these datasets is time-consuming and expensive. Furthermore, accurately annotating video frames (e.g., identifying objects, actions, or events) requires human effort or sophisticated automated tools, which adds another layer of cost. * **Expertise and Talent:** Developing advanced AI requires highly specialized skills. Machine learning engineers, AI researchers, data scientists, and computer vision experts are in high demand and command significant salaries. Building a skilled team is a major expense. * **Research and Development:** The field of AI is constantly evolving. Significant R&D is needed to push the boundaries of what's possible, experiment with new algorithms, and optimize existing ones. This ongoing innovation requires substantial investment in research grants, experimentation, and continuous learning. * **Algorithm Complexity:** Video AI often involves complex algorithms that need to understand motion, temporal relationships, context, and multiple interacting elements within a video stream. Developing and refining these algorithms is a technically challenging and resource-intensive process. * **Large Model Sizes:** State-of-the-art AI models, particularly generative models for video creation or advanced analysis, can be extremely large, requiring huge amounts of memory and storage. * **Infrastructure and Tools:** Setting up and maintaining the necessary cloud infrastructure, development environments, and specialized software tools for AI development incurs significant costs. * **Testing and Validation:** Rigorous testing and validation are crucial to ensure the reliability and accuracy of video AI systems. This involves extensive testing across various scenarios and edge cases, which can be a lengthy and costly process. * **Iterative Improvement:** AI models are rarely perfect on the first try. They require continuous iteration, retraining, and fine-tuning based on performance feedback and new data, further contributing to ongoing development costs.
- SHM Studio reading: real opportunity or premature hype?
- Operational implications for content and campaign managers
- The construction site still open: what to observe in the coming months
Snap has announced the spin-off of its internal video AI team. The new entity is called Dotmo. It is comprised of Snap employees who are leaving the parent company to focus exclusively on developing AI-based video technologies.
Therefore, the move is not just an internal reorganization. It is a clear signal: the costs of video AI are too burdensome for a social-first company like Snap. Consequently, separating the team allows for seeking dedicated funding and building a vertical product. However, for Italian marketing managers, the relevant point is different. Dotmo could become an independent player in the market for content creation automation, with direct implications for video campaigns, synthetic UGC, and scalable creative production.
We of SHM Studio We are closely monitoring these developments. AI video is maturing rapidly. Furthermore, the tools that stem from it are increasingly entering pipelines. digital marketing Italian SMEs. In summary, Dotmo is still a work in progress — but it's worth keeping on your strategic radar.
The chronology: from internal team to autonomous company
On June 18, 2026, TechCrunch reported the official news. Snap has decided to separate its video AI research and development group. The new company is called Dotmo. It is comprised of current Snap employees who are leaving the social company's structure to found an independent entity.
The stated motivation is clear: costs. The development of AI video models requires enormous infrastructure. Furthermore, it requires specialized teams that can hardly find their natural habitat within a company focused on Snapchat and social advertising. Therefore, the spinoff is a rational choice, not an escape.
This isn't the first time Snap has gone down this path. In the past, the company has already spun off internal units to reduce operational overhead. However, this operation has a different significance: it concerns generative AI applied to video, one of the most competitive and expensive segments in the entire tech landscape.
Who wins and who loses in this scenario
From the perspective of market equilibria, Dotmo's emergence has multi-level effects. Let's analyze them in order.
Snap It lightens the balance sheet. As a result, it can focus resources on the core advertising business. However, it gives up a technological asset that, if mature, could have differentiated it from competitors. It is a defensive choice, not an offensive one.
Dotmo earns operational freedom. In fact, as an independent entity, it can raise capital from AI-specialized VCs, forge partnerships with video platforms, and develop products without the constraints of a social roadmap. Conversely, it loses Snap's distribution network and data. The path will be anything but simple.
The martech market receives an important signal. More and more players are building vertical stacks on AI video. Among other things, the competition with Runway, Pika Labs, and the video models from Google and OpenAI is already fierce. Dotmo will need to find a precise positioning to survive.
Why is AI video so expensive to develop?
It's worth understanding the technical context. Generative models for video require significantly higher computational resources than models for text or images. McKinsey estimates that the training and inference costs for video models are still an order of magnitude higher than for language models.
Moreover, the quality expected by users and brands is very high. It's not enough to generate coherent videos; outputs ready for distribution on advertising platforms are needed, with precise resolution, timing, and brand safety standards. Therefore, the gap between research prototype and commercial product is still very large.
For this reason, many tech companies are choosing to outsource or spin off these units. Similar to what happened with some NLP research divisions in previous years, AI video is becoming a specialized field in its own right.
SHM Studio reading: real opportunity or premature hype?
We of SHM Studio We work daily with marketing managers who ask us when and how to integrate AI video into their creative pipelines. The honest answer is: it depends on the use case.
Today, available AI video tools are useful for specific scenarios, such as the production of creative variants for Google Ads campaigns o A/B testing on short video formats. However, for productions requiring brand consistency, recognizable actors, or complex narratives, the limitations remain evident.
If Dotmo manages to build a solid commercial product, it could fill some of these gaps. Specifically, it could do so for the SMB segment that doesn't have budgets for traditional video production. Therefore, there is strategic interest. But the product's maturation time is still uncertain.
For marketing managers considering investments in AI solutions, The advice is to monitor Dot's evolution without expecting ready-to-use outputs in the short term. Likewise, it is worth exploring existing tools today to identify the most suitable use cases for your sector.
Operational implications for content and campaign managers
Beyond corporate dynamics, the news has practical implications for those involved in digital marketing e content strategy. Let's look at the most relevant points.
- Content automation video The separation of specialized AI video teams is accelerating industry specialization. Consequently, in the next 12-18 months, we are likely to see new vertical tools that are more precise and less general.
- Synthetic UGC: One of the most interesting use cases for brands is the generation of video content that simulates the authenticity of user-generated content. However, platforms are refining their detection systems. Therefore, output quality will be a deciding factor.
- Integration with existing stacks: who manages campaigns on LinkedIn Meta knows how expensive creative production is. Well-integrated AI video tools could significantly reduce the cost per creative variant.
- SEO and video content: Google continues to index and value video content. Therefore, more scalable video production also has a direct impact on strategies SEO.
In summary, AI video is not yet plug-and-play for most Italian businesses. However, the market is rapidly structuring itself. Therefore, ignoring it today means being behind tomorrow.
The construction site still open: what to observe in the coming months
Dotmo has just been born. Therefore, it is premature to evaluate its final product or positioning. However, there are some signs to monitor closely.
First of all, the funding round. If Dotmo can attract significant capital from VCs specializing in AI, it will be an indicator of technical credibility. Following that, the choice of partnerships: integrating with existing video distribution platforms or building its own distribution are two paths with very different implications.
Finally, the type of clients Dotmo will serve. If it targets enterprise and large brands, the comparison will be with Runway and Adobe's tools. Conversely, if it targets SMEs and creators, it could carve out a more accessible and less crowded niche. Harvard Business Review has already analyzed how generative AI is redefining business creative models - and video is the next frontier.
To delve deeper into how to integrate AI tools and content automation into your strategy, you can explore SHM Studio services to read the in-depth articles about our blog. Additionally, the team is available for dedicated consultation through the page contacts.
Besides this, who is working on strategies for copywriting and content will find it useful to understand how video AI integrates with textual production—two increasingly complementary assets in multichannel campaigns. Similarly, those who manage the company's web presence through web solutions will have to consider how generative video content will impact user experience and time spent on the site.
Related articles
Discover other articles that explore similar topics in depth, selected to give you a more complete and stimulating view. Each piece of content is carefully chosen to enrich your experience.
