In the relentless environment of Silicon Valley, innovation often hits the hard wall of fiscal reality. Snap Inc., the parent company of Snapchat, has announced the spin-off of its AI video research team into a new, independent entity called Dotmo. This move is not merely an organizational reshuffle; it is a clear admission that the cost of developing cutting-edge Generative Video models has become prohibitively expensive, even for social media giants.
The Economic Hemorrhage of GPUs and the Hunt for Capital
Developing models like OpenAI’s Sora or Kuaishou’s Kling requires thousands of Nvidia H100 GPUs and millions of dollars in monthly cloud computing expenses. For Snap, a company that has struggled for years to convince Wall Street of its long-term profitability, keeping such a team on its balance sheet was becoming a liability. Dotmo, operating as a standalone company, will have the ability to raise external capital from venture capital firms without directly impacting Snap’s earnings reports.
According to analysts, the spin-off allows Snap to retain a significant equity stake in Dotmo, ensuring access to its technology, while transferring the risk of failure and the R&D burn rate to third-party investors. It is a "strategic offloading maneuver," as market insiders call it, signaling that the era of "growth at any cost" is definitively over.
Dotmo: A New Promise for Mobile Video
Dotmo is not starting from scratch. It inherits the intellectual property and the top-tier engineers from Snap who worked on "Dreams" and the AI lenses we know today. The new company’s goal is to build high-fidelity video generation tools optimized for mobile devices. While competitors focus on long-form cinematic shots, Dotmo appears to be targeting the "attention economy": short, impactful videos that can be generated in seconds by an average smartphone user.
Dotmo’s independence also allows it to collaborate with other platforms. It will no longer be exclusively tied to the Snapchat ecosystem, opening doors for partnerships with TikTok, Instagram, or even Hollywood production houses. This flexibility is critical for the survival of an AI startup in today's landscape.
The Broader Trend of AI Unbundling
The case of Snap and Dotmo is not an isolated incident. We are witnessing a broader trend where tech companies are unbundling their most expensive AI research divisions. Microsoft invested billions in OpenAI rather than developing everything in-house, while Apple is partnering with Google and OpenAI for Apple Intelligence. Snap’s strategy is a variation of this model: instead of buying an external solution, it creates its own and sets it free to find its own market path.
The remaining question is whether Dotmo can compete with the industry titans. With Snap’s backing and the agility of a startup, it has a solid starting point. However, the battle for GPUs and training data is fierce, and Dotmo will need to prove that its technology offers something unique that users and advertisers are willing to pay for.
Conclusions and Outlook
This move marks the end of the first, "romantic" phase of AI, where every company tried to build its own large models. We are now entering the phase of fiscal consolidation. For Snapchat users, this might mean fewer free AI features in the future, but for Snap as a corporation, it is a necessary step to ensure viability in an environment where capital is no longer cheap.
