Big Tech Bundles Video Generation
Mirage
The real risk is that video generation is quickly becoming a feature, not a standalone category. Mirage is selling a social video workflow, but larger companies can treat generation as a loss leader, absorb the compute bill, and drop similar capabilities into products users already open every day, like ChatGPT, Gemini, and Canva. That makes distribution and workflow ownership matter more than raw model quality alone.
-
OpenAI and Google are bundling text to video into broad consumer AI products, while Canva layers native AI generation and a plugin ecosystem into its design suite. In practice, that means many users can make clips without ever choosing a dedicated video tool first.
-
Incumbents also have more room to price aggressively. OpenAI is operating at roughly $25B in annualized revenue as of February 2026, and Canva reached $4B ARR at the end of 2025. Mirage has raised $100M, meaningful for a startup but small next to platforms funding AI from much larger revenue bases.
-
The market is already splitting into layers. Foundation labs own models, broad platforms bundle video into larger suites, and specialists like Mirage, OpusClip, and Velvet focus on narrow jobs such as social clips or launch videos. That usually compresses margins for the middle layer unless it owns a sticky workflow.
Going forward, the winners in AI video will be the companies that become the default place where work starts and finishes. For Mirage, that means pushing beyond generation into publishing, analytics, iteration, and social optimization, so customers stay for the workflow even as the underlying video model becomes easier to buy anywhere.