Video Rebirth, a Singapore-headquartered artificial intelligence video startup, has raised $50 million to develop software aimed at professional content producers, shifting away from consumer-grade generative tools that have dominated the market.
The company did not disclose its valuation or identify investors, describing them only as top-tier financial and strategic backers.
Proceeds will be used to build a high-fidelity, controllable platform for advertising, e-commerce, film, and animation, with a Version 1.0 release targeted for December 2025.
Founded by Wei Liu — a former Tencent distinguished scientist and fellow of the IEEE and AAAS — Video Rebirth is pursuing a “world model” approach to video generation that seeks to encode physical dynamics and scene logic.
The company argues that while generative video tools have seen rapid adoption, many struggle to meet studio requirements, citing weak motion consistency, limited shot control and unrealistic physics that increase production costs and limit reuse.
Video Rebirth’s technical pitch centres on a proprietary “Physics Native Attention” architecture combined with a curated data strategy.
Its “Bach” series of models is intended to improve realism in lighting, shadowing and object interactions, and to give creators finer control over movement and continuity between shots.
If successful, the company said, the approach could reduce the need for manual post-production and compositing, a key constraint for commercial deployments.
Generative video is a crowded field, with well-funded rivals racing to improve quality and guardrails amid concerns over intellectual property, deepfakes and model training data.
Delivery timelines in the sector have also slipped before, and Video Rebirth has not provided revenue metrics or detailed commercialization plans.
The company said the fresh capital will support research hiring and partnerships to build an ecosystem around what it calls AI-generated entertainment.
The firm said it aims to prove cinematic-grade output and physical consistency as it moves from research to product over the next year.
