en.Wedoany.com Reported - Video understanding AI company TwelveLabs Inc. announced it has secured $100 million in early-stage funding, bringing its total funding to over $207 million. The Series B round was co-led by NEA and Naver Ventures, with participation from Amazon, Radical Ventures, Korea Investment Partners, Index Ventures, Quadrille Capital, and Red Bull Ventures.
Company founder and CEO Jae Lee stated that the foundation of machine intelligence lies in dynamically recording reality, and video is the data that understanding must respond to. TwelveLabs is dedicated to building multimodal models that natively understand video, rather than merely treating it as a processing tool for large language models.
Its flagship products include the Marengo model series and Pegasus 1.5. Marengo supports real-world AI embeddings for video, audio, text, and combinations thereof, converting multiple content types into machine-readable data structures that facilitate large-scale AI understanding and information search; version 3.0 of this series was released late last year. Pegasus, on the other hand, transforms video into structured data, understanding scene boundaries, entities, and temporal segments, enabling large language models to reason across visual information. It operates similarly to how large language models process documents and images, summarizing videos in a markup language to make them more comprehensible. The model maintains a memory mechanism that persists across queries, accumulating knowledge as video content compounds.
The company plans to build a new video perception paradigm based on these models, enabling analysis, search, and manipulation of video material. This technology can be applied to industries such as security, advertising, sports, and automotive, where vast amounts of video information exist.
As part of the funding, TwelveLabs has deepened its partnership with Amazon Web Services. Its services have been available on AWS Marketplace since 2025, with foundational model hosting access provided through Amazon Bedrock. TwelveLabs has signed a multi-year commitment to optimize video inference workloads based on AWS Trainium chips, with future frontier models set to debut on the AWS platform.
