Twelve Labs Co., a San Francisco-based video search startup backed by Nvidia, is raising $100 million in Series B funding from investors including Amazon.com Inc., NEA Management Co. and Naver Ventures, strengthening its push to build artificial intelligence tools that can search, understand and organize large volumes of video.
The funding round was co-led by NEA and Naver, with participation from Radical Ventures, Index Ventures, Korea Investment Partners Co. and other investors. Alongside the investment, Amazon Web Services has signed a multiyear agreement to host Twelve Labs’ workloads on Trainium chips, with new models set to become available on AWS for developers building AI applications.
Twelve Labs is trying to solve a problem that has quietly grown with the explosion of video: most footage remains difficult to search beyond file names, folders, captions, transcripts or human memory. The company’s pitch is simple but ambitious. It wants video to become as easy to query as text.
Founded by Korean-born engineers and headquartered in San Francisco, Twelve Labs builds AI models designed to make video searchable and understandable. Its technology is aimed at companies sitting on large archives of footage, including entertainment studios, advertising firms, social media creators and sports organizations.
Jae Lee, the company’s co-founder and chief executive, said the company made an early bet that video would become central to how machines understand the world.
“Five years ago, we made a contrarian bet that video, and not text, is the most similar signal data that we receive as humans to learn about the world,” Lee said. He added that Twelve Labs’ work differs from newer frontier models that remain primarily language-based.
The company’s core products include Marengo 3.0, which converts raw footage so machines can search across sound, speech and motion, and Pegasus 1.5, which structures that output into a format that AI tools and applications can read. Together, the models are intended to improve how video archives are indexed, searched and used over time.
For a studio, sports franchise or media company, that could mean finding a precise moment inside thousands or millions of hours of footage without relying on manual tagging. The company points to examples such as locating a specific scene from On the Waterfront or Diego Maradona’s “Hand of God” goal from the 1986 World Cup.
Twelve Labs is also working on video agents that can search, explain, plan and execute tasks through text commands. Lee said the company has not yet named those agents.
The startup has about 200 employees, split evenly between Seoul and San Francisco. Its customers include Maple Leaf Sports & Entertainment Ltd., owner of the Toronto Raptors, AMC Global Media Inc. and UNICEF.
The funding comes at a moment when companies are looking beyond text-based AI and toward systems that can interpret richer formats such as video, audio and motion. For Twelve Labs, the opportunity lies in making every second of video easier to find, understand and use.
“Our goal is to make every second of video addressable, searchable and usable by agents,” Lee said.
Also Read: Alpha Review: Alia Bhatt’s Woman-Led Spy Film Steals the Show
