TwelveLabs Raises $100M to Build Full-Stack Video Cognition System

TwelveLabs, a video intelligence startup, has closed a Series B round of $100 million (KRW 155.4 billion), pushing its total funding above $200 million (KRW 310.8 billion).

The round was co-led by Naver Ventures and NEA, with Amazon joining as a major backer. Existing investors Radical Ventures, Korea Investment Partners, and Index Ventures returned for the round, while Quadrille Capital and Red Bull Ventures came on board as new investors.

TwelveLabs plans to use the fresh capital to build out its “Video Cognition System,” an architecture that unifies video perception, memory, and reasoning. The approach is designed to move past a core limitation of today’s large language models, which typically sample only a handful of frames from a video and must re-analyze the footage from scratch for every new query. Instead, TwelveLabs’ system understands a video once, stores the results as structured memory, and draws on that memory for subsequent reasoning tasks.

The round also deepens TwelveLabs’ partnership with AWS. The company has named AWS its top-priority cloud provider, is optimizing video inference for AWS’s Trainium AI chips, and will debut its upcoming video foundation models on AWS first.

TwelveLabs is expanding its global footprint alongside the raise. It will reinforce its San Francisco and Seoul offices while opening new locations in New York, London, and Los Angeles to broaden its international operations. Last month, the company also launched a closed beta of Rodeo, an AI-powered video creation tool.

“We’re moving beyond being a foundation model company to build a full-stack video cognition system for developers, everyday users, enterprises, and agents alike, ushering in the era of Video Superintelligence,” said Jae Lee, CEO of TwelveLabs.