TwelveLabs Raises $100 Million in Series B Funding to Build Video Superintelligence

Industry leader develops a first-of-its-kind system that unifies perception, knowledge, and reasoning to transform video into intelligence that grows more valuable with every interaction

TwelveLabs, a leading video intelligence company, today announced it has raised $100 million in Series B funding. The 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. The investment comes as TwelveLabs expands beyond video understanding models into a full-stack agentic intelligence system for video that combines perception, knowledge, and reasoning into a single architecture. For the first time, organizations and creators can put vast video archives to work as a living, searchable system, unlocking footage that was historically too hard to analyze, operationalize, or monetize.

TwelveLabs’ maturation and platform extension take place at a pivotal moment for the video intelligence market. Video represents the vast majority of the world’s data, upwards of 90%, yet most of it is opaque and inaccessible. Enterprises are rapidly moving from experimentation to production-scale deployment of video understanding technology; TwelveLabs has been at the center of this shift, establishing deep traction in media and entertainment while moving into the public sector as well, working with governments around the world to apply video intelligence to mission-critical workflows. Additional verticals, including advertising, security, sports, and automotive, continue to drive demand for the company’s platform.

Marketing Technology News: MarTech Interview with Miguel Lopes, CPO @ TrafficGuard

“NEA backed TwelveLabs in the early days of video intelligence — and that conviction has only deepened as they have shaped this exciting category,” said Tiffany Luck, Partner at NEA. “Jae and his team built the foundation models that set the standard for what video understanding can be. They’re purpose-built to turn millions of hours of footage into intelligence that compounds over time, and as video understanding moves from novel capability to essential infrastructure, we believe TwelveLabs is the company defining what comes next.”

The Foundation: Genuine Multimodality
TwelveLabs’ position is built on a foundation of category-defining research and a core principle: genuine multimodality. Not language models watching video, but models born in video. The company’s Marengo 3.0 model, released late last year, represents the world’s most powerful video embedding model. It understands every sound, every word, every motion on screen across time, turning raw video into a semantic layer that machines and AI can understand and search at scale.

Marketing Technology News: Is the Traditional CDP Already Out of Date?

Working alongside Marengo, the company’s recently released Pegasus 1.5 model turns video into structured data: scene boundaries, entities, temporal segments, and semantic context the system can reason with. Pegasus functions as a domain-specific language for video understanding, making raw footage parseable by any intelligent system built on top, the same way markup languages make raw documents parseable by browsers.

Together, these models form the perception that powers everything TwelveLabs builds, and everything its customers and partners build on top. Both models are distributed through Amazon Bedrock and through TwelveLabs’ own API.

“TwelveLabs was the first investment NAVER Ventures ever made, and co-leading their Series B is the strongest expression of conviction we can offer,” said YJ Park, General Partner at NAVER Ventures. “When we first met Jae, he described TwelveLabs as the visual cortex for future AI agents. That framing has only sharpened over time. As agents and machines move into roles where they need to perceive and reason about the physical world, video is the modality that matters most, and TwelveLabs is the team building that capability with the depth the problem demands. We are proud to co-lead this round.”

From Models to a Full-Stack Agentic Video Intelligence System
Most of the world’s data lives in video, yet the intelligence inside it remains out of reach. The tools of the LLM era were built for text. Applied to video, they sample a few frames, miss everything in between, and start from zero with every query. Brute-force fixes fail in both directions. Feeding entire video libraries into a model’s context window would require compute and technology that doesn’t exist, at a cost no enterprise could justify.

Converting video into a static database creates structure, but no intelligence capable of acting on it.

TwelveLabs’ new agentic architecture unites both halves. The system builds a structured, persistent memory of every video it ingests, then reasons across all of it. The more content it indexes and the more analysis it runs, the more capable it becomes. This is intelligence that compounds with every video processed, not a tool that resets with every query.

Delivering this requires owning the entire stack. Because TwelveLabs builds its own perception, knowledge, and reasoning layers, along with the orchestration that binds them from the moment a video enters the system, the result is cohesive intelligence rather than an approximation. That end-to-end ownership is what makes infrastructure reliable and durable.

With models proven in production by leading enterprises and the agentic infrastructure to unlock new user cases, TwelveLabs is now moving up the stack into applications that put the full system in the hands of creators, operators, and decision-makers, no integration required. The company took its first step earlier this month with the launch of Rodeo, its first application-layer product.

Partnering for the Future
TwelveLabs partners continue to play a crucial role in its growth. In particular, TwelveLabs’ relationship with Amazon goes beyond today’s investment. Amazon Web Services (AWS) is TwelveLabs’ preferred cloud provider, and the two companies have deepened their strategic partnership with a multiyear commitment that includes optimizing TwelveLabs’ video inference workloads on AWS Trainium chips. Additionally, new TwelveLabs models will launch first on AWS, reinforcing a shared vision for building the intelligence layer for video at scale.

“TwelveLabs has been pushing the boundaries of what AI can perceive and reason about since its earliest days, and we’ve had the privilege of partnering with them throughout that journey,” said Jason Bennett, VP and Global Head of Startups and Venture Capital at AWS. “Their models have been delivering real value to customers on Amazon Bedrock for more than a year, and as they scale their video cognition system on AWS infrastructure—including our purpose-built Trainium chips—we’re excited to deepen our partnership with a team that is defining video intelligence at production scale.”

To continue its rapid pace of innovation and adoption, TwelveLabs will use its Series B funding to invest even more heavily in research and development. It will continue its investments in San Francisco and Seoul and also support geographic growth, with new offices in New York and London to serve demand from global customers.

“Five years ago, we made a contrarian bet: the substrate of machine intelligence is recorded reality in motion, not language. Language is downstream of understanding. Video is the data understanding has to answer to,” said Jae Lee, CEO and co-founder of TwelveLabs. “We have spent half a decade building the perception, knowledge, and reasoning architecture to close that gap. Models commoditize. The intelligence layer that composes them does not. This funding lets us take TwelveLabs from foundation models to a full-stack video cognition system that meets every user, every agent, and every machine that needs to understand the world. The road to Video Superintelligence starts here.”