- TwelveLabs raised $100 million in a Series B co-led by NEA and NAVER Ventures.
- It’s shifting from selling AI models to a full platform that reasons across video archives.
- Amazon, a participating investor, named AWS its preferred cloud provider, with new models launching there first.
TwelveLabs wants to build what its chief executive officer, Jae Lee, calls “video superintelligence.” That phrase invites comparison to OpenAI’s Sora and Google’s Veo, two of the most talked-about names in AI video. But TwelveLabs is deliberately not fighting that battle.
The San Francisco- and Seoul-based startup raised $100 million in Series B funding to double down on a narrower, less flashy problem: making sense of existing video rather than generating new video from scratch.
The company has grown to roughly 178 employees as of June 2026, up from around 58 a year earlier.
Understanding video, not generating it
TwelveLabs was founded in 2021 by Lee, Dave Chung, Aiden Lee, Soyoung Lee, and SJ Kim, who met while serving in South Korea’s military cyber operations command. Lee previously worked as a lead data scientist at South Korea’s Ministry of National Defence before recruiting the founding team.
The startup, founded to solve a problem most large language models still can’t handle well, works by building models specifically for video rather than adapting text-based AI to it. Its Marengo model converts raw footage into searchable data, capturing speech, sound, motion, and the relationships among events over time.
A second model, Pegasus, reasons over that data continuously, tracking entities and causation across up to two hours of footage rather than sampling isolated frames. For example, a broadcaster sitting on years of unlabeled game tape could ask the system to pull every match-winning goal scored in the final five minutes, without a person ever having tagged a single clip.
That’s a meaningfully different bet than the one Sora, Veo, or Runway are making. Those tools compete on how realistic and controllable the AI-generated video looks. Runway, now valued at $5.3 billion after a $315 million round in February 2026, has pushed further still into world models that simulate 3D physical environments.
TwelveLabs’ argument is that the harder, more durable problem is extracting intelligence from existing video, most of which sits unindexed in surveillance systems, sports archives, and enterprise libraries.
“We have spent half a decade building the perception, knowledge, and reasoning architecture to close that gap. Models commoditise. The intelligence layer that composes them does not,” Lee says.
Why the AWS deal matters as much as the money
The most consequential detail in this round isn’t the $100 million — it’s whose writing part of that check. Amazon, already a repeat investor in TwelveLabs, used this round to formalise AWS as the company’s preferred cloud provider, with new models optimised for AWS Trainium chips and launched on them first.
It’s a pattern becoming familiar across AI: cloud providers using investment dollars to secure a customer’s compute spend and product roadmap, not just equity upside. Tech Funding News saw a nearly identical structure play out weeks earlier, when Amazon named itself the preferred cloud provider for Odyssey, another AI lab, just months after Nvidia’s venture arm had backed that company’s prior round.
The Series B was co-led by NEA and NAVER Ventures, the latter’s first investment since launching its Silicon Valley venture arm in mid-2025. Returning investors Index Ventures, Radical Ventures, Korea Investment Partners, and Quadrille Capital joined, alongside new backer Red Bull Ventures. The round follows a $50 million Series A in 2024 co-led by NEA and NVIDIA’s NVentures, bringing TwelveLabs’ total funding to roughly $150 million.
NEA has previously backed infrastructure-heavy companies, including Databricks, giving it a track record of conviction in unglamorous but foundational technology bets, a category TwelveLabs is now explicitly trying to join.
“TwelveLabs was the first investment NAVER Ventures ever made, and co-leading their Series B is the strongest expression of conviction we can offer. When we first met Jae, he described TwelveLabs as the visual cortex for future AI agents. That framing has only sharpened over time,” says YJ Park, general partner at NAVER Ventures.
The video-understanding category that TwelveLabs occupies is small compared to the broader AI video generation market, but it’s growing quickly as enterprises move from pilots to production: industry estimates put AI video search at roughly $3.2 billion by 2028.
TwelveLabs plans to use the new funding for research, product development, and new offices in New York and London, alongside its existing San Francisco and Seoul bases. Whether video superintelligence becomes a real category, or another AI framing that quietly resolves into which cloud provider owns the compute, is the question the next few funding rounds in this space will answer.
