TwelveLabs Inc., the developer of generative artificial intelligence foundation models that can understand videos like humans, today announced it has raised $100 million in early funding to expand beyond simple understanding to achieve holistic intelligence.

The Series B round was co-led by NEA and NAVER Ventures. Amazon, Radical Ventures, Korea Investment Partners, Index Ventures, Quadrille Capital and Red Bull Ventures also participated in the round. Today’s capital infusion brings the company’s total raised to more than $207 million.

“Five years ago, we made a contrarian bet: The substrate of machine intelligence is recorded reality in motion, not language,” said co-founder and Chief Executive Jae Lee. “Language is downstream of understanding. Video is the data understanding has to answer to.”

TwelveLabs is bringing genuine AI power to the field with its technology, having built frontier foundation models capable of video understanding. The company worked from the ground up to construct multimodal models that weren’t simply large language models that process video, but models that natively understand video.

The company’s flagship products include the Marengo model family, with the 3.0 version released late last year and Pegasus 1.5.

Maregno enables real-world AI embedding for videos, audio, text and composition. That means that it can parse numerous types of content and add it to machine-readable data structures, such as vector databases, so that AI models can understand and search the information at scale.

Pegasus works alongside the previous model to turn video into structured data. It understands scene boundaries, entities, time segments and what’s happening, enabling LLMs to reason across visual information. It operates similarly to how large language models process large documents and images, summarizing them in markup languages to make them easier to understand.

Today’s LLMs can’t consume video all at once. They need to segment it up into flashes – a series of screenshots – and then use that to reason across. TwelveLabs said it built a reasoning capacity that natively understands trends over time by maintaining a memory that persists between queries, rather than evaporating after each one: an intelligence that compounds with each video, allowing the model to become more knowledgeable.

The company said that, using these models, it intends to build a new paradigm of video perception, creating a system that allows machines to analyze, search and operationalize footage.

Use cases for this type of technology span numerous workflows across industries such as security, advertising, sports and automotive, where tremendous amounts of information reside in video.

As part of the funding, TwelveLabs is deepening its relationship with Amazon Web Services. The company’s service has been available on AWS Marketplace since at least 2025, alongside managed access to foundation models via Amazon Bedrock.

AWS is already TwelveLabs’ preferred cloud provider, and with today’s investment, the company has signed a multiyear commitment to optimize its video inference workloads for AWS Trainium chips. In addition, new frontier models will launch on AWS first.

Photo: Pixabay