TwelveLabs Video Understanding Comes to Snowflake AI Data Cloud

We believe the best way to make video intelligence useful is to bring it to the platforms our customers already run on. That principle is why our models are in Amazon Bedrock, and it is the same thinking behind the exciting update we are sharing today: Marengo, our video understanding model, now generates embeddings natively inside Snowflake AI Data Cloud.

Snowflake has been a strong partner of ours, and this deepens that relationship in a meaningful way. Many of the largest names in media and entertainment build on Snowflake, including Warner Bros. Discovery, Disney, and Paramount. Bringing video understanding into that environment means it shows up where the work is already happening.

Why Snowflake Matters

Snowflake AI Data Cloud brings a governed AI control plane to their data in the customer’s account. This changes AI enrichment from a separate data pipeline to part of the analytics workflow itself.

The more important shift is what it does for unstructured data. Most enterprise analytics has been built on neat rows and columns. The harder, more valuable signal has been locked inside documents, audio, and video that the warehouse could not read. Processing video files directly in Snowflake creates a richer context for enterprises to measure and act on.

Video is the Data Nobody Could Read

Video is now the majority of what moves across the internet. By some estimates it is more than 80% of all consumer traffic, and the share inside enterprises is climbing just as fast across marketing, training, security, and media libraries.

Like text, video is unstructured. Unlike text, it has been almost impossible to work with at scale. And it is extraordinarily information dense: a single clip carries what is on screen, what is said, what is heard, how it is paced, and how it makes someone feel, all at once across time. That density is exactly what makes it valuable, and what has kept it sitting in storage as a cost rather than an asset.

TwelveLabs Marengo reads video the way a person does, taking in visuals, audio, and speech together, and encodes it into embeddings that capture meaning rather than just labels. Inside Snowflake, those embeddings are generated with a single function call over a video file, and the resulting vectors land in a table ready to search and join. When the video and the numbers that describe it live in the same place, questions that used to require a separate tool can become a query.

New Dimensions for Media Performance

Campaign results, audience segments, and spend already live in Snowflake. With Marengo embeddings beside them, advertising and media teams can start to connect what is actually in the content to how it performed.

That means asking which on-screen elements, pacing, or audio patterns track with engagement, scoring inventory for brand suitability at the scene level instead of relying on coarse category tags, and curating large content and creator libraries by what a clip contains rather than by whatever metadata someone remembered to enter. While these are familiar goals, what is new is being able to pursue them as analytics, on dimensions drawn from the video itself.

Additionally, some of the most valuable work in media happens between partners who cannot share raw data with each other, which is what Snowflake's Data Clean Rooms are built for. Since Marengo embeddings are generated inside a customer's own Snowflake environment and land in standard tables, enterprises can bring content-derived signals into those clean room workflows alongside the audience and outcome data, without sending footage to an outside service.

More to Come at Cannes 2026

This is an early step, and the best applications will come from teams who know their own content and their own data far better than we do. That is the part we are most excited about.

We will be at Cannes Lions 2026 to dig into the problems worth solving first. If your video and your data already sit in Snowflake, we look forward to connecting with you! To request access to TwelveLabs Marengo in Snowflake AI Data Cloud, please share your interest here.