How National Geographic will move 15 petabytes of video to the cloud

The National Geographic Society has a lot of videos—like, oceans of them. The 138-year-old nonprofit org, most famous for supporting expeditions to the furthest flung corners of the world (and the stacks of magazines in your grandfather’s attic), annually produces the equivalent of about 100 days of 4K video alone, according to the group’s Chief Technology Officer Jason Southern.

Now Southern wants to lead an effort to send National Geographic’s cumulative 15 petabytes of video—or the data equivalent of over 3 billion high-res images from your phone—to the cloud, where the National Geographic’s editors and producers can quickly find clips and other assets they need to build future stories. With the migration, Southern aims to have the data not only backed up, but also positioned to fuel future AI efforts.

The end result, if everything goes right, could enable production teams to use a plain-text request rather than a manual search to find, say, that one video of cone-bearing conifers in Northern California that was previously buried on a random drive.

“You know, I’m looking for redwood trees in Northern California,” Southern told us. “We would be able to ask questions like that of a conversational interface in the media production platform, and then be able to pull up proxies of those almost immediately.”

But before the LLM can start sending over pictures of the right tree, Nat Geo needs a strategy of prioritization (what data goes first), unification (getting it all on one platform), categorization (adding metadata), and testing.

Video store. National Geographic’s video data—including deep-sea footage, archaeological records, drone footage, and much more—is currently all on‑premises at its Washington, DC, facility, split between network-attached storage and thousands of standalone drives.

Southern said the effort—a partnership with Amazon Web Services (AWS)—will focus on video for now, not photography.

“Fifteen petabytes is obviously a huge amount of data. But the irony of that is that it’s only the tip of the iceberg of what’s really available, and it really actually does represent an incremental opportunity,” Lauren Stovall, global head of nonprofit programs at AWS, told us.

To take an additional petabyte out of this massive job, he said he’ll work with his production teams to learn which data will be used most frequently this year; those will get processed (and tested) first.

Step 1: Consolidate. First, Southern and his team will create a unified asset inventory. Physical drives will be shipped, and their media categorized and then placed onto Amazon’s S3 storage system, with each object tied back to the drive, the project, and the time period.

Step 2: Labels. Metadata provides descriptive labels and context to help an LLM understand what’s in the footage: Is it day? Is it night? Is the video showing wildlife? Asking someone to make all those descriptions manually would be a tough ask.

“There’s no way that we could spend the time, with human beings, to go through and review all of that footage and manually put metadata tagging on all over that,” Southern told IT Brew.

Southern said he’s also working with AWS-offered AI services (as well as third-party services that AWS has used with other customers) to analyze the video for additional metadata. That’s the biggest unknown, according to Southern, and his production teams will likely review the metadata outputs in the most important video assets for inaccuracies and missing info. Depending on how this goes, they may need to manually tag the most important files, or rerun them through metadata tools.

“There’s a lot that we need to learn. We know…this is going to be really experimental for a while for us,” Southern said, predicting that full data migration will be done by May 2027. And if it doesn’t work out—hey, at least the data’s now backed up in the cloud, allowing future producers to more easily produce stories that reveal the splendor of the natural world.

“The power of storytelling in the National Geographic Society is: People won’t protect what they don’t care about. And the way to get people to care about things is to tell them in stories,” Southern added.