Versos AI & Tape Ark unlock taped video for training

Versos AI has partnered with Tape Ark to convert video archives stored on physical tape into data for AI training, aiming to bring offline footage into the AI market.

The partnership combines Tape Ark's tape digitisation and cloud migration services with Versos AI's software for tagging, indexing and structuring video files. That would allow archived footage to be turned into searchable, licensable datasets for AI systems.

Large volumes of video remain stored on ageing tape formats across broadcasters, studios, sports organisations and other enterprises. In many cases, those archives rely on older hardware and incomplete catalogues, making the material difficult to access or assess.

This creates a supply issue for AI companies seeking more training data, particularly as demand shifts beyond text and still images to video. It also leaves archive owners paying to store content without making practical use of it.

Versos AI said its software analyses digitised footage and tags objects, scenes, people and events across large video collections. Once processed, archives that had been sitting on shelves can be searched, reviewed and prepared for licensing.

Storage economics

Tape Ark presented the partnership as a way for organisations to cut the cost of retaining old media while opening a potential new income stream from existing collections. Many archive owners, it said, still pay monthly storage charges for tape libraries that are rarely viewed and may be deteriorating.

"Companies are paying between $0.40 and $1.20 per tape per month just to store content on a shelf. Many are spending millions of dollars holding on to content that's largely inaccessible, practically unusable and degrading every day," said Guy Holmes, Chief Executive Officer, Tape Ark.

"Once those tapes are digitized and moved to the cloud, it can be stored for a fraction of a cent - and more importantly, it becomes usable. Instead of paying to store it, organizations can start monetizing it as AI training data," Holmes said.

Tape Ark said it has processed more than an exabyte of data worldwide, including more than 200 petabytes of video. It said that volume is roughly equal to 29 million hours of high-quality 4K video.

Video demand

Versos AI linked the deal to a broader shift in the AI sector towards systems trained on more real-world video. It argued that a large amount of useful footage remains outside the datasets currently available to model developers because it has never been digitised or structured for machine use.

"The future of AI training data is video, and much of it is still trapped on tape," said Chris Keevill, Chief Executive Officer, Versos AI.

"This partnership with Tape Ark creates a direct pipeline to unlock that footage, and transform it into structured, usable data for AI systems, bringing decades of video that's trapped on tape into the AI economy for the first time," Keevill said.

The arrangement reflects a wider push across the AI industry to find new sources of training material as developers face tighter limits on what can be extracted from open web content. Private archives, historical media libraries and specialist collections are drawing increasing attention as alternatives.

For content owners, the offer is twofold: preserve vulnerable media by moving it off obsolete tape formats, and make those collections easier to catalogue and potentially license. For AI developers, the appeal is access to footage that has not been widely used in existing training datasets.

Tape remains a common storage and backup medium despite the spread of cloud systems. As a result, a significant quantity of legacy content remains outside modern search and analysis tools, especially where records were created manually or not maintained in detail.

By linking tape migration with automated video indexing, the companies are seeking to build a commercial route from archive shelves to AI datasets. Tape Ark said the process turns material that was once hard to find and expensive to maintain into digital inventory that can be examined and used.