Apple Acquires AI-Powered Video Compression Startup, WaveOne

Apple has quietly acquired WaveOne, a Mountain View-based business working on artificial intelligence (AI) algorithms for compressing video.

However, WaveOne's website was taken down in January, and several former staff members, including one of WaveOne's co-founders, are now employed by Apple's various machine learning divisions.

Bob Stankosh, the organization's former head of sales and business development, made the sale public in a LinkedIn post that was published a month ago.

“After almost two years at WaveOne, last week we finalised the sale of the company to Apple,” Stankosh wrote.

“We started our journey at WaveOne, realising that machine learning/deep learning video technology could potentially change the world.

"Apple saw this potential and took the opportunity to add it to their technology portfolio.”

Lubomir Bourdev and Oren Rippel established WaveOne in 2016 with the goal of transforming the antiquated paradigm of video codecs into an AI-powered system.

Bourdev was a founding member of Meta's AI research division before joining the company, and he and Rippel both worked on the Meta computer vision team, which was in charge of Facebook's content moderation, visual search, and feed ranking.

When it comes to common video compression and decompression algorithms, the compression takes place on the servers of the content provider (like YouTube), while the decompression is handled by the computers of the end users.

It's a good strategy, but because new codecs need technology designed to speed up compression or decompression, algorithmic advancements are slow to catch on.

The main advancement made by WaveOne was a "content-aware" video compression and decompression algorithm that could be utilised by PCs, smartphones, and an increasing number of tablets with AI accelerators.

The startup's technology could essentially "understand" a video frame by utilising AI-powered scene and object detection, which allowed it to, for instance, prioritise faces over other scene features in order to conserve data.

Moreover, WaveOne asserted that its video compression technology was resistant to unanticipated connectivity outages.

The movie wouldn't freeze when bandwidth was suddenly limited since it could make a "best guess" based on the bits it had available. Instead, it would display less information for the duration.

WaveOne asserted that its method, which was independent of technology, could cut the size of video files in half, with larger improvements in more complicated scenarios.

Evidently, investors recognised the potential. WaveOne received $9 million in funding from investors including Khosla Ventures, Vela Partners, Incubate Fund, Omega Venture Partners, and Blue Ivy prior to the Apple acquisition.

What then could Apple hope to achieve with an AI-powered video codec?

Well, more effective streaming is the obvious solution.

Depending on the sort of streaming content being used, even slight advancements in video compression could reduce bandwidth costs or allow services like Apple TV+ to deliver greater resolutions and framerates.

YouTube already carries out this. YouTube's data requirements dropped by 4% as a result of Alphabet's DeepMind's adaptation of a machine learning algorithm originally created to play board games to the challenge of compressing YouTube videos.

Perhaps the WaveOne team, owned by Apple, will soon produce innovations similar to these.