It hasn't been publicly announced, but Apple acquired Mountain View-based WaveOne, a startup that was working on algorithms for artificial intelligence to compress video.
The company declined to comment. WaveOne's website was, however pulled around January, and several of WaveOne's former employees, including one of its co-founders now work within Apple's various machine learning groups.
WaveOne's former head of sales and business development, Bob Stankosh, yesterday reported the sale on LinkedIn, posting it a month ago.
As Stankosh put it, "After almost two years at WaveOne, last week we finalized the sale of the company to Apple." He went on to write that he and his team started their journey at WaveOne realizing that machine learning and deep learning video technology would possibly change the world. Apple saw this potential and took the opportunity to add it to their technology portfolio.
In 2016, Lubomir Bourdev and Oren Rippel started WaveOne in a bid to transform decades-long video codec paradigms into AI. Bourdev was part of the founding team of Meta's AI research division prior to co-founding the venture. Before this, he and Rippel collaborated on Meta's computer vision team, which catered to content moderation, visual search, and feed ranking in Facebook.
Where standard compressing and decompressing video algorithms are concerned, the compression happens on the content provider's side, while the decompression happens on end-users' machines. It is an efficient approach but new codecs need new hardware specially built to accelerate compression or decompression; hence, improvements in that regard were slow to propagate.
Its most significant innovation was a "content-aware" video compression and decompression algorithm that could run on the AI accelerators integrated into millions of phones and growing numbers of PCs. Tapping AI-driven scene and object detection, the company's technology could accurately "understand" a video frame, which meant, for example, that it could favor faces over other features in a scene to economize on bandwidth.
WaveOne also said that its video compression technology was robust to sudden drops in connectivity. That is to say, it could make a best guess, based on whatever bits it had available, so when bandwidth suddenly dropped, the video wouldn't freeze: it'd just show less detail for the duration.
WaveOne said its hardware-agnostic approach could reduce video file sizes by as much as half in some cases, but the actual gains would really be more impressive, when the scenes are more complex.
Apparently, that is enough potential to attract $9 million from the backers like Khosla Ventures, Vela Partners, Incubate Fund, Omega Venture Partners, and Blue Ivy before Apple acquisition.
So what's it that Apple might want from an AI-powered video codec? An obvious motivation is better efficiency in streaming. Even the smallest improvements in video compression might save in bandwidth costs or enable services like Apple TV+ to stream at higher resolutions and higher framerates, depending on the content being streamed.
YouTube's already doing this. Last year Alphabet's DeepMind adapted a machine learning algorithm originally developed to play board games to the problem of compressing YouTube videos, leading to a 4% reduction in the amount of data the video-sharing service needs to stream to users.
Perhaps we'll soon see similar innovations from the WaveOne team, which is owned by Apple.