Apple hasn't officially confirmed the purchase yet, but WaveOne's website was shut down in January, and several previous employees, including one of its co-founders, are now working for the Cupertino company's machine learning teams.
Bob Stankosh, WaveOne's former head of sales and business development, wrote a blog post about the sale of the business a month ago.
Lubomir Burdev and Oren Ripel founded WaveOne in 2016. Prior to that, Burdev was one of the founding members of Meta's AI Research and Development division, where he coordinated content management, visual search, and Facebook feed ranking.
Video compression and decompression are performed on the consumer's devices in standard algorithms. However, new codecs require new hardware specifically designed to speed up compression or decompression, making the improvement slow to spread.
WaveOne's major innovation was a video encoding and decoding algorithm that could be plugged into smartphones and PCs, which enabled the startup's technology to effectively "understand" a video frame, prioritize faces by considering other elements in the scene, and save bandwidth.
WaveOne claimed that its video compression technology was unavoidable when the connection suddenly became inaccessible. The video did not freeze, but simply displayed less detail temporarily. Video file sizes could be reduced by about half with better quality in more complex scenes.
WaveOne received $9 million from sponsors including Khosla Ventures, Vela Partners, Incubate Fund, Omega Venture Partners, and Blue Ivy.
Apple has apparently acquired the business in order to improve streaming. Even minor improvements in video compression can save bandwidth or result in higher resolutions and frame rates, depending on the content being streamed, for services like Apple TV+.
YouTube is already doing this, as an example, when Alphabet's DeepMind integrated a board game learning algorithm into YouTube's video compression process last year, which allowed YouTube to reduce the amount of data it sends to users by 4%.
Apple is also testing Siri's "unwieldy design" which compels basic changes, like adding new suggestions, to the entire Siri database, which can take up to six weeks, while adding more advanced features, like new searching tools, can take up to a year.