Apple buys WaveOne to shrink videos with AI

There is nothing more annoying than a video that freezes the second your Wi-Fi drops by a single bar. Traditional video codecs have worked the same way for decades, and fixing that lag usually requires waiting years for new hardware chips to roll out.

Apple just quietly snapped up a Mountain View startup trying to solve that exact headache with machine learning. The company, WaveOne, was building AI algorithms designed to completely rethink how video gets compressed and streamed.

Apple has not officially confirmed the purchase, but WaveOne shut down its website around January. Multiple former employees, including one of its co-founders, now list roles across Apple's machine learning teams. WaveOne’s former head of sales, Bob Stankosh, confirmed the sale in a LinkedIn post, noting that Apple saw the potential to add their tech to its portfolio.

WaveOne was founded in 2016 by Lubomir Bourdev and Oren Rippel, both Meta AI veterans who worked on Facebook’s computer vision and feed ranking. Before getting scooped up, the startup raised $9 million from investors like Khosla Ventures and Vela Partners.

Smarter compression that prioritizes faces

Standard video compression happens on servers like YouTube's, while your phone handles the decompression. The catch is that new, more efficient codecs usually require dedicated silicon to decode without melting your battery, meaning upgrades roll out painfully slow.

WaveOne took a different route. Its content-aware algorithm is hardware-agnostic and runs directly on the AI accelerators already packed into modern phones and laptops.

Basically, the AI understands what is happening in a frame. It can spot a person talking and prioritize crisp details on their face while cutting data on a blurry background wall to save precious bandwidth.

Even better, the tech handles bad connections gracefully. Instead of freezing and buffering when the network drops, WaveOne’s algorithm makes a best guess with the bits it has available, lowering the resolution temporarily while keeping playback smooth. The team claimed it could shrink video files by up to half.

Why it matters

Bandwidth is expensive, and streaming efficiency is an invisible war. Last year, DeepMind adapted a board-game AI to compress YouTube videos and squeezed out a 4% data reduction, which was already considered a massive win. If WaveOne’s tech can deliver even a fraction of its promised 50% reduction, Apple could save millions on Apple TV+ infrastructure.

For everyday users, this could quietly supercharge services like FaceTime and Apple TV+. It means fewer dropped calls in poor-coverage areas and higher frame rates on cellular connections without burning through data caps.

That said, we think the real hurdle will be power efficiency. Running real-time generative or predictive models on on-device neural engines consumes battery. Apple will need to prove that AI-driven decompression can save bandwidth without quietly cooking your pocket.