Google Pixel 10 and 11 leak reveals Tensor G5 AI camera upgrades

Google is reportedly gearing up for a massive leap in AI-powered photography and video editing with its upcoming Pixel 10 and Pixel 11 flagships.

According to leaked documents from Google's internal gChips division, both upcoming generations will see deep architectural and functional overhauls. The heavy lifting will be driven by the anticipated Tensor G5 chip, which is expected to shift to TSMC's 3nm manufacturing process for drastically better thermal and power efficiency.

Voice-guided edits and sketch-to-image

For the Pixel 10, Google is heavily focused on expanding generative tools directly inside the Google Photos app, powered by an upgraded TPU inside the Tensor G5.

One headline feature in the leak is dubbed Video Generative ML. The goal is to make post-capture video editing intuitive by letting generative models handle adjustments on the fly, without manual timeline tweaking.

Google is also working on Speak-to-Tweak, a voice-activated editing tool that lets you alter photos simply by talking to your phone. Another feature, Sketch-to-Image, will allow users to turn rough hand-drawn doodles into polished, detailed images right on the device.

100x zoom and candlelight night video

The leak reveals even more ambitious camera hardware updates planned for the Pixel 11.

Google plans to introduce up to 100x zoom for both photos and videos, powered by a next-gen telephoto lens teamed up with machine learning. The leak does note, however, that quality gaps between still photography and moving video will likely persist at extreme focal lengths.

Video recording under rough conditions will also get a major pass. Google is prepping Ultra Low Light Video, also referenced as Night Sight Video, designed to record clean footage in environments as dim as candlelight entirely on-device. Meanwhile, Cinematic Blur will receive a power consumption reduction.

Why it matters

For creators and casual mobile photographers, intuitive post-processing without third-party desktop software has long been the dream. If Google pulls off direct voice adjustments and generative video tweaks on-device, it will eliminate massive friction in mobile content creation.

On the hardware front, switching the Tensor G5 to TSMC's 3nm node addresses the single biggest complaint Pixel owners have had for generations: heat and battery throttling under load. Running low-light video enhancement and real-time generation locally requires immense thermal headroom.

We think Google is smart to double down on video. Apple has owned the mobile video crown for years largely because real-time multi-frame computational video eats too much power. If Google can pull off 100x machine learning zoom and candlelight recording on-device without melting the phone, it finally gives the Pixel a genuine punch against the iPhone in moving pictures.