Qualcomm · Filed Dec 4, 2024 · Published Aug 6, 2026 · verified — real USPTO data

Qualcomm Patents a Video Chip That Cuts Its Own Power During Low-Quality Streams

Qualcomm has filed a patent for a video processing chip that monitors the complexity of a video stream in real time and automatically slows down or speeds up parts of itself to match, saving power when it doesn't need to run at full speed.

Qualcomm Patent: Bitrate-Based Video Chip Power Scaling — figure from US 2026/0230652 A1
Figure from the official USPTO publication.
See all 15 drawings from this filing ↓
Publication number US 2026/0230652 A1
Applicant QUALCOMM Incorporated
Filing date Dec 4, 2024
Publication date Aug 6, 2026
Inventors Naveen Kumar Ponnusamy, Shengqi Yang, Kishore Chamarthi
CPC classification 375/240.18
Grant likelihood Medium
Examiner LIMA, FABIO S (Art Unit 2486)
Status Notice of Allowance Mailed -- Application Received in Office of Publications (May 26, 2026)
Document 20 claims

How Qualcomm's video chip reads the stream and adjusts itself

Imagine your car automatically shifting into a lower gear on a flat road to save fuel, then revving back up when you hit a hill. Qualcomm's patent describes a video chip that does exactly that kind of thing, but for playing or recording video on your phone.

Right now, video chips in phones often run at a fixed speed regardless of whether you're streaming a low-quality video call or a high-bitrate 4K movie. Qualcomm's design splits the chip into two parts: one that handles the structural information in a video file, and one that handles the actual pixels. The pixel-crunching part can have its clock speed or power adjusted on the fly based on how demanding the video stream actually is.

If you're watching a simple, low-bitrate clip, the chip eases off. If you switch to something dense and detailed, it ramps back up. The result is a chip that works only as hard as it needs to, which in theory means less heat and longer battery life without sacrificing video quality.

How the controller splits syntax work from pixel work

The patent describes a video coding chip with two distinct processing engines and a controller that mediates between them.

Video Syntax Processing (VSP) engine: This part handles the structural layer of a video stream, things like headers, flags, and the encoded instructions that describe how frames are organized. This work is relatively lightweight and steady.

Video Pixel Processing (VPP) engine: This is where the heavy lifting happens. The VPP takes the decoded instructions from the VSP and actually reconstructs or encodes pixel data, the raw picture information. Pixel work scales directly with bitrate: a high-bitrate 4K stream demands far more from this engine than a compressed 480p clip.

Controller: A controller sits above both engines and continuously measures the bitrate of the current video session. Based on that measurement, it raises or lowers the clock speed (how many operations per second the chip runs) or adjusts the power state of the VPP engine specifically. The VSP keeps running at whatever pace it needs to, since its workload doesn't change as dramatically.

The key insight is that the pixel-processing workload is tightly correlated with bitrate, so bitrate becomes a practical proxy for deciding how much power the chip should draw at any given moment.

We find one patent like this every day. Get the best of each week in your inbox, free →

What this means for phone battery life and video quality

For phone makers and chip designers, power consumption during video playback and recording is a persistent problem. Video apps, camera apps, and streaming services all push the hardware hard, and that drains batteries and generates heat. A chip that can throttle its pixel-processing engine during low-bitrate sessions without touching the rest of the pipeline could reduce power draw in a meaningful, targeted way.

For you as a user, this could mean your phone stays cooler during a long video call, or your battery lasts a bit longer when you're streaming a compressed clip over a weak connection. It also has implications for Qualcomm's chips inside streaming devices, laptops, and automotive screens, not just phones, anywhere video decoding runs continuously for long periods.

Editorial take

This is solid, unglamorous chip engineering. Splitting a video pipeline into syntax work and pixel work, then scaling only the expensive part, is a logical approach to a real problem. It won't make headlines the way a new AI feature does, but dynamic power scaling in dedicated video hardware is exactly the kind of incremental work that adds up to noticeable improvements in real-world battery life.

There are more where this came from

We read every patent application Big Tech publishes and send you the ones worth knowing. Plain English, free, every week.

The drawings

15 drawing sheets from US 2026/0230652 A1 · click any drawing to enlarge

Patent filing page

Source. Full patent text and figures from the official USPTO publication PDF.

Editorial commentary on a publicly published patent application. Not legal advice.