Google’s Pixel 10 on-Device Sustainable AI: Explained

Google is betting that consumers want smartphones that are both intelligent and environmentally responsible as the search giant unveils its Pixel 10 device. 

The phone showcases how AI capabilities can coexist with sustainability initiatives in consumer electronics manufacturing.

The company has built the device around its custom Tensor G5 processor, developed alongside Google DeepMind, its AI research arm. 

This collaboration has produced a chip capable of running Gemini Nano, Google’s compact AI model designed specifically for smartphone hardware.

The approach demonstrates the broader shift towards edge computing, where AI processing happens locally on devices rather than in remote data centres. 

This method reduces the need for constant internet connectivity while addressing privacy concerns that arise when personal data gets transmitted to external servers for processing.

Kate Brandt, Chief Sustainability Officer at Google, says: “The new Pixel 10 is built with at least 32% recycled materials by weight, the most recycled content of any Pixel generation yet. Plus, all of Google’s packaging remains 100% plastic-free.”

How Tensor G5 powers Magic Cue AI 

The sustainability strategy tackles electronic waste through three main areas: building devices that last longer, sourcing recycled materials and making repairs easier. 

The phone’s shell uses 100% recycled spacecraft-grade aluminium, which provides strength without excess weight while reducing demand for newly mined materials.

Google has committed to seven years of software updates for the Pixel 10, far exceeding typical Android device support periods. 

This extended timeline addresses a key factor in smartphone obsolescence, where devices become unusable due to outdated software rather than hardware failure.

The real innovation lies in how the Tensor G5 chip processes AI tasks without relying on cloud services. 

This on-device approach allows the phone to analyse user data from Gmail, Calendar, Messages and Photos while keeping information secure on the handset.

Stephanie Scott, Group Product Manager at Google, says: “Tensor G5, our new custom silicon, is the most significant upgrade to our chip since Tensor’s debut. 

“It delivers snappy performance while bringing deeply helpful experiences to Pixel users first.”

The chip powers Magic Cue, an AI assistant that proactively surfaces helpful information across different apps. 

Rather than waiting for user commands, the system anticipates needs by connecting data from various sources to provide timely suggestions.

Stephanie explains: “For instance, when you’re calling an airline, it can automatically find your flight details from your email and display it during your phone call. 

“It helps you both recall information and take action, too – whether it’s providing an address for dinner or finding the right photos to share with a friend, all without leaving your chat.”

This functionality demonstrates practical applications of on-device AI that go beyond novelty features to address real user frustrations with managing information across multiple apps.

WHy circular design principles guide hardware decisions

Google has redesigned the Pixel 10 with repairability in mind, incorporating high-contrast screws and colour-coded repair points that simplify component replacement. 

The company has also streamlined its spare parts ordering process to encourage repairs rather than replacements.

These design choices support circular economy principles, where products are built for durability and easy maintenance rather than disposal. 

The approach helps address the smartphone industry’s contribution to electronic waste, where consumers typically replace devices every 2.5 years despite remaining functionality.

Kate says: “This is a milestone that reflects a decade of dedicated work in recycled materials from our teams. Our work isn’t just about big goals – it’s about the intentional efforts of countless people across our organisation who are committed to building for everyone and the planet.”

Stephanie concludes: “And, thanks to co-design with Google DeepMind, the newest Gemini Nano model will run first on Tensor G5 to unlock many on-device generative AI experiences that make everyday life easier.”