Google is gearing up to address persistent issues with its Pixel smartphones through its latest custom chip, the Tensor G5. According to Android Authority, the new processor, slated to power the upcoming Pixel 11, is being designed to improve thermal performance and battery life, which have long been common sources of user frustration.
Background: The Path from Tensor G1 to G4
Understanding the development of Google’s custom silicon provides crucial context for the G5. The journey began with the Tensor G1 in 2021, co-developed with Samsung, emphasizing AI features like computational photography and real-time language translation.
However, it fell short in power efficiency and heat management. The Tensor G2 (2022) followed with minor architectural tweaks but failed to solve battery and thermal problems which Google tried to address in Android 14.
By 2023, the Tensor G3 adopted a 64-bit architecture built on Samsung’s 4nm process. While it strengthened AI capabilities, users still experienced heat issues and limited CPU upgrades. The Tensor G4 in 2024 delivered slight improvements, mainly in mid-cluster cores, but retained an older GPU design, limiting graphical advances.
Heat and Battery Performance: A Persistent Issue in Pixel Phones
One of the most notable challenges faced by Google’s Pixel line is thermal inefficiency. Internal reports highlighted that approximately 28% of Pixel phone returns were due to overheating, making it a critical area for improvement.
To counter this, the Tensor G5 will include a “Cinematic Rendering Engine“, which aims to cut power use during video recording with blurred backgrounds by 40%, reducing the heat generated during demanding tasks.
Battery life has also been a point of contention. According to user feedback, fewer than 86% of Pixel 6 and 7 owners could get through a full day on a single charge. This shortcoming has driven Google to make battery efficiency a key focus for upcoming Tensor chips.
Design and Efficiency Changes in Tensor G5 and G6
The latest leaks indicate that the Tensor G5 will feature one Cortex-X4 core, five Cortex-A725 mid-cluster cores, and two Cortex-A520 efficiency cores. These changes reflect Google’s focus on improving multitasking performance while maintaining power efficiency. This configuration suggests a slight boost in speed but does not indicate a leap in overall performance. Notably, these cores will be fewer in number compared to previous generations, which suggests Google is prioritizing mid-cluster performance over efficiency cores.
The Tensor G6, slated for 2026, will use a more compact design, measuring around 105 mm², compared to the 121 mm² of the Tensor G5. This reduction is achieved through TSMC’s N3P process node, known for either lowering power consumption by 9% or boosting performance by 4% at the same power level. These attributes will make it ideal for Google’s focus on enhancing battery life and heat management.
However, the downsizing comes with compromises. According to a recent leak, the GPU in the Tensor G6 will be an older IMG CXT model, originally planned for the Tensor G4, allowing for a 12% die area reduction but potentially limiting performance. The Tensor G5 will feature a stronger Imagination Technologies DXT GPU with ray tracing support. This means that the digital signal processor (DSP) will lose one core in the Tensor G6, and the system-level cache will be halved to 4 MB to conserve space.
Integrated AI and Software Synergy
Google’s use of custom silicon provides it with the advantage of deeper integration between hardware and software. This is particularly evident in features like Night Sight photography, real-time transcription, and voice recognition, which leverage the Tensor Processing Unit (TPU) built into the chip. Such integration allows for faster machine learning processes and a more cohesive user experience compared to devices running third-party processors.
Security enhancements, including the Titan M2 security chip, are expected to remain part of the architecture, ensuring that sensitive data processing stays secure within the hardware. This level of integration positions the G5 as more than just a chip—it is a central part of Google’s strategy to offer unique user-centric features.
Competitive Context: How Tensor G5 Stacks Up
While the Tensor series has historically been designed to optimize specific use cases, it has lagged behind Qualcomm’s Snapdragon series and Apple’s A-series chips in raw performance. For example, Qualcomm’s Snapdragon 8 Elite and Apple’s A18 Pro, both built on TSMC’s N3E process node, boast superior power efficiency and battery life. The A18 Pro, with its 105 mm² die size, exemplifies how a smaller chip can maintain high performance.
Google’s reliance on TSMC’s N3P process for the G5 could bridge some of these gaps by focusing on power efficiency and improved thermal management. However, the emphasis on user experience rather than benchmark results might not appeal to performance-focused users.
The move to a more efficient chip could extend the device’s lifespan by reducing thermal wear on components, potentially lowering the frequency of replacements and reducing electronic wasteImproved battery life and reduced heat mean longer recording times and better performance during demanding tasks, such as video processing.
