Loading article…
Google unveils Tensor G5 with 34% faster CPU and 60% stronger TPU, powering Pixel 10 AI features and on‑device privacy.
Google’s new Tensor G5 chip, revealed at the Made by Google 2025 event, delivers a 34% CPU speed increase and a 60% TPU boost over its predecessor, immediately raising expectations for AI‑driven features on the Pixel 10 lineup【1】.
| At a glance | |
|---|---|
| CPU speed | 34% faster than Tensor G4 |
| TPU performance | 60% stronger than Tensor G4 |
| Process node | 3 nm TSMC fabrication |
| AI model support | Gemini Nano runs 2.6× faster, 50% less energy |
The Tensor G5 is built on TSMC’s 3‑nanometer process, allowing more transistors per square millimeter and better power efficiency than the Samsung‑fabricated Tensor G4【2】. Google claims the CPU delivers a 34% average speed uplift, while the dedicated TPU gains a 60% performance edge, translating into quicker app launches and smoother multitasking【1】. The chip also introduces a new Matryoshka transformer architecture that can handle up to 32,000 tokens of context—roughly the content of a month’s worth of emails—enabling more ambitious on‑device AI tasks【1】.
Tensor G5 powers over 20 new AI‑driven tools in the Pixel 10 series, including Magic Cue prompts, Voice Translate with speaker‑matched voices, and Call Notes that auto‑generate action items【1】. All these functions run locally, meaning personal data never leaves the device, a claim Google backs with the integration of the Titan M2 security chip for hardware‑level protection【1】. The new ISP adds 10‑bit video capture at 1080p and 4K 30 fps, while the Pro Res Zoom feature on Pixel 10 Pro leverages a diffusion model with nearly a billion parameters to achieve 100× hybrid zoom without sacrificing detail【1】.
Benchmark tests show the Tensor G5 outpaces the prior Tensor G4 and Samsung’s Exynos 2500 in single‑core CPU scores, but it still trails Qualcomm’s Snapdragon 8 Elite and Apple’s A18 Pro in multi‑core performance and synthetic benchmarks【2】. GPU performance appears limited; early reports suggest a PowerVR‑based GPU that omits ray‑tracing support, potentially reducing appeal for high‑end mobile gamers【1】. Nonetheless, the chip’s efficiency gains—longer battery life despite heavier AI workloads—address past overheating concerns that plagued earlier Tensor generations【2】.
The Tensor G5 marks Google’s most ambitious push toward an AI‑first smartphone, emphasizing on‑device processing and privacy. Whether the performance gains narrow the gap with rival chipmakers remains to be seen, but the chip sets a new baseline for integrating generative AI into everyday mobile experiences.
Coverage is mostly measured — 15 of 15 reports stay neutral.
Every Monday — the token unlocks, Fed dates & catalysts set to move crypto and markets this week. So you’re never blindsided.
Free · 3-min read · one-click unsubscribe
AI-assisted synthesis by the TrendWatcher Editorial Desk · sourced from 2 outlets · Jun 23, 2026 · How we report
It is an inflatable sleeping pad intended for backcountry camping, offering lightweight, packable, and insulated sleeping comfort.
By using a refreshed, older GPU design and a 7‑core CPU on the 2nm N2 process, Google seeks to control costs while improving efficiency over the previous generation.
It incorporates two ultrathin layers of Thermal Mirror metalized film within each baffle to reflect body heat and minimize crinkling sounds.
The launch is anticipated for August 2026.
Its 3.5‑inch thickness requires more air to inflate, which can increase lung effort and may necessitate an electric pump for convenience.