Google Research has unveiled a new edge AI chip that reduces power consumption by 40% compared to previous versions. The announcement took place during the annual Research@ event at the company's headquarters in Mountain View, California, on October 14, 2026.
What happened?
The edge AI chip is part of the Astra line, developed in partnership with Synaptics. It uses the M2, successor to the Coral Neural Processing Unit (NPU), Google's low-power architecture for on-device AI processing. Edge AI technology allows calculations for voice, video, motion, and context detection to occur on wearables without sending data to the cloud, which reduces latency and increases privacy.
Avinatan Hassidim, vice president of Google Research, stated that conventional edge chips are slow and consume too much energy, making an always-on (ambient) AI experience impossible. The new Coral NPU M2 version uses 40% less energy. More details can be found in reports such as CHOSUNBIZ and Seoul Economic Daily.
Why does this matter?
Wearable devices such as smartwatches and smart glasses require extreme efficiency because they have small batteries and generate little heat. Processing AI locally avoids dependence on connectivity and reduces the risk of sensitive data leaks. Rahul Patel, CEO of Synaptics, said the chip will not remain in the lab and will be built into Google devices, although without specifying which ones or when.
Synaptics plans to disclose detailed specifications and demonstrations at its own event on October 20. This advance in the edge AI chip reinforces the trend of on-device processing for privacy and autonomy in everyday gadgets, enabling more practical ambient AI experiences in consumer products.
Sources: CHOSUNBIZ and Seoul Economic Daily reports based on Google's Research@ event.
By GeekikiBot
