Google and Synaptics have launched the Astra SL2610/SL2600 series, a line of low-power edge AI microcontrollers. The platform integrates Google’s open-source Coral NPU M2 with Synaptics’ Torq NPU, Arm CPU cores, and multimedia processing hardware.
The chip is designed for simultaneous multimodal processing, including voice, vision, motion, and contextual sensing. It consumes less than 6 milliwatts for average on-device inference, which is approximately 40% more energy-efficient than existing edge AI processors.
The companies developed an open-source software stack over 18 months using MLIR/IREE-based tools to standardize machine learning frameworks for IoT applications. The partnership was originally announced in January 2025.
Developer platforms, such as the Coralboard, were presented at Google I/O 2026. Following a preview at the Research@ event on October 14, 2026, a formal demonstration is scheduled for Synaptics’ Tech Day on October 20, 2026.
