Next-Generation Edge AI Chipset Announced for Low-Power IoT Devices
Revolutionary Multimodal Gen AI Processor Delivers Sub-Watt Operation and Industry-First Coral NPU Integration for Intelligent Edge Devices
SAN JOSE, Calif. – October 21, 2025 – Synaptics Incorporated today announced the Astra™ SL2610, the first product in its breakthrough Astra SL2600 series of multimodal generative AI processors designed specifically for power-constrained Internet of Things (IoT) edge devices. The chipset represents a fundamental architectural advancement, enabling intelligent sensing and AI-driven decision making while consuming less than one watt during continuous operation.
The SL2610 redefines the traditional embedded SoC hierarchy by centering system control around a dual Arm Cortex-A55 CPU cluster operating at 2 GHz, complemented by a dedicated Cortex-M52 subsystem featuring Helium DSP extensions for real-time workloads. This inverted architecture allows devices to boot directly into ultra-low-power microcontroller (MCU) mode without initializing the application processor or operating system, eliminating wasted idle power in sensor-monitoring states.
“This allows our power-sensitive customers to actually boot the machine directly into low power mode,” said John Weil, Vice President and General Manager for Synaptics’ IoT and Edge AI Processor Business. “Directly in a microcontroller mode, you can shut off the A cores, or never even start them if you want.”
Power efficiency metrics demonstrate the chipset’s breakthrough capabilities. While competitive solutions like the NXP i.MX 8M Plus deliver 2.3 TOPS within a 3-8W envelope, and the Hailo-8 accelerator achieves 26 TOPS at 2.5W, the Astra SL2610 maintains persistent AI readiness below one watt through its novel mode-switching technology. When workload demands increase—such as invoking vision or speech models—the SoC dynamically activates the A-class cores and integrated neural processing unit, executes the AI task, and immediately returns to MCU mode, achieving deterministic real-time control while slashing average power consumption by up to 73 percent compared to conventional always-on architectures.
Market dynamics underscore the critical timing of this launch. The global IoT chip market reached $525.21 billion in 2024 and is projected to grow at a 9.3 percent CAGR to $895.05 billion by 2030 . This explosive growth is fueled by accelerating demand for ambient intelligence in consumer appliances, industrial automation, and wearable devices where battery life and thermal constraints dominate design requirements. Research indicates that edge AI chip revenue will surpass cloud AI chip revenue for the first time in 2025, driven by the availability of ultra-low-power processors purpose-built for distributed intelligence.
The SL2610 integrates Synaptics’ Torq™ NPU subsystem, the industry’s first production implementation of Google’s newly announced Coral NPU architecture. This hardware partnership addresses three critical barriers limiting powerful always-on AI: the performance gap between complex ML models and edge power budgets, the fragmentation tax of optimizing for diverse proprietary processors, and the user trust deficit requiring hardware-enforced privacy. The Coral NPU’s transformer-capable design supports dynamic operators, enabling future-ready edge AI systems that can evolve with emerging generative models.
Target applications span smart consumer appliances that maintain persistent environmental awareness, industrial IoT gateways performing predictive maintenance, and wearable devices delivering contextual assistance without compromising battery life. The chipset supports hardware-enforced security through PSA Level 3 certification, ECC-protected memory paths, and emerging CHERI technology for fine-grained memory safety, ensuring sensitive AI models and personal data remain isolated in hardware-enforced sandboxes.
“Embedding ambient AI onto battery-constrained devices requires rethinking silicon from first principles,” said Weil. “The Astra SL2610 doesn’t just add an NPU to a conventional SoC—it architecturally reimagines how edge devices wake, process, and sleep to make always-on intelligence practical at the milliwatt level.”
Manufacturing readiness positions the SL2610 for rapid OEM adoption. Samples are available now with mass production scheduled for Q1 2026. The platform includes an open-source compiler and runtime based on IREE and MLIR, providing developers a unified toolchain that eliminates vendor lock-in and accelerates model deployment across the Astra ecosystem.
About Synaptics
Synaptics is a leading developer of human interface solutions and edge AI processors for consumer electronics, IoT, and enterprise markets. With over 35 years of innovation in sensing and connectivity, Synaptics’ Astra platform is advancing the next generation of intelligent, context-aware devices that anticipate user needs while prioritizing privacy and power efficiency.
Media Contact
Sarha Al-Mansoori
Director of Corporate Communications
G42
Email: media@g42.ai
Phone: +971 2555 0100
Website: www.g42.ai
