News

Artificial Intelligence

Americas

Ambiq Launches Atomiq as the First Ultra-Low Power NPU SoC for Edge AI

The new SPOT-based processor enables complex vision and reasoning on battery-powered devices with industry-leading efficiency.

The new SPOT-based processor enables complex vision and reasoning on battery-powered devices with industry-leading efficiency.

The new SPOT-based processor enables complex vision and reasoning on battery-powered devices with industry-leading efficiency.

NewDecoded

Published Jan 7, 2026

Jan 7, 2026

3 min read

Image by Ambiq

Ambiq Micro (NYSE: AMBQ) has introduced Atomiq, a new system-on-chip that integrates a Neural Processing Unit for real-time AI at the edge. Built on the company's SPOT platform, the processor allows battery-powered devices to perform complex tasks like vision and reasoning without frequent recharging. This launch marks the first time an NPU has been optimized for sub-threshold voltage operation to achieve significant energy savings.

The hardware combines high-performance compute with extreme efficiency by operating transistors at voltage levels far below industry standards. This architecture enables the Arm Ethos-U85 NPU to deliver over 200 GOPS of on-device performance for demanding AI workloads. Manufacturers can now deploy sophisticated models for computer vision and speech recognition in compact devices that were previously limited by thermal or battery constraints.

To support rapid adoption, Ambiq is providing the Helia AI platform and the modular neuralSPOT software kit. These tools are designed to reduce the memory footprint and optimize power consumption for a faster product development cycle. This hardware-software integration ensures that developers can deploy energy-efficient intelligence across diverse markets without complex custom coding.

Adoption of this technology is already visible through strategic partnerships with companies like Bravechip. Their new chiplet platform for smart rings has reduced manufacturing costs by 85 percent while increasing production yields by 20 percent. This progress allows for continuous health monitoring and gesture detection in wearables that last all day on a single charge. In the industrial sector, Ambiq has worked with Ronds to deploy more than 400,000 intelligent sensors for predictive maintenance. These devices operate independently of the cloud in harsh environments where running power cables is often impossible. The Atomiq platform will soon allow these sensors to utilize high-fidelity audio analysis to identify machinery faults automatically before they occur. Looking toward the future, Ambiq plans to share more details about its 12nm SPOT roadmap at the Embedded World Conference in Germany this March. This next step highlights a commitment to scaling ambient intelligence across healthcare and smart infrastructure. With over 280 million devices powered to date, the company continues to drive the transition toward decentralized intelligence.


Decoded Take

Decoded Take

Decoded Take

The introduction of Atomiq signals a shift from simple low-power microcontrollers to high-performance AI processors that live entirely at the edge. By integrating a dedicated NPU into a sub-threshold voltage domain, Ambiq is addressing the primary barrier to widespread AI adoption: the extreme energy requirement of complex models. This move decentralizes intelligence, shifting heavy workloads away from power-hungry data centers and onto the local device. For the industry, this represents a significant step toward sustainable AI, where data privacy and energy conservation are achieved through localized, always-on processing.

Share this article

Related Articles

Related Articles

Related Articles