STEPX Neo Phone Adds Syntiant Neural Processing Power
The integration enables the phone to perform continuous audio sensing at lower power consumption for LLM-native tasks.
Updated on Oct. 15, 2026 in Artificial Intelligence
Live Poll
Do you trust smartphones that use always-on sensors to monitor your surroundings for context?
Irvine-based Syntiant Corp has integrated its NDP120 Neural Decision Processor into the STEPX Neo, a new LLM-native agentic phone. The integration utilizes a six-microphone array to enable low-power, always-on audio sensing for the device.
Why it matters
This integration enables edge devices to capture context before a user prompt is issued, potentially reducing the frequency with which a device must wake its main processor to perform tasks. By running audio sensing locally and continuously at low power, the system aims to improve efficiency while maintaining user privacy.
The STEPX Neo utilizes the NDP120 Neural Decision Processor and an array of six SiSonic MEMS microphones. This hardware setup facilitates continuous, low-power audio sensing, distinguishing it from standard architectures that typically require higher power consumption for constant interaction.
The players
Syntiant Corp
An Irvine-based semiconductor company that designs neural processing hardware for low-power edge AI applications.
STEPX
A Shanghai-based AI firm that develops LLM-native agentic smartphone hardware.
The details
The system utilizes the NDP120 — a neural decision processor specialized for ultra-low power inference — to process audio signals at the edge. The array of SiSonic MEMS microphones (micro-electro-mechanical systems, which are tiny sensors that convert sound into electrical signals) continuously monitors for ambient context. By capturing this data locally, the phone can provide context to the agent before a prompt, minimizing the time the primary power-hungry application processor stays active.
Timeline
Syntiant Corp was founded in 2017.
The integration announcement occurred on October 15, 2026.
The Tech Race
The use of the NDP120 in the STEPX Neo follows a growing industry trend toward offloading voice processing to dedicated edge silicon. This strategy positions Syntiant as a key supplier in the race to develop LLM-native phones that require constant, low-latency audio awareness.
Users of the STEPX Neo may experience improved battery life and faster agent response times due to the pre-processing of audio context. The technology currently remains specific to this device, with no timeline provided for integration into broader consumer hardware ecosystems.
The takeaway
The move demonstrates the increasing necessity of dedicated silicon to sustain local, always-on AI agents in mobile form factors. Observers should watch for future benchmark reports comparing the power efficiency of the NDP120 setup against traditional high-power processor wake-word architectures.
Further reading
For broader trends in device intelligence, explore the Artificial Intelligence section.
Live Poll
Do you trust smartphones that use always-on sensors to monitor your surroundings for context?
