NXP microcontrollers and applications processors with machine learning at the edge enable smart thermostats, industrial robotics and intelligent vehicles. Machine learning at the edge allows NXP processors to learn, adapt and make decisions. This allows NXP hardware to overcome cloud latency and bandwidth challenges. NXP edge AI processor portfolio and eIQ software toolkits enable faster responses, reduced power consumption, enhanced privacy and lower operating costs.
Artificial Intelligence
Empowering the intelligent edge with AI and machine learning
Innovating at the Edge with AI
Expansive Edge AI Processor Portfolio
Heterogeneous AI acceleration for scalable edge intelligence.
eIQ® AI Software Stack and Development Tool
Enables a unified workflow with tools and pipelines for edge AI model optimization, deployment, explainability and security.
Responsible Edge AI Enablement
Making AI work in the real world requires looking beyond tools and technology to ensure it is safe, reliable, and responsible. That’s where Responsible AI takes center stage-bringing together technology, government, and business leaders to make it a reality
Embed Artificial Intelligence at the Edge
NXP AI solutions enable machines and vehicles to learn and reason like humans with NXP toolkits for machine learning and deep learning.
Embedded Vision
Technologies
Enable devices that process visual data locally at the edge, enabling real-time, private and low-latency decisions.
Audio
Technologies
Audio and voice processing solutions that offer high-quality, reliable voice control and communication software for low-power, small-footprint applications.
Time Series
Technologies
An AI/ML toolkit for automated machine learning (autoML) for time series AI generation and optimization to microcontrollers and apps processors.
Generative AI
Technologies
eIQ Gen-AI Flow brings LLMs/SLMs and Gen-AI models to NXP processors, supporting RAG and on-device speech features.
Agentic AI
Technologies
NXP’s Agentic AI solutions enhance autonomy at the edge by enabling AI-driven Agents for real-time, autonomous decisions.
eIQ NPU
Technologies
The eIQ® Neutron NPU is a scalable ML accelerator core, optimized for power and performance, supporting various neural networks and fully integrated with NXP's microcontrollers and processors.
Explore Retrieval Augmented Generation (RAG) Methodology
eIQs GenAI Flow provides building blocks like optimized Large Language Models (LLMs) and Retrieval Augmented Generation (RAG) to make generative AI applications accessible on edge devices.
Artificial Intelligence Applications
-
Industrial
Devices that observe, learn and adapt in real-time can give your business better foresight and insight, informed control and new capabilities. Adding intelligence to devices can enhance performance for products across a full spectrum of commercial applications.
-
Consumer
ML-enabled devices such as door locks, thermostats and lights make autonomous decisions based on user preferences and real-time data for more safety, convenience and security.
-
Automotive
AI models deployed directly in the car enable real-time responses, enhanced privacy and security, and reduced bandwidth usage.
-
Healthcare
Smart system solutions to advance medical IoT devices and medicinal products to enable efficient processes, reliable product protection and improved patient safety and outcome with low-power efficiency, intelligence, security and wireless connectivity.
Design Resources
Documentation
| Document | Type | Date |
|---|---|---|
|
Responsible Edge AI Enablement Making AI work in the real world requires looking beyond tools and technology to ensure it is safe, reliable, and responsible. That’s where Responsible AI takes center stage—bringing together technology, government, and business leaders to make it a reality. |
White Paper | May 5, 2025 |
|
eIQs GenAI Flow provides building blocks like optimized Large Language Models (LLMs) and Retrieval Augmented Generation (RAG) to make generative AI applications accessible on edge devices. |
White Paper | Oct 25, 2024 |
|
Intellectual Property Aspects of ML This white paper sets out to add clarity to the legal background and challenges for the IP aspects of ML for data collection and model building. |
White Paper | Jul 13, 2023 |
|
Our ebook ‘Essentials of Edge Computing' shares insights into machine learning inference at the edge, such as uninterrupted processing, lower latency and user privacy. |
Brochure | Jan 21, 2022 |
FAQs
How does NXP empower edge AI development across hardware and software?
NXP accelerates edge AI through high-performance processors that integrate NPUs, GPUs, CPUs, and DSPs—enabling efficient machine learning at the edge. Its unified eIQ AI Software Stack supports "bring your own model" and "bring your own data" workflows across MCUs and MPUs, while promoting privacy by processing data locally. Strategic partnerships further streamline AI integration into embedded systems.
Can NXP products run LLMs, VLMs, or GenAI models at the edge?
Yes, NXP enables LLM, VLM, and GenAI deployment on embedded devices through eIQ® GenAI Flow—a modular pipeline for real-time, voice-based AI interaction. Integrated NPUs deliver low-latency, energy-efficient inference, and for larger models, NXP supports discrete AI accelerator integration to meet high-throughput requirements.
What AI applications are powered by NXP processors across industries?
NXP processors support a wide range of edge AI applications:
- Industrial: Machine vision, predictive maintenance, voice/gesture HMI.
- Smart Home & IoT: Voice assistants, behavior-aware appliances, facial recognition.
- Consumer Electronics: Health wearables, smart displays, AR interfaces.
- Automotive: ADAS, in-cabin monitoring, predictive maintenance, battery management.
What is the eIQ AI Software Stack and how does it support AI development?
NXP’s eIQ AI SW Stack enables flexible AI development with support for TensorFlow, PyTorch, ONNX, and BYOM/BYOD workflows. It includes tools like eIQ Time Series Studio for sensor-based AI, eIQ GenAI Flow for LLM fine-tuning, and model watermarking for IP protection. Optimized for low-latency, energy-efficient inference, it supports full pipelines for embedded vision and multimedia across MPUs and MCUs.
Does NXP offer pre-trained models and application examples for edge AI?
Yes, NXP provides a model zoo with pre-trained models for vision, audio, and other domains, optimized for its MCUs and MPUs. These include conversion and fine-tuning recipes. The Application Code Hub (GoPoint) offers demonstration applications and tutorials for object recognition, face detection, voice AI, handwritten digit recognition, and LLMs—accelerating development with real-world examples.
How does NXP ensure safe, ethical, and responsible AI at the edge?
NXP promotes responsible AI through a five-principle ethics framework: non-maleficence, human autonomy, explainability, vigilance, and privacy-by-design. It emphasizes fairness, robustness, and inclusivity, advocating for diverse datasets to prevent bias. NXP also supports AI literacy and workforce upskilling to ensure ethical deployment and long-term reliability in edge environments.