Azure Percept Pricing And Features Edge Ai Computing Solution

Azure Percept: Pricing, Features, and Retirement Guide

Azure Percept was Microsoft's integrated edge AI hardware and software platform. Microsoft retired Azure Percept in March 2023. The underlying edge AI capabilities now live in Azure IoT Edge, Azure AI Services, and partner hardware through the Azure Certified Device program. EPC Group helps organizations migrate from Azure Percept to the modern Azure edge AI stack. 29 years of Microsoft experience.

Key facts

• Azure Percept was retired in March 2023. Microsoft ended support at that date.
• Azure Percept DK hardware: carrier board with an Intel Movidius Myriad X VPU for computer vision.
• Azure Percept Audio accessory: Arm Cortex-M audio processor for speech and voice AI.
• Modern replacement stack: Azure IoT Edge + Azure AI Services + Azure Certified Device hardware.
• Migration path: Azure Machine Learning → ONNX export → Azure IoT Edge module.
• EPC Group: 29 years Microsoft consulting, 10,000+ enterprise deployments.

What Was Azure Percept?

Azure Percept launched in 2021 as an integrated edge AI platform. It combined hardware, software, and cloud services in one solution for organizations exploring edge AI without deep embedded systems expertise.

The Azure Percept DK was a development kit. It included a carrier board with a vision System-on-Module (SoM) featuring an Intel Movidius Myriad X VPU. The Azure Percept Audio accessory added speech and voice AI capabilities.

Microsoft retired Azure Percept in March 2023. The capabilities have been absorbed into the broader Azure IoT and AI ecosystem.

Azure Percept Retirement: What Changed

Microsoft retired Azure Percept because the edge AI ecosystem matured. The specific capabilities Percept provided now exist as more flexible, mature services:

• Vision AI at the edge → Azure Custom Vision + Azure IoT Edge (Docker container deployment)
• Speech AI at the edge → Azure Cognitive Services Speech SDK on Azure Certified Device hardware
• Hardware platform → Azure Certified Device program with 1,000+ certified partner devices
• Cloud management → Azure IoT Hub and Azure IoT Central (unchanged)
• AI development workflow → Azure Machine Learning (model training and ONNX export)

Modern Azure Edge AI Stack (Post-Percept)

The recommended workflow for edge AI in 2026 — replacing Azure Percept:

1. Train the model — Use Azure Machine Learning or Azure AI Custom Vision to train computer vision or speech models.
2. Export to ONNX — Export the trained model in ONNX format for edge deployment.
3. Package as a container — Build a Docker container with ONNX Runtime and your model weights.
4. Deploy via Azure IoT Edge — Push the container to edge devices through Azure IoT Hub's deployment manifests.
5. Monitor and update — Azure IoT Hub monitors device health and deployment status. Update models remotely without physical device access.

This stack is more flexible than Azure Percept. It runs on any Azure Certified Device — not just proprietary Percept hardware.

Selecting Edge AI Hardware (Post-Percept)

With Azure Percept retired, organizations choose hardware from the Azure Certified Device catalog. Key hardware categories:

• NVIDIA Jetson — Best for GPU-accelerated vision AI inference (object detection, classification, segmentation).
• Intel NUC and UP boards — General-purpose x64 edge compute with Intel OpenVINO for vision acceleration.
• Raspberry Pi (Certified) — Lower-cost option for lightweight edge AI workloads without strict real-time requirements.
• Advantech and Kontron — Industrial-grade rugged hardware for manufacturing and OT environments.
• Qualcomm AI Hub — AI-optimized NPU hardware for power-efficient edge inference on ARM devices.

EPC Group Edge AI Consulting

EPC Group helps organizations migrate from Azure Percept to the modern edge AI stack, and design new edge AI deployments on Azure IoT Edge. Our services:

• Hardware selection: evaluate Azure Certified Devices for your compute, power, and ruggedization requirements.
• Model development: train computer vision or speech models in Azure Machine Learning.
• ONNX export and container packaging for Azure IoT Edge deployment.
• Azure IoT Hub configuration: device management, deployment manifests, and remote updates.
• Compliance: HIPAA, NERC CIP, and FedRAMP controls at the edge device level.

Frequently asked questions

Is Azure Percept still available?

No. Microsoft retired Azure Percept in March 2023. The hardware and dedicated cloud services are no longer available for new purchases. Existing deployments should migrate to Azure IoT Edge with Azure Certified Device hardware and Azure AI Services.

What replaced Azure Percept?

Azure Percept capabilities are now delivered through Azure IoT Edge (runtime and deployment), Azure AI Services (Custom Vision, Speech), Azure Machine Learning (model training and ONNX export), and the Azure Certified Device program (1,000+ certified hardware platforms).

How do I migrate from Azure Percept?

Retrain or export your Azure Percept models to ONNX format. Package the ONNX model with ONNX Runtime in a Docker container. Deploy the container to your new hardware via Azure IoT Edge. Configure Azure IoT Hub for device management. EPC Group provides fixed-fee migration engagements.

What hardware should I use for edge AI after Azure Percept?

Choose from the Azure Certified Device catalog. NVIDIA Jetson for GPU inference. Intel NUC or UP boards for x64 workloads. Advantech or Kontron for industrial/OT environments. Qualcomm AI Hub for power-efficient ARM deployments. The right choice depends on compute requirements, power budget, and environmental conditions.

Does EPC Group help with Azure Percept migration?

Yes. EPC Group provides fixed-fee Azure Percept migration engagements covering hardware selection, model migration to ONNX, Azure IoT Edge deployment, and Azure IoT Hub configuration. Contact us for a scoping call.

Start your edge AI modernization

Talk to an EPC Group architect about Azure Percept migration or new edge AI architecture on Azure IoT Edge. Call (888) 381-9725 or request a 30-minute discovery call.