Edge AI & IoT platforms enable artificial intelligence deployment and operation on edge devices and Internet of Things networks. These tools optimize AI models for resource-constrained environments while providing seamless integration with IoT infrastructure.
41 verified AI-first sites in Edge & IoT.
Last reviewed .
Leading edge AI development platform joining Qualcomm. Build datasets, train models, and optimize for any edge device from MCUs to gateways. Partners with NVIDIA, AWS, Nordic, STMicroelectronics, Arduino. Used by Chamberlain, Polar, Ultrahuman. Ultimate Guide to Edge AI available.
Pricing: Free tier. Enterprise: Custom pricing.
On-camera face recognition for resorts, casinos, venues, and smart home. Privacy-safe edge inference so video stays on the device; Orchestrait for supervision and on-prem deployment.
Pricing: Enterprise and OEM programs; contact Xailient.
Edge AI platform now part of Infineon's DEEPCRAFT suite. Features DEEPCRAFT Studio for end-to-end ML model development, Model Converter, and Ready Models. Supports tinyML, computer vision with YOLO, and PSoC Edge microcontrollers. Newsletter and Startup Challenge available.
Pricing: Contact for pricing. Enterprise solutions available.
Analog compute-in-memory edge AI accelerators for high-efficiency neural inference. Targets vision and always-on workloads where TOPS-per-watt dominates discrete GPUs.
Pricing: Hardware and SDKs; see mythic.ai.
Ultra-low-power edge AI chip solutions. Features neural decision processors for audio and sensor applications. Includes development tools and reference designs. Specializes in efficient AI processing for battery-powered devices.
Pricing: Contact for pricing. Hardware and development kit pricing varies.
TDK edge AI platform for industrial machine health (formerly Qeexo). Features edgeRX predictive maintenance, on-device sensor analytics, and real-time decisions for manufacturing, energy, and buildings. Privacy-first edge deployments; AWS partnership.
Pricing: Enterprise pricing. Contact for demo.
Nordic's TinyML stack from the Neuton acquisition. Auto-grown models typically under 5KB for nRF54 SoCs, plus Edge AI Lab and Axon NPU for heavier on-device workloads.
Pricing: Free via Nordic Edge AI Lab for Nordic customers.
LEIP software factory for deploying and updating edge AI models on existing hardware. Used for disconnected and mission-critical fleets, including defense and industrial sites.
Pricing: Enterprise licensing; contact Latent AI.
On-device retail AI for product recognition and loss prevention on existing scales, scanners, and self-checkout. Runs in-store without a cloud round-trip; live in 2,000+ grocery stores.
Pricing: Enterprise pilots and deployments; contact Edgify.
NVIDIA edge AI computing platform spanning Jetson Orin Nano, Orin NX, AGX Orin, and Thor for robotics, vision, and on-device generative AI. JetPack SDK bundles CUDA, TensorRT, and agent-ready tooling for deployment from compact modules to high-TOPS autonomous systems.
Pricing: Dev kits from roughly $249; modules and production systems vary by SKU.
Hybrid router that sends AI requests to local models or cloud APIs based on cost, privacy, and complexity. Works with OpenAI, Anthropic, Gemini, and local runtimes such as Ollama.
Pricing: Free trial; subscription for production-see oblix.ai.
High-performance edge AI accelerator featuring in-memory computing architecture delivering up to 214 TOPS of throughput with 15 TOPS/W power efficiency. The Metis AIPU (AI Processing Unit) enables server-class AI performance directly at the edge for computer vision applications. Can process multiple high-resolution video streams simultaneously for applications like surveillance, retail analytics, and industrial quality control while maintaining low power consumption. Available in PCIe card and M.2 module form factors.
Pricing: Developer kits available. Enterprise pricing for production deployments.
Akida neuromorphic processors for ultra-low-power event-based edge AI. On-device learning and inference for sensors, audio, and vision without heavy GPU power budgets.
Pricing: Chips and modules; see brainchip.com.
Google's edge AI platform for on-device ML. Hardware (Edge TPU, Dev Board, USB/M.2 accelerators) plus compiler tooling to deploy PyTorch, JAX, and LiteRT models locally.
Pricing: Hardware via Coral and partners; software tools free-see developers.google.com/coral.
Edge AI platform for building ML on resource-constrained IoT devices. Analytics Toolkit with Data Studio and AutoML Analytics Studio; generates ultra-compact inference code for microcontrollers (models as small as 5KB). Open-source Piccolo AI with no-code GUI and Python SDK. Supports Arm Cortex-M, RISC-V. For predictive maintenance, anomaly detection, industrial IoT. Partnership with onsemi.
Pricing: Toolkit and services. Contact for pricing.
Edge AI NPU and SoC vendor for on-device inference. Reconfigurable NPUs (KL1140 for edge Mamba/LLMs, KL730/KL720/KL630/KL530) for autonomous driving, security, robotics, IoT. Reconfigurable Artificial Neural Network (RANN) for audio and visual recognition. Supports TensorFlow, PyTorch, ONNX. San Diego-based.
Pricing: Hardware and SDK. Contact for pricing.
STMicroelectronics edge AI toolchain for STM32 MCUs and Neural-ART NPUs. Convert, optimize, and deploy neural networks on-device with ST Edge AI Suite and TinyML workflows.
Pricing: Free tools; silicon and kits via ST and distributors.
NPU processor IP and software toolchain provider for AI chips and SoCs. Enables hardware teams to build programmable edge AI silicon with efficient on-device inference.
Pricing: IP licensing and enterprise agreements; contact for terms.
Edge AI inference accelerators and software for vision-heavy embedded systems. M.2 and PCIe form factors support multi-stream video analytics in cameras, NVRs, and industrial compute nodes with low-latency on-device inference.
Pricing: Developer kits and enterprise programs; contact.
Fabless edge AI processor company. Energy-efficient accelerators for real-time inference on devices and gateways, with compiler and runtime support for vision, sensor fusion, and low-power generative workloads at the edge.
Pricing: Commercial and enterprise pricing; contact for hardware and programs.
Machine learning system-on-chip (MLSoC) platforms for edge and embedded AI. Targets robotics, industrial vision, drones, and smart cameras with integrated pipelines for CNNs, transformers, and multimodal models on-device.
Pricing: Dev kits and modules priced on site; enterprise programs on request.
Tiny AI software for intelligent cameras and ultra-constrained MCUs. People, vehicle, face, and motion analytics designed for low memory, coin-cell or solar power, and mass-market IoT silicon.
Pricing: Commercial licensing; contact for integration and volume.
Edge AI platform focused on incremental learning and optimization on connected devices. Industrial IoT use cases such as condition monitoring, predictive maintenance, and per-device model adaptation without shipping raw data to the cloud.
Pricing: Enterprise and OEM licensing; see site for evaluation.
Ultra-low-power edge AI SoCs for battery devices. Apollo and Atomiq families run always-on sensing and inference for wearables, healthcare, and industrial IoT. SPOT platform emphasizes sub-threshold power efficiency for long-life edge deployments.
Pricing: Hardware and SDK tiers; contact for enterprise.
Edge AI compute platform with GSP architecture and software stack for smart vision and industrial deployments. Delivers real-time inference on cameras and edge servers with power-efficient processing for enterprise edge workloads.
Pricing: Enterprise and platform pricing; contact.
South Korean edge AI semiconductor company shipping the DX-M1, DX-V1, DX-V3, and DX-H1 NPUs (up to 25 TOPS) for on-device generative AI, robotics, smart factories, and physical AI. Delivers up to 10x better power efficiency than GPU-based edge boards with LLM, vision, and multimodal model support. DXNN software stack integrates PyTorch, ONNX, and Hugging Face. Partners include Hyundai, Nvidia (Metropolis), and Amkor; over $80M raised.
Pricing: Modules and PCIe cards through partners; enterprise licensing on request.
Qualcomm on-device AI hub for deploying optimized models across Snapdragon and IoT platforms. Hosts ready models, conversion and profiling tools, and workflows for vision, audio, and generative inference on phones, gateways, and edge devices.
Pricing: Developer access free; commercial terms via Qualcomm programs.
Edge AI vision SoCs for cameras, automotive, and robotics. CV-series processors combine ISP and neural acceleration for low-power multi-camera perception, security, and autonomous systems at the edge.
Pricing: Chip and module programs; contact for OEM pricing.
Google's high-performance on-device ML framework (successor to TensorFlow Lite). Runs optimized models on mobile, microcontrollers, and edge hardware with conversion, delegation, and tooling for production edge inference.
Pricing: Free and open source.
PyTorch-native runtime for deploying models on mobile and edge devices. Exports programs from PyTorch, runs on-device with portable backends, and targets phones, embedded Linux, and microcontroller-class hardware.
Pricing: Free and open source.
Arm Ethos-U NPUs for microcontrollers and edge SoCs, including Ethos-U85 with transformer support up to ~4 TOPS. IP licensed into MCU and IoT silicon for efficient on-device neural inference at milliwatt-to-watt scale.
Pricing: IP licensing via Arm partners; see arm.com Ethos pages.
NVIDIA managed platform for deploying and operating AI applications across Jetson and GPU edge fleets. Centralized provisioning, monitoring, and updates for distributed edge sites.
Pricing: NVIDIA subscription; see nvidia.com Fleet Command.
Renesas tools for TinyML and sensor AI on MCUs and MPUs. Reality AI Studio and related software for building on-device analytics in industrial and consumer IoT.
Pricing: Tools via Renesas; see product pages.
Licensable DSP and NPU IP for edge AI in sensors, audio, and vision SoCs. Neural network and signal-processing cores used across mobile, automotive, and IoT silicon vendors.
Pricing: IP licensing; see ceva-ip.com.
AI depth cameras and edge vision systems (OAK) for robotics and industrial perception. On-device neural inference with spatial AI pipelines for autonomous machines.
Pricing: Cameras and kits on luxonis.com; volume OEM pricing available.
Industrial-grade edge AI platform for robotics, medical, and factory systems. Combines enterprise GPU compute, safety, and long-lifecycle support for on-device generative and sensor AI.
Pricing: Platform and enterprise software; see NVIDIA IGX product pages.
Ultra-low-power AI sensing processors (WiseEye2) for always-on vision on battery devices. Cortex-M55 + Ethos-U NPU class silicon for cameras, wearables, and IoT endpoints.
Pricing: Silicon and modules via Himax and partners; contact for OEM.
Analog machine-learning chips for always-on edge sensing. Runs tiny ML in the analog domain so voice, vibration, and anomaly detection wake digital systems only when needed - ultra-low-power edge AI silicon.
Pricing: Chip and module pricing via Aspinity; contact for OEM.
NVIDIA SDK for multi-stream edge vision AI: decode, infer, track, and visualize camera pipelines on Jetson and dGPU. Used for smart cities, retail analytics, and industrial vision at the edge.
Pricing: Free SDK; hardware and enterprise support via NVIDIA.
Python-programmable machine-vision cameras for microcontrollers, built for TinyML and on-device AI models. Low-power OpenMV Cams for robotics, sensing, and embedded vision prototypes through production.
Pricing: Hardware from ~$65+; open firmware. See OpenMV shop.
Edge AI accelerators for high-throughput on-device vision inference. Hailo-8 and modules deliver strong TOPS-per-watt for cameras, gateways, PCs, and industrial systems without cloud round-trips.
Pricing: Chip and module pricing via Hailo and partners; contact Hailo.