The choice of IoT cloud platform is one of those decisions that seems tactical—“we just need somewhere to send data”—but turns out to be deeply strategic. Once 10,000 devices are provisioned against a platform’s certificate authority, once your backend code is wired to its device shadow API, once your team has built operational muscle around its monitoring tooling, migrating is expensive. The major platforms—AWS IoT Core, Azure IoT Hub, Google Cloud IoT, and Particle—have converged on a common feature set (device registry, MQTT ingestion, security, telemetry routing) but diverge sharply in pricing model, ecosystem integration, developer experience, and the operational sophistication required to run them well. This comparison gives you the context to make the right call for your product.
AWS IoT Core: The Most Powerful, Most Complex
AWS IoT Core is the default choice for teams already building on AWS, and for good reason. It’s the most feature-rich IoT platform available, with a massive ecosystem of integrated services: IoT Events for complex event detection, IoT Greengrass for edge computing, IoT Device Defender for security auditing, IoT Fleet Hub for operational dashboards, and IoT TwinMaker for digital twins. No other platform offers this breadth out of the box.
At the core, IoT Core provides:
- MQTT, HTTPS, and WebSocket endpoints for device communication
- Device Shadow for storing and syncing device state even when offline
- Rules Engine for routing messages to Lambda, Kinesis, DynamoDB, S3, SNS, SQS, and 15+ other AWS services with SQL-like rule expressions
- Just-In-Time Provisioning (JITP) and Fleet Provisioning for automated certificate-based device onboarding at scale
- AWS IoT Device Management for grouping, OTA updates, and remote operations
The trade-off is complexity and cost. AWS IoT Core’s pricing is message-based: you pay per million messages (a “message” is a 512-byte unit, so 1 KB messages = 2 message units), per minute of device connectivity, and per rule execution. At 100,000 devices sending one message per minute, costs can reach $500–2,000/month before you account for the downstream services (Kinesis, Lambda, DynamoDB) that process and store the data.
AWS IoT Core is the right choice when you need the full AWS service catalog and are prepared to invest in learning its many abstractions. AWS IoT Core documentation is comprehensive but dense.
Azure IoT Hub: Enterprise-Grade with Tight Microsoft Integration
Azure IoT Hub is Microsoft’s flagship IoT service and a strong competitor to AWS IoT Core, particularly for enterprises already standardized on Azure Active Directory, Azure Monitor, and the broader Microsoft ecosystem. It’s the platform of choice for manufacturing, energy, and healthcare use cases where enterprise identity management and compliance certifications (HIPAA, SOC 2, FedRAMP) matter.
Key capabilities:
- Device-to-Cloud (D2C) and Cloud-to-Device (C2D) messaging with configurable message retention
- Device Twin (analogous to AWS Device Shadow) for desired/reported state synchronization
- Azure IoT Edge for containerized edge workloads on gateway devices, with native integration to Azure ML for edge inference
- Azure DPS (Device Provisioning Service) for zero-touch, at-scale device enrollment using TPM, X.509, or symmetric keys
- IoT Hub routing to Event Hubs, Azure Blob Storage, Service Bus, and Cosmos DB
Azure IoT Hub’s pricing tiers (Free, S1, S2, S3) are capacity-based rather than strictly pay-per-message at the lower tiers, which can be more predictable for mid-scale deployments. The S1 tier supports 400,000 messages/day per unit at ~$25/unit/month—straightforward for planning.
The Azure IoT Hub documentation and the broader Azure IoT Central (a higher-level SaaS IoT application platform built on Hub) give teams options from managed to fully custom.
Google Cloud IoT Core: Deprecated but Its Successor Matters
It’s important to note: Google Cloud IoT Core was deprecated in August 2023. If you’re evaluating GCP for IoT, the current architecture is to use Pub/Sub for MQTT or HTTP ingestion (via the MQTT Bridge or direct HTTP), combined with Dataflow (managed Apache Beam) for stream processing, BigQuery for analytics, and Cloud Bigtable or Firestore for device state.
This is a more DIY approach than AWS or Azure, but the individual GCP services are excellent. Google Cloud Pub/Sub is highly scalable for IoT ingestion. BigQuery is arguably the best analytical query engine available, making GCP attractive for IoT platforms with heavy analytics requirements. The lack of a purpose-built IoT service means more integration work but more flexibility.
GCP is worth considering if:
- Your team already uses GCP and BigQuery for analytics
- You need BigQuery’s analytical capabilities for your device data
- You’re building a platform where Dataflow’s stream processing power is relevant
Particle: Developer-Friendly, Hardware-to-Cloud
Particle occupies a different position from the hyperscalers: it’s a vertically integrated IoT platform that provides its own cellular and Wi-Fi hardware modules, a cloud backend, a device OS, and a developer toolchain in a single integrated product. If AWS IoT Core is a set of infrastructure building blocks, Particle is a product you can have running in an afternoon.
Particle’s strengths:
- Integrated hardware: The Photon 2 (Wi-Fi), Boron (LTE-M/NB-IoT), and Tracker SoM (LTE + GNSS) modules ship with pre-provisioned credentials and a working cellular data plan
- Particle OS: A real-time operating system with a familiar Arduino-like API, OTA firmware updates, and a cloud-connected RTOS in one package
- Particle Cloud: Device management, event publishing, function calling, variable reading, webhook integration, and fleet-level OTA all included
- Console & APIs: A polished developer console and REST API make it easy to build a product backend without deep cloud infrastructure expertise
The trade-off is lock-in and scale cost. Particle’s per-device pricing (~$2.99/device/month for cellular) is cost-effective for small fleets but becomes expensive at tens of thousands of devices. And because Particle’s cloud is tightly coupled to its hardware, migrating to AWS or Azure later requires hardware changes.
Particle’s documentation is excellent, and the platform is a strong choice for IoT product teams that want to move fast without hiring cloud infrastructure specialists.
Feature-by-Feature Comparison
| Feature | AWS IoT Core | Azure IoT Hub | GCP (DIY) | Particle |
|---|---|---|---|---|
| MQTT | Yes | Yes | Via Bridge | Yes |
| Device Shadow/Twin | Device Shadow | Device Twin | DIY (Firestore) | Variable/Function |
| Edge Computing | Greengrass | IoT Edge | Edge TPU | No |
| OTA Updates | IoT Jobs | ADU (Azure) | DIY | Built-in |
| Fleet Provisioning | JITP/Fleet Provisioning | DPS | DIY | Automatic |
| Analytics | Kinesis + QuickSight | Stream Analytics + Power BI | BigQuery (excellent) | Webhook → external |
| Security | X.509 + policy engine | X.509 + AAD | IAM + X.509 | Pre-provisioned certs |
| Pricing model | Per message + connection minute | Per unit/tier | Per service | Per device/month |
| Complexity | High | High | High | Low |
| Best for | Large-scale, AWS-native | Enterprise, Azure-native | Analytics-heavy | Rapid product dev |
Hybrid and Multi-Cloud Approaches
Many production IoT platforms don’t rely on a single cloud vendor exclusively. Common patterns include:
- Particle for device management, AWS S3/Athena for data storage and analytics: Use Particle’s ease of use for device operations while leveraging AWS for scalable data storage
- Azure IoT Edge on-premises + Azure IoT Hub cloud: Edge processing for latency-sensitive operations, cloud for persistence and analytics
- Self-hosted MQTT (EMQX/HiveMQ) + Kafka + InfluxDB: Full control over the stack, portable across clouds, no per-message fees—but requires significant operational expertise
The decision between managed platform and self-hosted stack usually comes down to team size and expertise. A 3-person startup shipping a connected product benefits enormously from Particle’s integration. A 30-person platform team with cloud engineers can run a self-hosted stack that costs 60% less per device at scale.
For deeper understanding of the data pipeline within these platforms, read our Managing IoT Data at Scale article. Our team at UABit helps clients evaluate and implement IoT cloud platforms tailored to their product’s requirements—visit our IoT Connectivity Integration services page to learn more.
Conclusion
AWS IoT Core wins on breadth and ecosystem for teams already on AWS. Azure IoT Hub wins on enterprise identity integration and compliance for Microsoft-shop organizations. Google Cloud wins on analytics if you’re already using BigQuery and willing to assemble your own IoT stack. Particle wins on developer experience and time-to-market for hardware-forward teams. The right answer depends on your team’s existing cloud skills, your device fleet’s scale, your analytics requirements, and how much operational complexity you can absorb. Lock-in is real on all platforms—evaluate the migration path before committing, and prototype under realistic load conditions before you provision 10,000 devices.
IoT & AIoT Weekly
Get the best IoT development content delivered weekly. No noise, just signal.