ThingsBoard vs AWS IoT Core & Azure IoT Hub

ThingsBoard is a complete IoT platform; AWS IoT Core and Azure IoT Hub are primarily a connectivity layer. That is the heart of the comparison. ThingsBoard ships device management, a rule engine and ready-made dashboards in one product and runs self-hosted on your infrastructure for a flat server bill. AWS IoT Core and Azure IoT Hub get devices reliably into the cloud but bill per message or per unit, and visualization, analytics and storage are assembled from extra services. This article compares features, current 2026 pricing with a worked example, data sovereignty, lock-in and operations, and says when each path fits.

Features: platform vs. pure connectivity

The biggest difference is scope. ThingsBoard covers the whole application layer: device management via device profiles, native transports (MQTT, HTTP, CoAP, LwM2M, SNMP), a visual rule engine (rule chains), freely buildable dashboards, alarms, multi-tenancy and RBAC. What ThingsBoard is and does is laid out in What is ThingsBoard?. The current release is ThingsBoard 4.3 (January 2026); the Community Edition is Apache 2.0.

AWS IoT Core and Azure IoT Hub are scoped more tightly. Their core job is connectivity: a scalable MQTT/HTTP endpoint, device authentication, a device shadow (Device Shadow on AWS, Device Twin on Azure) and routing logic (Rules Engine on AWS, Message Routing on Azure) that forwards messages to downstream services. Device management in the sense of provisioning, jobs and updates is a separate building block (AWS IoT Device Management, Azure Device Provisioning Service). Dashboards, analytics and long-term storage are not included: on AWS you combine Timestream, Lambda, S3 and QuickSight; on Azure, Stream Analytics, Functions, Storage and Power BI. That is powerful and deeply integrated, but each block has its own billing, its own configuration and its own lock-in.

In practice: what ThingsBoard delivers as one platform becomes, on a hyperscaler, an architecture of five to eight services. For pure connectivity inside an existing cloud stack, that is a good fit. For an IoT solution with dashboards and business logic, ThingsBoard is the shorter path.

Pricing: flat server bill vs. per message and per unit

This is where the models diverge fundamentally.

AWS IoT Core bills per component, with no minimum fee (US East N. Virginia, as of aws.amazon.com/iot-core/pricing, June 2026):

• Connectivity: USD 0.08 per 1M connection minutes.
• Messaging: USD 1.00 per 1M messages (first 1B/month, then tiered down to about USD 0.70), metered in 5 KB blocks (an 8 KB message counts twice).
• Device Shadow and Registry: USD 1.25 per 1M operations, metered in 1 KB blocks.
• Rules Engine: USD 0.15 per 1M rules triggered plus USD 0.15 per 1M actions.

Azure IoT Hub bills per unit and tier (as of azure.microsoft.com/pricing/details/iot-hub, June 2026):

• Free F1: 8,000 messages/day, metered in 0.5 KB blocks.
• Basic B1/B2/B3: about USD 10 / 50 / 500 per unit/month, but without cloud-to-device, device twins and device management.
• Standard S1/S2/S3: about USD 25 / 250 / 2,500 per unit/month for 400,000 / 6M / 300M messages/day, metered in 4 KB blocks.

ThingsBoard has neither a per-message nor a per-device fee. The Community Edition (Apache 2.0) is license-free, including for commercial use. You pay for server, storage and operations, a flat bill that does not grow with every message. An optional Professional Edition is available as a perpetual license from USD 4,999 one-off (as of thingsboard.io/pricing, June 2026). The full breakdown is in ThingsBoard pricing.

Worked example: 10,000 devices, one message per minute

A rough calculation makes the model visible. 10,000 devices each sending one small message (under 4 KB) per minute generate about 432M messages/month.

• AWS IoT Core: messaging 432M x USD 1.00/M = USD 432, plus connectivity (10,000 devices always connected = 432M minutes x USD 0.08/M = about USD 35), together around USD 467/month, before the Rules Engine and Device Shadow are added. If one rule with one action fires per message, add about USD 130/month.
• Azure IoT Hub: 432M/month is about 14.4M messages/day. One S2 unit holds 6M/day, so three units = 3 x USD 250 = around USD 750/month.
• Self-hosted ThingsBoard: this load (about 167 messages/second) is handled by a solid server (for example 8 vCPU, 32 GB RAM on Proxmox or Hetzner) for around EUR 100 to 150/month plus operations, independent of message volume. The Community Edition is license-free.

The decisive point is scaling: double the fleet and the cloud bill doubles. The server bill only steps up when the hardware hits its limit, in steps rather than linearly. This linearity is the real cost driver of hyperscaler IoT and is framed more broadly in self-hosted IoT vs. cloud.

Data sovereignty, GDPR and the EU

With AWS IoT Core and Azure IoT Hub, telemetry, master data and often control commands run through a US hyperscaler. Even in an EU region, US providers fall under the CLOUD Act, which can grant authorities access regardless of storage location. For personal or business-critical machine data this is a real compliance issue.

Self-hosted ThingsBoard flips it: the platform runs on your infrastructure inside the EU, with no third-country transfer and no shared responsibility with a hyperscaler. You decide on encryption, retention policies, deletion concepts and access rights yourself. For many European industrial and utility customers this is the deciding point, long before it comes to cents per message. The EU Data Act (Regulation (EU) 2023/2854), applicable since 12 September 2025, reinforces this direction: a right to data access and, from 12 January 2027, a ban on cloud switching fees.

Lock-in and operations

AWS IoT Core and Azure IoT Hub bind you on several levels: proprietary APIs and SDKs, vendor-specific services around them (Lambda/Functions, Timestream/Stream Analytics, QuickSight/Power BI) and the sheer migration effort once thousands of devices have to move. The exit is often more expensive than the entry. ThingsBoard relies on open protocols (MQTT, HTTP, CoAP), an open database (PostgreSQL, optionally TimescaleDB) and full access to the source code, which is portability by design.

To be honest: self-hosting has an operational price. Updates, backups, monitoring, security patches and database scaling all need an owner. But that effort is predictable and bounded, while the cloud bill keeps running with every new device. In practice many customers run operations as a managed service and still keep data sovereignty, open interfaces and the source code.

When self-hosted ThingsBoard, when a hyperscaler?

As a rule of thumb: for a quick test the cloud is convenient. For anything that grows, runs permanently or is regulatorily sensitive, a sovereign, self-hosted platform is the more sustainable choice. A broader view of the open-source options is in the open-source IoT platforms comparison.

How we work at WZ-IT

We plan, build and operate sovereign, self-hosted IoT platforms with ThingsBoard on your infrastructure (Proxmox, Hetzner, on-prem), GDPR compliant from Germany and without cloud lock-in. We model the real cost honestly against your AWS or Azure alternative and, on request, take over the full operation. More on our ThingsBoard page, in the IoT hub and specifically on IoT platform development. To book a no-obligation intro call, pick a time slot.