Every hardware company evaluating IoT platforms eventually asks: should we just build on AWS IoT Core? Amazon offers a massively scalable MQTT broker, device shadows, rule engines, and integrations with the entire AWS ecosystem. It sounds like everything you need. And for some companies, it is. But for most hardware companies and SMBs, building on AWS IoT Core is like buying a pile of lumber when you need a house.
What AWS IoT Core gives you
AWS IoT Core is an MQTT message broker with device authentication, message routing, and integration with other AWS services. Here is what you get out of the box:
A managed MQTT broker that scales to millions of concurrent connections. Device registry for tracking connected devices. Device shadows (digital twins) that store the last reported state. A rules engine that routes messages to other AWS services (DynamoDB, S3, Lambda, Kinesis). X.509 certificate-based device authentication. Pay-per-message pricing (USD 1 per million messages for the first billion).
What AWS IoT Core does not give you
This is the list that catches most hardware companies by surprise.
No dashboards. AWS IoT Core does not include any visualization layer. You need to build dashboards using QuickSight, Grafana, or a custom React application. No time-series database. Messages are routed but not stored in a query-friendly format. You need to set up Timestream, InfluxDB, or DynamoDB with appropriate schemas and retention policies. No alerting system. The rules engine can trigger Lambda functions, but building a complete alert pipeline (thresholds, escalation, delivery via SMS/WhatsApp/email) is a separate engineering project.
No firmware SDK. AWS provides basic MQTT client libraries, but not firmware SDKs that handle reconnection logic, offline buffering, OTA updates, and device provisioning workflows. No OTA update system. AWS IoT Jobs can push firmware updates, but you need to build the entire pipeline: firmware signing, staged rollouts, rollback logic, and version management. No predictive maintenance or ML inference. AWS has SageMaker for ML, but integrating it with IoT data requires building a data pipeline, training models, deploying endpoints, and connecting the results back to your devices.
No edge AI. For on-device inference, you would need AWS Greengrass, which is a separate product with its own learning curve, deployment model, and pricing. No compliance templates. There are no pre-built reports for BEE, CPCB, FAME II, or any India-specific regulatory frameworks. No WhatsApp or SMS alerts. Amazon SNS can send SMS, but WhatsApp Business API integration is a separate project.
The hidden costs of building on AWS IoT Core
The per-message pricing of AWS IoT Core looks cheap. USD 1 per million messages. For a fleet of 1,000 devices sending data every 30 seconds, that is about USD 87 per month just for message brokering. But message brokering is perhaps 10 percent of your total IoT infrastructure cost. Here is what the full picture looks like.
Time-series storage: Timestream costs USD 0.50 per GB ingested and USD 0.01 per GB-hour for storage. For 1,000 devices with moderate telemetry, expect USD 200-500 per month. Dashboard hosting: A managed Grafana instance on AWS costs USD 9-24 per editor per month plus data source fees. Lambda compute for rules processing: variable, but typically USD 50-200 per month for a moderate workload. S3 for firmware storage and logs: USD 20-50 per month. Certificate management and IoT policy administration: engineer time, not AWS costs.
The real hidden cost is engineering time. A senior engineer in India costs 15-25 lakh per year. Building the dashboard layer, alert system, OTA pipeline, and device management console takes a minimum of 2-3 engineers working for 3-6 months. That is 15-40 lakh in salary costs before your first device ships. Then you need those same engineers to maintain and scale the system indefinitely.
When managed platforms like Akraniq make more sense
Managed IoT platforms bundle everything AWS IoT Core leaves out: dashboards, time-series storage, alerting, firmware SDKs, OTA updates, and operational monitoring. The trade-off is less customization flexibility in exchange for dramatically faster time to market.
Hardware companies that build sensors, EVs, or industrial equipment should almost always use a managed platform. Your competitive advantage is in your hardware and your domain expertise, not in your ability to configure VPC peering between IoT Core and Timestream. A managed platform like Akraniq gets you from first prototype to production fleet in 3-4 weeks, not 3-6 months.
SMBs and mid-size manufacturers who need IoT for their own operations (energy monitoring, predictive maintenance, fleet tracking) should use a managed platform. You do not have a cloud engineering team, and you should not build one just to monitor your machines.
When building on AWS IoT Core makes sense
There are legitimate reasons to build on AWS IoT Core directly.
You are building an IoT platform as your product. If you are a software company building a multi-tenant IoT platform to sell to other businesses, you need the flexibility of raw AWS services. You have a dedicated cloud engineering team. If you already employ DevOps engineers, backend developers, and data engineers who are comfortable with AWS, the incremental cost of building on IoT Core is lower. You have strict data residency requirements. Some enterprises and government contracts require data to stay in specific AWS regions with specific compliance certifications. You need deep integration with other AWS services. If your application already runs on AWS and you need tight integration with services like SageMaker, Kinesis, or Redshift, building natively on AWS makes architectural sense.
Cost comparison: AWS IoT Core DIY vs Akraniq
For a typical deployment of 500 devices (EV fleet or factory monitoring), here is a realistic 12-month cost comparison in INR.
AWS IoT Core DIY: AWS services (IoT Core, Timestream, Grafana, Lambda, S3, SNS) at approximately 3-5 lakh per year. Engineering team (2 backend engineers for 6 months build plus ongoing maintenance) at 15-25 lakh per year. Firmware development (if not already done) at 5-10 lakh. Total first year: 23-40 lakh. Ongoing: 8-15 lakh per year for infrastructure plus at least one dedicated engineer.
Managed platform (Akraniq): Platform license plus deployment services at 8-15 lakh per year including dashboards, alerts, OTA, edge AI, and support. Firmware SDK integration at 2-3 lakh one-time. Total first year: 10-18 lakh. Ongoing: 8-15 lakh per year with no dedicated engineering staff required.
The managed platform costs 40-55 percent less in the first year and requires zero cloud engineering headcount. The gap widens in year two because the DIY approach still needs an engineer on payroll.
Five questions to ask before choosing
Before deciding between AWS IoT Core and a managed platform, answer these honestly.
One: Do you have backend engineers who are comfortable with AWS IoT, MQTT, and time-series databases? If no, a managed platform saves you 6 months of hiring and learning. Two: Is your competitive advantage in cloud infrastructure or in your hardware and domain expertise? If hardware, stop building cloud. Three: Do you need to be live in weeks or can you wait months? Managed platforms deploy in 2-4 weeks; DIY takes 3-6 months minimum. Four: Will you maintain this system for years? IoT infrastructure is not build-and-forget. Security patches, scaling, and feature requests require ongoing engineering. Five: What is your total budget including engineering salaries? The cheapest AWS bill means nothing if you are spending 20 lakh per year on engineers to manage it.
The bottom line
AWS IoT Core is an excellent building block for companies that want to build IoT platforms. It is a poor choice for companies that want to use IoT to make their products smarter. If you are a hardware company, factory owner, or fleet operator, you need a managed platform that handles the cloud so you can focus on what you actually sell. That is the build-vs-buy decision in one sentence.