IoT Platforms for Indian Manufacturing: Build vs Buy
Indian manufacturers are connecting their factories. Sensor data flows to cloud. AI surfaces insights. The choice of IoT platform shapes what's possible. Here's the framework.
Key takeaways
- Buy: AWS IoT, Azure IoT, GCP IoT for cloud-native; PTC ThingWorx, Siemens MindSphere for industrial-specific.
- Build: only for highly custom needs or at scale where vendor pricing hurts.
- Most manufacturers should buy and customize on top.
- The hidden cost is integration with PLCs, SCADA, MES, and data normalization.
What IoT platforms do
Device management
Provision, configure, update thousands of devices remotely.
Data ingestion
High-volume time-series ingestion from sensors.
Edge compute
Run logic at the edge (factory floor) for latency-sensitive operations.
Cloud analytics
Aggregate and analyze data centrally.
Visualization
Dashboards, alerts.
Integration
Connect to ERP, MES, SCADA, business systems.
Cloud-native options
AWS IoT
Mature, broad services, ecosystem. Good for AWS-aligned shops.
Azure IoT
Strong industrial features, IoT Central for low-code.
GCP IoT
Strong analytics (BigQuery), Cloud IoT Core was sunset; alternatives via Pub/Sub.
Industrial-specific options
PTC ThingWorx
Strong manufacturing focus, OPC-UA integration, digital twin.
Siemens MindSphere
Deep manufacturing, integrates with Siemens equipment.
Litmus, Cisco Edge Intelligence
Niche players for specific industrial scenarios.
When to build
- Scale where vendor pricing hurts (millions of devices)
- Highly custom protocols
- Specific edge compute requirements
- Strong in-house IoT team
Otherwise, buy and customize.
What to plan for
PLC/SCADA integration
Most factory data comes from PLCs and SCADA systems. OPC-UA is the integration standard. Plan for adapters.
Data normalization
Sensor data from different vendors arrives in different formats. Normalization layer is real work.
Time-series storage at scale
Millions of data points per minute. Use TimescaleDB, InfluxDB, or cloud time-series.
Connectivity
Factory networks vary. 4G/5G, Wi-Fi, wired, mesh. Plan for resilience.
Edge vs cloud decision per workload
Latency-sensitive: edge. Aggregate analytics: cloud. Hybrid is real.
What works
Start with one production line
Don't try to instrument everything. One line, real value, then scale.
Define the use case first
Predictive maintenance, OEE tracking, quality monitoring, different use cases shape platform choice.
Plan for OT/IT separation
Industrial control networks separate from corporate networks. Security boundaries.
Common pitfalls
Platform shopping without use case. "We need IoT" doesn't tell you what to buy.
Underestimating integration. PLC/SCADA work is months.
Cloud-only when latency matters. Some workloads need edge.
No data model. Sensor data without taxonomy is data debt.
What we recommend
Pick a platform aligned with your cloud preference. Start with one pilot line. Prove value. Scale. Add edge components for specific latency-critical workloads.
FAQs
Open source IoT platforms? Eclipse IoT, ThingsBoard. Viable for some.
5G/private cellular? Emerging; not yet default.
MQTT vs OPC-UA? Both, MQTT for cloud-bound, OPC-UA for industrial.