LabHub

Blog

Industrial IoT (IIoT) 2026 Complete Guide - Siemens Insights Hub, AWS IoT, Azure Digital Twins, PTC ThingWorx, Litmus, Balena, Particle, GE Vernova Deep Dive

2026-05-16 · 23 min read · #industrial-iot#iiot#siemens#aws-iot#azure-digital-twins#ptc-thingworx#litmus#balena#particle#opc-ua#mqtt#english

한국어English日本語

Introduction: May 2026, IIoT Has Entered the "OT/IT Convergence Era"

In the early 2020s, IIoT was a simple slogan: "send sensor data to the cloud." By May 2026 the story is far more nuanced. The boundary between OT (Operational Technology, factory automation) and IT (Information Technology, enterprise IT) has dissolved, and predictive maintenance, digital twins, and edge AI inference are simultaneously standard requirements.

This article is not a marketing slide. It honestly maps the IIoT stack going into actual factories, power plants, and logistics warehouses as of May 2026 — how Siemens rebranded MindSphere as Insights Hub, where Predix went after GE's 2024 split into GE Aerospace / Vernova / HealthCare, how Azure IoT Operations and AWS IoT TwinMaker filled the gap left when Google Cloud IoT shut down in August 2023, and how far open source projects like EdgeX Foundry and Eclipse Hono have come.

IIoT Stack: Decomposed Into 7 Layers

The standard IIoT stack in 2026 splits into the following 7 layers.

  1. Field devices: sensors, PLCs, motors, drives
  2. Edge gateway: protocol conversion, local buffering, edge inference
  3. OT network: PROFINET, EtherCAT, EtherNet/IP, OPC UA, MQTT
  4. Cloud gateway: MQTT brokers, IoT Hub, message queues
  5. Data platform: time-series DB, data lake, digital twin
  6. Analytics + AI: predictive maintenance, quality inspection, optimization
  7. Apps + UI: SCADA, MES, ERP integration, AR work instructions

As inter-layer interfaces standardize, mixed-vendor stacks have become the norm. Combinations like Siemens PLC + AWS IoT Core + Azure Digital Twins + PTC ThingWorx apps are common in the wild.

Siemens Industrial Edge + Insights Hub: "MindSphere" Is Gone and Reorganized

Siemens rebranded its cloud IoT platform MindSphere as Insights Hub in 2023, and simultaneously split out the edge track as Industrial Edge and the AI track as Industrial AI Suite. As of May 2026, the lineup is:

A typical Insights Hub API call looks like this.

import requests

tenant = "mycompany"
token = "<oauth_token>"
asset_id = "asset-123"

resp = requests.get(
    f"https://gateway.eu1.mindsphere.io/api/iottimeseries/v3/timeseries/{asset_id}/Temperature",
    headers={"Authorization": f"Bearer {token}"},
    params={"from": "2026-05-15T00:00:00Z", "to": "2026-05-16T00:00:00Z"},
)
for point in resp.json():
    print(point["_time"], point["temperature"])

The strength of Industrial Edge is its deep integration with Siemens PLCs. Data from SIMATIC S7-1500/1200 and ET 200SP distributed IO is standardized via OPC UA and shipped to the cloud. The downside is vendor lock-in — Insights Hub forces its own time-series model on you.

GE Vernova / GE Digital Predix: The Landscape After the 2024 GE Split

GE officially split into GE Aerospace (aircraft engines), GE Vernova (power/energy), and GE HealthCare (medical) in April 2024. The IIoT assets mostly went to GE Vernova.

GE Vernova is now positioned as an IIoT vendor specialized in power plants and utilities. The era when Predix was advertised as the "Android of industry" is over. In general manufacturing, Siemens Insights Hub, PTC ThingWorx, and AWS/Azure took its seat.

PTC ThingWorx + Vuforia + Kepware: Industrial Data Platform + AR

PTC's IIoT portfolio rests on three pillars.

The strength of ThingWorx is that, paired with Kepware, it absorbs nearly every industrial protocol. Modbus, PROFINET, EtherNet/IP, S7, Allen-Bradley, Honeywell Experion, DNP3 — all unified to OPC UA from a single gateway.

PTC also pushes the digital thread narrative by combining with PTC FlexPLM (product lifecycle management). The picture is a single line of data from design through manufacturing through service.

Hitachi Lumada: Strong IIoT Player in Asia

Hitachi Lumada has a strong presence in Japan and the wider Asian IIoT market. As of May 2026 its components are:

Hitachi reinforced its digital engineering capability with the 2023 GlobalLogic acquisition, and is expanding grid-side IIoT around Hitachi Energy through 2025. Major customers include Nippon Steel, JFE, and Kawasaki Heavy Industries.

Schneider EcoStruxure: Unified Energy + Building + Industry Platform

Schneider Electric's EcoStruxure is an IIoT platform that layers three tiers (Connected Products, Edge Control, Apps/Analytics) on top of four domains (Building, Data Center, Industry, Power).

With the AVEVA acquisition, Schneider brought OSIsoft PI System (the de-facto time-series data standard) into its camp. PI System is the de facto standard in power plants, oil fields, and mines.

ABB Ability + Honeywell Forge + Rockwell FactoryTalk

Each of these three vendors put a cloud IIoT track on top of its existing DCS/SCADA line.

Rockwell strengthened the OT cybersecurity track via the 2023 Verve Industrial Protection acquisition, and extended into the AMR (autonomous mobile robot) space via the 2024 Clearpath Robotics acquisition.

Bosch IoT Hub + Bosch Phantom Edge: IIoT From an Automotive OEM Heritage

Bosch's IIoT story starts in automotive OEM parts supply chains.

Bosch open-sourced a large part of its IoT code into the Eclipse Foundation. This is the origin of Eclipse Hono (IoT messaging protocol adapter), Eclipse Ditto (digital twin), Eclipse Hawkbit (OTA update), and Eclipse Kura (Java-based edge gateway).

AWS IoT Core / SiteWise / TwinMaker / Greengrass: Hyperscaler Number One

AWS IoT is built from five services.

A SiteWise asset model definition looks like this.

{
  "assetModelName": "TurbineModel",
  "assetModelProperties": [
    {
      "name": "Temperature",
      "dataType": "DOUBLE",
      "type": {
        "measurement": {}
      }
    },
    {
      "name": "AvgTemperature15min",
      "dataType": "DOUBLE",
      "type": {
        "metric": {
          "expression": "avg(temp)",
          "variables": [
            {"name": "temp", "value": {"propertyId": "Temperature"}}
          ],
          "window": {"tumbling": {"interval": "15m"}}
        }
      }
    }
  ]
}

The strength of AWS IoT is integration with the wider AWS ecosystem. The flows SiteWise → S3 → Athena/QuickSight and Greengrass → SageMaker Edge Manager (phased out in 2024, replaced by Greengrass ML Component) just work end to end.

Azure IoT Hub / Digital Twins / IoT Operations: New K8s-Based Line in 2024

Azure IoT's core is:

The arrival of IoT Operations is a big shift. Kubernetes (managed via Azure Arc) is installed at the edge, and on top of it, an MQTT broker (EMQX-based), OPC UA Connector, Data Processor, and Akri (edge device discovery) are deployed as containers.

A DTDL digital twin definition looks like this.

{
  "@context": "dtmi:dtdl:context;3",
  "@id": "dtmi:com:example:Turbine;1",
  "@type": "Interface",
  "displayName": "Wind Turbine",
  "contents": [
    {
      "@type": "Property",
      "name": "model",
      "schema": "string"
    },
    {
      "@type": "Telemetry",
      "name": "rotationSpeed",
      "schema": "double"
    },
    {
      "@type": "Relationship",
      "name": "rel_has_blades",
      "target": "dtmi:com:example:Blade;1"
    }
  ]
}

Azure's differentiator is the DTDL standard. The spec influenced ISO/IEC 30173 and stands alongside AAS (Asset Administration Shell) as one of the digital twin standardization axes.

Google Cloud IoT Shut Down in August 2023: The Vacuum Since

Google announced the end of Google Cloud IoT Core in August 2022, and the service was fully terminated on August 16, 2023. Existing customers were guided to migrate to ClearBlade, EMQX Cloud, or HiveMQ Cloud.

As of May 2026, Google Cloud does not directly provide an IoT device gateway SaaS. Instead it recommends:

With Google exiting the IoT device gateway market, the AWS-vs-Azure duopoly has strengthened. That said, Google's BigQuery + Vertex AI track is still frequently used as an IIoT analytics backend.

EdgeX Foundry: LF Edge's Open Edge Platform

EdgeX Foundry is the open-source IoT/IIoT edge platform under LF Edge. Starting in 2017 from code Dell contributed, the stable line as of May 2026 is Odessa (3.1).

The strength of EdgeX is vendor-neutral open source. While AWS Greengrass and Azure IoT Edge bind you to a specific cloud, EdgeX can send data anywhere.

Bringing up EdgeX with Docker Compose goes like this.

git clone https://github.com/edgexfoundry/edgex-compose
cd edgex-compose/compose-builder
make run no-secty
curl http://localhost:59880/api/v3/ping
curl http://localhost:59881/api/v3/reading/all

Besides EdgeX, LF Edge also hosts Akraino (telco edge), Project EVE (edge virtualization), and Fledge (industrial data OS).

Eclipse Hono · Ditto · Hawkbit · Kura · Kapua

The Eclipse Foundation hosts the core IIoT infrastructure as open source.

Bosch IoT Suite is a commercial SaaS layered on Hono + Ditto + Hawkbit, and VMware Tanzu's IoT track is also built on the Eclipse stack.

Apache PLC4X · StreamPipes · OpenEMS: The Apache Foundation Industrial Track

The Apache Foundation also hosts growing IIoT projects.

PLC4X Modbus read example.

PlcConnectionManager mgr = PlcConnectionManager.getDefault();
try (PlcConnection conn = mgr.getConnection("modbus-tcp:tcp://192.168.1.10:502")) {
    PlcReadRequest req = conn.readRequestBuilder()
        .addItem("temperature", "holding-register:100:UINT")
        .addItem("pressure", "holding-register:101:UINT")
        .build();
    PlcReadResponse resp = req.execute().get();
    System.out.println("temp=" + resp.getInteger("temperature"));
    System.out.println("pressure=" + resp.getInteger("pressure"));
}

PLC4X is the friendliest entry point for developers unfamiliar with industrial protocols.

OPC UA: IEC 62541, the De Facto Standard of Industrial Automation

OPC UA (Open Platform Communications Unified Architecture) is an industrial automation communications protocol standardized as IEC 62541. As of May 2026, OPC UA has become the lingua franca of IIoT.

Its key characteristics are:

An OPC UA Python client (open62541 or asyncua) example.

import asyncio
from asyncua import Client

async def main():
    url = "opc.tcp://192.168.1.20:4840"
    async with Client(url=url) as client:
        root = client.get_root_node()
        temp_node = await client.nodes.objects.get_child(["2:Plant", "2:Reactor1", "2:Temperature"])
        value = await temp_node.read_value()
        print(f"temperature={value} C")

asyncio.run(main())

OPC UA Pub/Sub mode layers OPC UA messages on top of MQTT 5.0 and has become the primary field-to-cloud communication protocol.

MQTT 5.0 + Sparkplug B: The IIoT Message Layer

MQTT is the simplest option for an IoT message protocol. MQTT 5.0 (2019) added properties, reason codes, shared subscriptions, and request/response, and from there it became a serious option for industrial use too.

Sparkplug B is an Eclipse Foundation standard that defines industrial message semantics on top of MQTT.

Sparkplug B is supported by Inductive Automation Ignition, HiveMQ Sparkplug Edition, AWS IoT Sparkplug Adapter, and Cirrus Link IoT Bridge. The standard line as of May 2026 is Sparkplug B 3.0.

An MQTT 5.0 + Sparkplug B publish in Python (paho-mqtt with a simplified payload).

import paho.mqtt.client as mqtt
import json
from datetime import datetime

client = mqtt.Client(client_id="edge-01", protocol=mqtt.MQTTv5)
client.username_pw_set("plant-user", "secret")
client.connect("broker.factory.local", 1883)

payload = {
    "timestamp": int(datetime.now().timestamp() * 1000),
    "metrics": [
        {"name": "Temperature", "value": 75.3, "type": "Double"},
        {"name": "Pressure", "value": 1.04, "type": "Double"},
    ],
    "seq": 42,
}
client.publish("spBv1.0/Plant1/DDATA/EdgeNode01/Reactor1", json.dumps(payload), qos=1)

In real Sparkplug B you use Protobuf encoding, but the JSON above shows the semantic model in a simplified form.

Modbus / PROFINET / EtherCAT / EtherNet/IP / DDS / LwM2M

Before OPC UA and MQTT — and still today alongside them — there are the industrial protocols.

Gateways typically translate these to OPC UA or MQTT using products like Kepware, HMS Anybus, Moxa MGate, Red Lion, and WAGO PFC.

Embedded + Gateway Hardware: Simatic IPC, Advantech, Beckhoff

Hardware that goes into an industrial site is different from a regular PC. IP65/67 dust/water sealing, -40 to +70 C operation, vibration resistance, and 100,000-hour MTBF are baseline.

The 5G + industrial PC combination has entered factories in earnest since 2024. 5G NR-Reduced Capability (RedCap) and 5G Advanced (Release 18) are the new circuit options for industrial IoT.

Balena · Particle.io · Golioth · Memfault: Device-to-Cloud SaaS

Beyond enterprise IIoT, there is a SaaS layer that targets OEM device makers.

Creating a Balena fleet and adding a device looks like this.

balena login
balena fleet create my-fleet --type raspberrypi4-64
balena os download raspberrypi4-64 -v default -o resin.img
balena os configure resin.img --fleet my-fleet
balena push my-fleet

Memfault has become the de facto standard for firmware crash reporting. Zephyr, FreeRTOS, and Linux are all supported, and firmware updates plus boot logs plus metrics live in one view.

Litmus Edge · HiveMQ · EMQX: MQTT Brokers + Industrial Data OS

MQTT brokers are the heart of IIoT message infrastructure.

Deploying an EMQX cluster on Kubernetes goes like this.

apiVersion: apps.emqx.io/v2beta1
kind: EMQX
metadata:
  name: emqx-cluster
spec:
  image: emqx/emqx:5.7.0
  coreTemplate:
    spec:
      replicas: 3
      resources:
        limits:
          cpu: "2"
          memory: 4Gi
  replicantTemplate:
    spec:
      replicas: 2
  listenersServiceTemplate:
    spec:
      type: LoadBalancer

Litmus Edge has grown quickly by carving out the industrial data standardization niche. The company drew attention with Snowflake's failed acquisition attempt in 2024.

NVIDIA Jetson + Metropolis + Omniverse: Edge AI + Digital Twin

NVIDIA has a foot in both IIoT axes.

Omniverse + USD has emerged since 2024 as a candidate standard for industrial digital twins. The picture is to standardize models built in Siemens NX, PTC Creo, and Autodesk Inventor as USD and merge them in Omniverse.

Jetson edge inference is typically structured as NVIDIA TensorRT + DeepStream + Triton. Compile an ONNX model with TensorRT, then run vision inference in a DeepStream pipeline (GStreamer-based).

Intel OpenVINO + TinyML + Edge Impulse: Non-NVIDIA Edge AI

There is also a non-NVIDIA edge AI track.

OpenVINO often appears in Intel NUC + industrial vision inspection. GPU performance is weaker than Jetson, but the CPU + iGPU + Movidius VPU combination delivers low power and reasonable performance.

ISA-95 · IEC 62443 · NAMUR · OPC UA Companion Specs: Industrial Standards

IIoT sits on top of standards.

ISA-95 is cited almost always in ERP-MES-PLC integration projects. As of May 2026, the three axes AAS + DTDL + OPC UA are competing side by side as the digital twin standardization.

OT Cybersecurity: Claroty · Dragos · Nozomi · Armis · TXOne

As IIoT spreads, OT cybersecurity has become a market in its own right. As of May 2026 the top five are:

IEC 62443 is the consistency baseline for these tools. An assurance that layers SOC 2 + ISO 27001 with IEC 62443 is now the standard requirement in OT security RFPs.

The core of OT security is passive monitoring. Active scans can knock out PLCs, so the standard approach is to read traffic from a network mirror port and infer assets and vulnerabilities from there.

Korean IIoT: Posco, Hyundai, CJ Logistics, LG CNS, Samsung SDS

The Korean IIoT ecosystem is built around the following:

LS ELECTRIC (formerly LS Industrial Systems), with its LG Industrial Systems heritage in PLCs and HMIs, is also a notable axis of Korean IIoT. They own the XGB PLC series and their own SCADA (XGI Studio).

Japanese IIoT: Mitsubishi Connected Industries, Yokogawa, NTT Com

The Japanese IIoT ecosystem stands out for:

Japan has its own EDGECROSS standard complementing OPC UA, and OPC UA / Sparkplug B Japanese tutorials on Qiita and Zenn have proliferated rapidly through 2026.

Adoption Roadmap: From Zero to IIoT Production

The safest order when starting an IIoT project from scratch is:

  1. Data acquisition first. Pull data from one PLC over OPC UA. 1-2 weeks.
  2. Time-series DB for storage. AWS Timestream, Azure Time Series Insights, InfluxDB, TimescaleDB.
  3. Simple dashboard. Grafana or Power BI. Line charts and alarms.
  4. Edge gateway. EdgeX, Eclipse Kura, AWS Greengrass, Azure IoT Edge.
  5. MQTT broker. HiveMQ/EMQX or a managed IoT Hub.
  6. Digital twin. Azure Digital Twins, AWS IoT TwinMaker, Eclipse Ditto.
  7. AI/ML. Last. Train a predictive maintenance model after 6 months or more of data.

Installing all seven from day one is a recipe for failure. The settled wisdom is to start with one PLC + one metric + one dashboard, then expand.

Frequently Asked Questions (FAQ)

Q1. Is MindSphere gone?

A. It was rebranded. Siemens renamed MindSphere as Insights Hub in 2023, and simultaneously split off Industrial Edge (edge) and Industrial AI Suite (AI). Existing MindSphere customers were migrated to Insights Hub.

Q2. Why did Google Cloud IoT Core shut down?

A. The stated rationale in Google's August 2022 announcement was that "customers' IoT needs can be met by our core cloud data analytics and ML services." The service was fully terminated on August 16, 2023. As a result, the AWS-vs-Azure duopoly in IoT has strengthened.

Q3. Where did GE Predix go?

A. After the 2024 GE split, Predix moved under GE Vernova (power/energy). In the general manufacturing IIoT market, Siemens Insights Hub, PTC ThingWorx, and AWS/Azure took its place.

Q4. OPC UA or MQTT — which should I choose?

A. Use both. OPC UA is typically used in field communications (PLC to gateway), and MQTT is typically used for gateway-to-cloud communications. OPC UA Pub/Sub mode layers OPC UA messages on MQTT and lets you use both at once. Sparkplug B is the industrial-semantic standard on top of MQTT.

Q5. Industrial Edge vs Greengrass vs Azure IoT Operations — which is best?

A. It depends on your preferred cloud. If AWS, go with Greengrass; if Azure, IoT Operations (K8s-based) or IoT Edge (classic); if you live in the Siemens world, Industrial Edge. They all integrate deeply with their respective clouds, but you can reduce lock-in by abstracting data through a neutral layer like EdgeX.

Closing: May 2026, IIoT Cannot Be Finished With One Vendor

The conclusion of this article is simple. IIoT cannot be finished with a single vendor. Mixed stacks like Siemens PLC + AWS IoT Core + Azure Digital Twins + PTC ThingWorx are the new normal, and standards (OPC UA, MQTT, Sparkplug B, DTDL, AAS, ISA-95, IEC 62443) are what make that mixing possible.

The biggest shift is the acceleration of edge computing. During 2024-2025, NVIDIA Jetson, Azure IoT Operations (K8s), and AWS Greengrass all firmed up their edge tracks simultaneously. The era of shipping everything to the cloud is over; the standard pattern is to run ML inference at the edge and send only the result.

Don't get stuck on tool selection. Just by starting with "one PLC + OPC UA + a time-series DB + one dashboard", you've solved 80% of the problem. The rest is a question of priorities and security requirements.

References

Comments

No comments yet.

Sign in to leave a comment