23 Jul 2026
Why the automation industry is moving beyond vendor lock-in
automation · opc-ua · mqtt · industry-4.0
For nearly three decades, industrial automation ran on a simple philosophy: pick one vendor, buy everything from that vendor, and stay with them for the next twenty years.
It wasn't a lazy strategy. It built some of the most reliable factories in the world. It also built one of the strongest forms of vendor lock-in anywhere in enterprise technology — and that model is now coming apart. Not because Rockwell, Siemens, Schneider or Emerson started making poor products. Quite the opposite. The industry around them changed.
The old world
Until recently, an automation project was really a choice between a small number of complete worlds:
One vendor · one ecosystem
Same story, different logo
Everything worked together, because everything was tested together, and one vendor stood behind all of it. For a production plant worth hundreds of millions, that logic was — and still is — hard to argue with.
The hidden cost
The downside never shows up on day one. It shows up years later, the first time someone proposes replacing just the PLC.
Suddenly the engineering software changes. The HMI projects need migration. The historian connectors change, the alarms need rebuilding, asset management moves, the maintenance team needs retraining, and the spare-parts inventory turns over. Anyone who has actually costed one of these migrations learns the same lesson: the hardware was never the expensive part. The ecosystem was.
Proprietary doesn't always mean closed
One of the most repeated lines in this industry is "Rockwell uses proprietary protocols" — and it's only partially true. EtherNet/IP is an open standard managed by the ODVA consortium. OPC UA is supported. MQTT is supported. REST APIs exist.
What's actually proprietary is the engineering experience: Studio 5000, FactoryTalk, the project formats, the AOIs and UDTs, the motion and safety configuration. These things carry genuine value — they're a big part of why the integrated stacks are so productive. They are also exactly what makes migration expensive. The moat was never the wire protocol. It's the years of engineering effort stored in formats only one toolchain can open.
The historian tells the whole story
Here's my favourite example of how tangled this world really is. Plenty of engineers assume FactoryTalk Historian is a Rockwell product through and through. It isn't — it's built on the OSIsoft PI System, licensed and private-labeled.
Follow the ownership chain: OSIsoft was acquired by AVEVA. AVEVA is owned by Schneider Electric. Which means one of Rockwell's biggest competitors ultimately owns the historian technology inside Rockwell's own ecosystem.
If the vendors themselves are this interdependent under the hood, the "one closed world" framing was never quite real to begin with.
What modern architectures look like
The interesting shift is in the question being asked. Ten years ago a project kicked off with "which PLC should we buy?" Increasingly it starts with "how do we stop any single vendor from controlling the architecture?"
The reference answer looks like this:
Look at where the PLC sits. It's still essential — but it's one component in the architecture, not the centre of it. The centre of gravity has moved up into the data layers, and those layers are deliberately vendor-neutral.
Four technologies made this possible:
- OPC UA — a vendor-neutral access layer the whole industry now supports, however grudgingly.
- MQTT with Sparkplug B — event-driven industrial messaging that behaves the way cloud-native systems expect.
- The Unified Namespace — instead of every application polling every PLC, everything publishes once into a shared tree and everything else subscribes. One source of truth instead of a spiderweb of point-to-point integrations.
- Open time-series databases — TimescaleDB, InfluxDB, QuestDB — doing the historian's job at a fraction of the licensing cost, in formats you can take with you.
And the newest industrial software companies are revealing in what they don't build. Ignition, Beckhoff, HiveMQ, HighByte — none of them are trying to build a better complete ecosystem. They're competing on interoperability: how little they lock you in, not how much.
In fairness: lock-in exists for a reason
It would be dishonest to write this as a simple liberation story. Integrated ecosystems reduce risk — one vendor, one support contract, one architecture that was tested as a whole. In pharmaceutical manufacturing, food processing, or critical infrastructure, that still matters enormously, and it should.
Open architecture isn't automatically better. It optimises for different priorities: flexibility and data portability over single-throat-to-choke accountability. Knowing which of those your plant actually needs is the real architecture decision.
The next decade
The industry isn't abandoning its traditional vendors, and it shouldn't. What's changing is where the value sits. Twenty years ago, the PLC was the platform. In the coming decade, the data architecture is the platform — and the hardware, excellent as it is, plugs into it rather than defining it.
The vendors that thrive will be the ones that accept that inversion. The plants that thrive will be the ones that design for it now — because the future of industrial automation isn't vendor-specific.
It's vendor-independent.
Running an architecture decision like this right now? Tell me about it.