What SmartWorX Told Us About the Next 10 Years of Industrial Data
We spent two days at SmartWorX listening to engineers, architects and operations leaders talk about what's actually working and what isn't, in industrial AI. While the sessions covered a lot of ground, one idea surfaced repeatedly from different speakers in different ways:
Context is the next 10 years.
Not agentic, connectivity or data mobility. Context.
Three Generations, But Most Companies Are Still in Generation One
One of the clearest frameworks from the event described industrial data maturity in three generations:
Connectivity: Can I access the data at all?
Data mobility: Can I move it where I need it?
Context: Why is this happening? What does it mean? What caused it?
Most companies we spoke with are in generation one or two: They've connected machines and are moving data but when their AI system flags an anomaly, no one can explain why because the data lacks the semantic layer that would make it interpretable.
Context is what's missing and until it's there AI projects fail in production. The statistics from the conference put it bluntly: 70-95% of industrial AI initiatives never make it past the pilot stage. The bottleneck is the data foundation.
The Five Blockers Nobody Wants to Admit
One session laid out five specific reasons that industrial AI fails before it ever ships, that we've heard echoed in sales conversations.
Fragmented silos: No unified view across systems
Missing semantics: The same event means different things to maintenance and finance
Poor data quality: Garbage in, garbage out, no matter how good the model
Tribal knowledge: Critical context lives in people's heads, not in systems
Inconsistent definitions: No shared vocabulary across teams or plants
A Unified Namespace (UNS) solves all five simultaneously: one semantic source of truth, governed definitions, context attached at the source. Rather than replacing existing systems, UNS requires organizing them.
You Can't Skip Rungs on the Autonomy Ladder
Another speaker introduced a framework that mapped almost exactly onto the way we think about the HiveMQ platform:
Visibility → trusted data → contextualized data → AI assistance → agents → autonomy.
The argument: autonomy is earned iteratively. Companies that try to jump straight to agentic AI without building the foundational rungs keep ending up back at the drawing board. The technology is right, but the sequencing is not.
Connect first. Contextualize the data. Analyze at scale. Only then does acting autonomously become safe and reliable.
This isn't just HiveMQ framing. It's what the market arrived at independently and it's the architecture we've built.
The Organizational Problem Nobody Budgets For
One of the most uncomfortable truths from the event: 70% of transformation programmes have failed for 40 years because the organizational structure that buys and deploys technology is 150 years old.
IT and OT are funded from separate budgets, they speak different languages and serve different masters. A platform that sits between IT and OT - connecting, contextualizing and making data legible to both sides - changes the calculus. HiveMQ is designed to be that connective tissue.
What This Means If You're Building Industrial AI Today
The three-generation model is a useful diagnostic. If your AI initiatives are stalling, the question worth asking isn't "which model should we use?" it's "which generation are we actually in?"
Most organizations are somewhere between generation one and two. Machines are connected and data is moving, but the semantic layer - the part that makes data interpretable across teams, systems and use cases - hasn't been built yet. That gap is where pilots go to die.
The good news is you don't have to boil the ocean to get started. A criticality assessment, identifying the 20% of data sources that drive 80% of the value, is a tractable first step. The foundation doesn't need to be complete to be useful, it needs to be right one step at a time.
The organizations pulling ahead treat context as infrastructure - something you build once, maintain continuously and compound on top of.
The industrial data market spent two days at SmartWorX arriving, independently, at the same conclusion: Context is the next 10 years.
The companies that build the right foundation now will be the ones with a meaningful head start when the rest of the market catches up.
Start on your data foundations now, speak to the HiveMQ team about Unified Namespace today.
Shashank Sharma
Shashank Sharma is Director of Product Marketing at HiveMQ, focusing on the company’s MQTT-based Industrial AI data platform across cloud and self-managed deployments. He is passionate about technology and developer-centric workflows, with 12+ years’ experience across software development, sales, and marketing for platforms and tools in numerical computing, autonomous driving, robotics, and AI.
