IMTS 2026 Key Learnings: Physical AI Needs Good Factory Data First

I went to IMTS 2026 expecting to hear a lot about AI.

And I did.

Physical AI. Industrial AI. Smart machines. Autonomous systems. AI-powered everything. But somewhere between the robots, machine demos, and conversations about the future of manufacturing, one thing kept pulling my attention back to something much less flashy: DATA.

IMTS 2026 took place September 14–19 at McCormick Place, Chicago, bringing together 90,000+ attendees and 1,700+ exhibitors from across the manufacturing world.

I was there with the Azilen team, walking the floor, talking to manufacturers, and seeing the latest technologies up close.

And the more I looked, the clearer it became: every AI story eventually comes back to the data underneath it.

What IMTS 2026 Was Really About

The theme for IMTS 2026 was “Achieve the Impossible.” And on the show floor, that meant seeing just how far AI and automation can go inside a real factory.

Six days of machining, automation, additive, metrology, and a new Industrial AI Arena. But one idea kept coming up: AI on the shop floor is no longer just a demo. It is a data test.

What was happening on the ground:

→ Demand is rising, while workforce constraints are pushing manufacturers toward AI and automation.

→ U.S. manufacturing technology orders grew 28.9% in the first four months of 2026.

→ Buyers are asking tougher questions: Where will AI create real value? Can it work with the factories we already have?

→ And underneath it all, factory data is still scattered, incomplete, or simply not ready to use.

The Industrial AI Conference was not a concept talk. It was built around sector case studies, implementation trade-offs, and measurable impact on uptime, quality, cost, and throughput. Physical AI sessions, like “From Automation to Autonomy“, showed where robots are heading.

The recurring pattern was clear. AI on the factory floor only delivers when it:

→ Runs on clean, usable machine data
→ Works at the edge when speed matters
→ Fits the machines already on the floor
→ Keeps people in control of decisions

The factories that win will not be the ones with the most AI demos. They will be the ones with sensors that see, gateways that connect, and telemetry they can trust.

The conversation has shifted from “Which AI should we buy?” to “Is our data ready for it?”

Now, below are the key learnings I gathered from IMTS 2026, and what each one means for manufacturers building connected, AI-ready operations.

5 Key Takeaways: IMTS 2026

1. Industrial AI Now Has to Prove Itself

This year IMTS added a dedicated Industrial AI Arena and a full-day Industrial AI Conference. AMT’s goal was to help manufacturers move past experiments and see where AI delivers real business value.

That tells you where buyers are. They are not asking, “What can AI do?” They are asking, “What will it do for my uptime, quality, and cost?”

The Arena focused on deployed systems for quality inspection, predictive maintenance, and process optimization. The demand is clear: show the result.

2. The Real Problem Is Usable Data, Not the AI Model

The Industrial AI Conference kept coming back to one idea: start with the data. Its program stressed data-centric, edge-first approaches, and measurable impact on uptime, quality, cost, and throughput.

I see the same thing in our projects. A model is only as good as the signals feeding it. Machines speak different protocols, and data often arrives late or not at all.

This is why gateways matter. We built a multi-protocol IIoT gateway ecosystem for 100+ manufacturing machines. It handles 70+ GB of telemetry per machine each year across 80+ gateways. The gateway filters the noise, holds data during outages, and passes clean telemetry forward. Our Industrial IoT gateway guide explains how.

3. Physical AI Is Coming, and It Runs on Sensor Data

Physical AI was one of the biggest topics at the show. FANUC America’s CEO described it as robots that can see, reason, and act in real production environments. Conference sessions also covered how Physical AI learns by mixing simulation with real-world feedback.

The simple point is this: a machine can only act on what it can sense. No sensors means no signals. No signals means no smart action.

That is where sensorization comes in. Sensors capture how a machine is doing. Telemetry streams that data live. Physical AI sits on top of both.

4. Old Machines Can’t Be Left Behind

Most factories run machines of many ages. One session argued that connectivity should be planned at the machine design stage to support lights-out production and predictive maintenance. That works well for new machines. But what about the ones already running?

The stakes are real. AMT shared that 71% of high-performing job shops run some unattended operation. They average 15 machine hours a day, versus 8.5 for shops without automation. You can’t run what you can’t see.

The answer is retrofitting. We took a retrofit-first approach for 1,000+ commercial dishwashers, combining hardware, edge processing, and secure connectivity. It gave the client real-time monitoring, remote diagnostics, and $1M in potential five-year savings. The same thinking applies on the factory floor.

5. Manufacturers Are Ready to Scale, Not Just Test

U.S. manufacturing technology orders grew 28.9% in the first four months of 2026. And the questions buyers ask are practical. Will it work in the factories we already run? How much oversight does it need? One session even focused on standardizing automation across multiple sites.

That is a scale question. Scaling means every plant and machine speaks the same data language.

We did this with a vendor-agnostic energy platform. It unified telemetry from 25,000+ industrial assets and standardized 60+ parameters. Once data is standardized, adding the next plant gets much easier.

Here Is Where We Fit Into All of This

IMTS is the kind of show where the machines steal the spotlight. We came to talk about what sits behind them: the data.

Factories make a huge amount of machine data every day. But much of it stays stuck. It sits inside old controllers, in different formats, on different networks. Teams still make big calls with half the picture.

We came to IMTS 2026 to put our IoT and Industrial AI work in front of people living that reality. You could find us at Booth 236374 at McCormick Place.

Most manufacturers do not lack ambition. They lack a clean, connected data layer that AI can trust. That is the gap our IoT team works on every day.

What We Brought: Connected Factory Intelligence

Picture a normal day on a busy shop floor. A machine starts running hot. The signal exists, but it sits in one system. The maintenance team looks at another. Nobody connects the two until the machine stops. Then the line is down, and everyone is chasing answers.

This is not rare. It is what happens when machines, sensors, and software do not talk to each other.

Our approach fixes it step by step. Capture the signal. Stream it. Clean it. Turn it into a decision.

Here are three real projects behind our work:

Azilen Works Inside What You Already Have

The first question factory teams ask is simple: do we have to replace our machines?

The answer is NO.

Our gateways sit between your current machines and modern cloud platforms. They translate between protocols like Modbus, OPC UA, and MQTT, so older PLCs, SCADA systems, and CNC machines can share data with newer tools. Our approach also connects with systems like MES, ERP, SCADA, PLM, QMS, and CMMS.

No rip-and-replace. It simply adds intelligence to what you already have. Learn more about our IoT development services.

What Makes It Different for Manufacturing

A factory is not an office. Networks drop. Machines speak many languages. Data volumes are huge. A tool built for a normal app will not last long here.

Here is the stack we brought to the booth:

Connected Machines: We capture machine data in real time.

IoT Sensorization: We add sensors so assets become visible and measurable, even older ones.

Factory Telemetry: We stream operational signals as they happen, so you see problems early.

Industrial AI: We turn factory data into decisions. AI spots patterns, odd readings, and early warning signs.

Digital Twins: We build a virtual copy of your process. You test changes there before touching the real line.

Our gateways are also built for real factory conditions:

→ They filter repeated readings, so the network is not flooded
→ They store data locally during outages and send it when the connection returns
→ Every gateway gets a secure identity
→ You can manage and update gateways remotely

Behind all this are 17+ years of engineering, 100+ manufacturing solutions delivered, and 500+ product engineers.

Key Takeaway

IMTS 2026 was six days of proof that manufacturing is getting serious about real results, not just new technology. Here is what stood out:

AI has to prove itself: The show built a whole Industrial AI Arena and conference around measurable value. Demos alone are not enough anymore.

Data comes first: A smart model cannot fix messy, missing, or late data. Clean data comes before smart AI.

Physical AI runs on sensor data: A machine can only act on what it can sense. Sensors and telemetry are the base layer.

Older machines matter: Most plants run machines of many ages. Retrofitting is how they join the connected factory.

The market is ready to scale: Manufacturers now ask how to roll out across plants, not whether to try it.

Six days in Chicago confirmed what we already believed. The manufacturers who came to IMTS 2026 are done with experiments. And our IoT team was built for exactly this.

Missed us at the show? Download our brochure, explore our Industrial IoT solutions, or book a meeting.