28 posts tagged with "thought-leadership"

The sustainability conversation in manufacturing usually starts with solar panels on the roof, LED lighting, and maybe a heat recovery system on the compressors. These are important investments. They're also insufficient.

The largest single source of waste, excess energy consumption, and avoidable emissions in most manufacturing plants isn't the HVAC system or the lighting — it's equipment running inefficiently because nobody noticed the bearing was failing, the seal was leaking, or the motor was drawing 15% more current than it should.

Predictive maintenance is, quietly, one of the most effective sustainability initiatives a manufacturer can implement. Not because it was designed for ESG — but because preventing failures systematically eliminates the waste, energy overconsumption, and material losses that failing equipment creates.

The data on this is surprisingly clear, and almost entirely overlooked by sustainability teams.

The pilot worked beautifully. Your AI model predicted bearing failures on Line 3 with 94% accuracy. The CEO saw the demo. The board heard about "digital transformation." Budget was approved for a plant-wide rollout.

That was eighteen months ago. The model still runs on Line 3. Maintenance still uses clipboards everywhere else. The data scientist who built the pilot left for a fintech startup. And nobody can explain why a model that worked perfectly on one line won't work on the other seven.

If this sounds familiar, you're not alone. According to a McKinsey survey on AI in manufacturing, 87% of manufacturing AI projects never make it past the pilot phase. Not because the AI doesn't work — but because the organizational, data, and infrastructure challenges of scaling from one line to a full plant were never addressed.

The AI isn't the problem. The pilot model is the problem.

The manufacturing execution system (MES) market hit $16.7 billion in 2025. By 2030, analysts project $28.3 billion. And yet, the most interesting thing happening in manufacturing software isn't MES growing — it's MES being absorbed.

IIoT platforms are eating MES functionality from the bottom up. What started as simple machine monitoring (connect a sensor, see a dashboard) has expanded to include OEE tracking, downtime analysis, quality management, production scheduling, and work order management — the traditional domain of enterprise MES.

Meanwhile, MES vendors are adding IIoT features — edge connectivity, real-time machine data, predictive analytics — from the top down. The two categories are converging, and the result is a fundamental disruption of how manufacturers think about their factory software stack.

If your plant is running a 10-year-old MES — or worse, if you're about to sign a 7-figure MES contract — this convergence matters to you. Here's what's actually happening and what to do about it.

Every IIoT deployment hits the same wall within the first week: IT. The factory floor needs to send machine data to the cloud. The IT department needs to approve network access, configure firewall rules, set up VLANs, conduct security reviews, and integrate the new traffic into their existing network architecture. What should be a two-day deployment becomes a three-month project — not because the technology is complex, but because the organizational process around network access was designed to prevent exactly the kind of connectivity that IIoT requires.

Cellular IIoT connectivity eliminates this wall entirely. Instead of routing machine data through the plant network, cellular-connected edge devices use their own mobile data connection to send data directly to the cloud. No IT involvement. No network configuration. No security review. No firewall rules. The machine data never touches the plant network at all.

This is not a workaround or a compromise. For a growing number of manufacturers, cellular connectivity is the architecturally superior approach to IIoT deployment — faster to deploy, more secure in practice, and cheaper when you account for the true cost of IT-dependent deployments.

Walk through any manufacturing plant in 2026 and you'll still see them: clipboards. Stacks of paper forms. Operators writing down temperatures, pressures, cycle counts, and quality measurements every hour. Data that gets entered into a spreadsheet the next day — if it gets entered at all.

This ritual persists because it feels free. The forms cost pennies. The operators are already there. What's the harm in a few minutes per hour with a clipboard?

The harm is enormous. And it's invisible precisely because nobody tracks the cost of tracking.

ISA-95 was created in the late 1990s to solve a simple problem: how should enterprise systems (ERP) communicate with plant floor systems (PLCs and SCADA)? Two decades later, IIoT platforms have disrupted the neat hierarchical model that ISA-95 defined. Data now flows from sensors directly to the cloud, bypassing every layer in between. The question for manufacturing engineers in 2026 isn't whether ISA-95 is still relevant — it's how to reconcile a framework built for hierarchical, on-premises architectures with the reality of cloud-native, edge-computing IIoT platforms.

Every manufacturing plant claims to be "doing predictive maintenance." In reality, most are somewhere between reactive and preventive, with a few vibration sensors they call "predictive" because a vendor told them to.

This isn't a criticism — it's a diagnostic. Understanding where you actually are on the maintenance maturity model is the first step to getting where you need to be. And more importantly, understanding which level makes sense for your plant, because not every operation needs to reach the peak.

McKinsey calls it "pilot purgatory." Gartner calls it "the trough of disillusionment." Plant managers call it something less polite.

The data is brutal: according to McKinsey's Global Lighthouse Network research, approximately 70% of Industry 4.0 pilots never make it past the pilot phase. They generate interesting data, produce impressive presentations, and then quietly die — the budget reallocated, the champion promoted to a different role, the hardware gathering dust in a server closet.

This isn't because Industry 4.0 doesn't work. It's because most pilots are designed to fail from day one. Here are the seven reasons why — and how to avoid each one.

Every major IIoT conference in 2025 and 2026 has had at least one vendor breathlessly promoting 5G private networks as the future of manufacturing connectivity. "Ultra-reliable low-latency communication! Network slicing! Massive machine-type communication! One million devices per square kilometer!"

The hype is real. But so is the technology — when applied to the right use cases. The problem is that most manufacturers don't need a 5G private network. They need reliable, low-latency connectivity to their PLCs. And for the vast majority of factory IIoT deployments, existing cellular (4G LTE) and industrial Ethernet already deliver that.

Let's separate the genuine use cases from the marketing noise.

Autonomous Maintenance (AM) — the TPM pillar where operators take ownership of basic equipment care — has been practiced in manufacturing for decades. The idea is sound: operators who run machines every day are best positioned to detect early signs of degradation. They hear subtle changes in sound, feel unusual vibrations, and notice when something doesn't look right.

The problem is execution. In most plants, autonomous maintenance means laminated checklists, clipboards, and handwritten logs that sit in a binder until audit time. Operators dutifully check boxes ("Lubrication points — OK") without the tools to quantify what "OK" actually means. Is the bearing temperature 65°C (fine) or 85°C (about to fail)? The clipboard doesn't say.

IIoT is transforming autonomous maintenance from a human-only discipline into a data-augmented system where operators combine their physical presence and intuition with real-time machine data. The result: better detection, faster response, and maintenance culture that actually sticks.