Vol. 1 · Issue 7 · April 27, 2026
The Connection Point
Automation, packaging, and CPG news — with Colin's take on what it means for you.
This week's issue covers a lot of ground, but one thread runs through most of it: the gap between what the industry is investing in and what the infrastructure underneath can actually support. The Iran conflict is pushing CPG input costs higher. Packaging materials suppliers are navigating a world where their OEM customers want sustainable materials running on existing machines. And the automation sector is having a serious conversation about whether the AI it wants to deploy can even move across the networks currently installed in most plants.
The ARC Advisory Group piece on AI-ready networking this week is worth sitting with. It doesn't make a lot of noise, but the argument it makes is one that controls engineers and machine builders should be paying close attention to: AI doesn't run well on aging infrastructure, and you can't just bolt capability onto a network that was never designed for it. There's a deeper thought this week along those same lines.
Good reading.
— Colin Cartwright, Connectivity Colin
Smart Factory: The Rise of PoE in Industrial Environments
Power over Ethernet is gaining serious traction on the factory floor. EDN's deep-dive looks at how PoE is being adopted across industrial automation and IIoT deployments, driven by the need to simplify physical infrastructure while accommodating the growing density of connected devices. A single cable carrying both power and data is an appealing proposition when you're wiring a machine or a production line where every extra conduit adds time and potential failure points.
Colin's Take
PoE is not new, but its relevance in industrial settings is growing precisely because of IIoT density. When you're running 40 or 50 sensors and edge devices across a packaging line, the cabling conversation becomes a real design constraint fast. PoE helps where you need to power smaller devices without pulling a separate circuit, but it has real limitations at higher power draw and in washdown environments where IP ratings matter as much as bandwidth. For packaging machine builders, this is worth watching as a supplementary architecture tool rather than a wholesale wiring solution. The question is always: what's actually at risk if that cable fails, and can you diagnose it from the HMI at 2 a.m.?
Mondi Brings Sustainable Packaging Innovation to Interpack 2026 Through 15 OEM Collaborations
Mondi isn't just showing materials at Interpack 2026. They're running sustainable packaging live on machinery built by 15 different OEM partners, demonstrating how fiber-based and recyclable materials can run on existing equipment without forcing a line redesign. The breadth of the OEM partnership signals a deliberate strategy: meet machine builders where they are rather than asking their customers to buy new machines to support greener materials.
Colin's Take
This is smart positioning from Mondi. The sustainability conversation in packaging always runs into the same wall: CPGs want greener materials, but OEMs can't always guarantee the new substrate runs cleanly on existing tooling. By putting sustainable films and fiber packaging on 15 different machines in a live environment, Mondi is doing the qualification work that makes adoption easier for both the machine builder and the brand. For OEMs evaluating material suppliers, this kind of partnership model is worth paying attention to. It shifts the conversation from "will this material work?" to "here's proof it already does."
The Evolution of Pouch Packaging
Packaging Strategies looks at how pouch format machinery has broadened well beyond its origins in stand-up pouches and retort applications. Today's pouch lines integrate spout insertion, blown film extrusion support, and full-line solutions that cover far more product types than the early adopters ever expected. The format's flexibility is driving adoption across food, beverage, home care, and personal care, with machine builders now offering increasingly integrated systems.
Colin's Take
Pouching is a format that keeps expanding its footprint precisely because it's lighter, cheaper to ship, and increasingly compatible with recyclable films. The interesting machine design challenge is that as formats multiply, so does the demand for changeover speed and flexibility. A machine that runs five different pouch configurations needs a controls architecture that can handle that variation without burying your operator in a manual. The lines that handle format diversity well are almost always the ones where the machine builder invested in decentralized I/O and smart diagnostics, because that's what lets a technician understand what changed when a format switch goes wrong at the start of a run.
Missing Diet Coke to Costlier Groceries: How the Iran War Is Hitting Your Daily CPG Life
It's not an abstraction anymore. The Iran conflict is pushing energy costs higher, which is flowing directly into packaging material pricing, ingredient sourcing, and transport logistics for CPG companies. Some product lines are being pulled from shelves. Others are getting more expensive. The piece is a useful consumer-facing lens on pressures that CPG manufacturers have been managing internally for weeks.
Colin's Take
When geopolitical disruption starts showing up in grocery aisles, it concentrates minds at the CPG level in ways that internal cost reports often don't. Packaging is usually one of the first places manufacturers look for savings when margins compress, which historically means two things: shorter run lengths as SKUs get rationalized, and increased pressure on machine OEMs to support faster changeover and smaller batch sizes without losing throughput. If you're building or selling packaging equipment right now, understanding your customer's cost environment isn't just context. It shapes what they'll actually buy.
Modern Networks Are the Foundation of AI-Ready Manufacturing
ARC Advisory Group makes a straightforward argument that doesn't get enough attention: AI-powered manufacturing systems place fundamentally different demands on industrial networks than traditional automation does. High bandwidth, low latency, deterministic communication, and the ability to move large volumes of data in real time are not optional features when AI is doing the work. Plants that invested in legacy networking infrastructure will hit a ceiling before they hit their AI goals.
Colin's Take
This is the piece I'd put in front of any CPG manufacturer or machine builder currently planning an AI deployment. The instinct is usually to focus on the AI application layer, the software, the dashboards, the predictive models. But the network underneath determines whether any of that can actually run in a production environment. Manufacturers who skipped network modernization because the traditional automation was "working fine" are going to discover the limitation at exactly the wrong moment. Getting AI decisions from sensors to controllers to the cloud and back again in real time isn't a software problem. It's an infrastructure problem, and it starts on the plant floor.
ABB Launches High-Speed PoWa Cobot Family to Bridge the Gap Between Collaborative and Industrial Robots
ABB Robotics released its PoWa cobot family this week, a range designed to operate at speeds that traditional collaborative robots can't match while remaining safe enough for close human interaction. The gap between cobots and full industrial robots has been a persistent frustration for machine builders who need throughput that cobots can't deliver but don't want the guarding, footprint, and integration complexity of a full industrial arm. The PoWa series is ABB's attempt to close that gap.
Colin's Take
The cobot speed problem is real. I've sat in rooms with packaging engineers who wanted collaborative robots for the safety and flexibility benefits but couldn't justify the throughput compromise for their application. If ABB has genuinely moved the needle on speed while keeping the cobot safety profile intact, this matters for secondary packaging applications where pick-and-place cycle times are often the bottleneck. The bigger question for machine builders isn't just the robot itself. It's how the I/O architecture around it gets designed, because a faster robot feeding a slow control loop doesn't solve the throughput problem. It just moves it.
Wispr Flow — Voice Dictation That Actually Thinks
I've been using Wispr Flow for a while now, and it's changed how I handle the parts of my day that used to eat time: follow-up emails, CRM notes after a customer visit, drafting LinkedIn posts while thoughts are still fresh. You talk, it writes. Not a transcript of what you rambled, but clean, structured, punctuated text that reads like you meant to write it that way. It works across every app on Mac, Windows, iPhone, and Android, so wherever the cursor is, you can just speak. For anyone in sales or technical roles who spends too much of their day typing things that could be said in half the time, this is one of those tools you won't want to work without after the first week.
Try Wispr Flow free for 30 days →Disclaimer — if you sign up you get a free month and I get a free month at no cost to you.
You Can't Run the Future on Yesterday's Infrastructure
The ARC Advisory Group piece this week makes a point that the industry keeps dancing around without quite landing on: AI in manufacturing is fundamentally a data movement problem, and most plants were not wired to move data the way AI requires. Traditional automation networks were designed for deterministic control traffic, small packets, predictable timing. AI applications want to pull from many sources simultaneously, send data to cloud or edge compute systems in bursts, and receive decisions back fast enough to act on them. Those are different demands, and aging network infrastructure wasn't built to accommodate both at once.
This comes up in conversations I have with machine builders more than it probably should. The question usually isn't whether the AI tool works in the demo. It's whether the plant network it's being deployed on can support it under production conditions, with other systems running, across a full shift. More often than not, the answer reveals a limitation that nobody budgeted for because it wasn't visible until the AI deployment was already underway.
There's a parallel here to what happened with IIoT a decade ago. Companies bought sensors and gateways and found out that their Ethernet topology, their switch capacity, and their cable plant couldn't carry the load cleanly. Some retrofitted. Many paused indefinitely. The ones who got ahead of it, who treated network modernization as a prerequisite rather than a concurrent workstream, ended up with something they could actually build on.
AI readiness is a phrase getting thrown around a lot right now. The honest version of it isn't a software checklist. It's a question about what's physically installed between your sensors and your servers, and whether it was designed for what you're asking it to carry. That audit tends to be less exciting than the AI roadmap, but it's the one that determines whether the roadmap goes anywhere.
Could your machine pass the 2 A.M. Test?
Book a Quick Connectivity Chat with Colin.
Book a Connectivity Chat35 years in automation. One focus: simpler wiring for packaging OEMs and CPGs.