Podcast: Crystal Ball 2026 – AI and everything else

In this episode of Great Question: A Manufacturing Podcast, Smart Industry content chief Scott Achelpohl wraps up the annual prediction series with highlights from SI's roster of subject-matter experts, who had a lot to say about how AI was moving well beyond the hype and coming to a production line near you.

Hello, everyone, and welcome to another episode of Great Question, a manufacturing podcast, and another brought to you by Smart Industry. I'm Scott Achelpohl, SI's head of content, and I'm joined for this episode by me. I'm flying solo on this installment for the sole purpose of encouraging you to check out our just-concluded 10-story Crystal Ball series on smartindustry.com.

The second—and the second successful—year, I might add, we featured the series on our site instead of as a downloadable ebook. So getting into it, it's interesting how, with an amazing lineup of subject matter experts, or SMEs, many of these predictions came to fruition in 2025 after the last series, and we expect the same out of 2026 when the next year is all said and done in December.

What we also expect is—and this is reflected in the new series—that AI, again, will be at the forefront of manufacturing technology talk. AI indeed was on the minds of almost all of our SMEs for Crystal Ball 2026. In fact, AI was hard to avoid, as much as we need to talk about other topics.

All these SMEs told us this year they believed AI was well beyond the hype stage, as we learned even before the series began from a RunSafe survey and a story we quoted from U.S., U.K., and German engineers and security professionals. And they said—and we reported December of ’24 before the series debuted—that AI is solidly, that is, in the middle of implementation.

Our IndustryWeek colleague Anna Townshend wrote a piece also before Christmas, and one that we featured, that held the same idea about AI implementation, as did our friends at ARC Advisory Group—Colin Masson, Craig Resnick, Inderpreet Shoker, and Patrick Arnold—in our Smart Operations Playbook, which was almost a preview of a lot of the ideas that emerged in the Crystal Ball series.

We also received sufficient warning during the series this year that data was proving to be an obstacle to AI adoption and implementation. Plenty of tech company–commissioned studies have also been showing this long before the series debuted, and it was definitely reinforced by some of the ideas in the Crystal Ball series.

Indeed, Nate Powery of MainPoint wrote for us January 7 and probably said it best: that governance of data will emerge as the new bottleneck breaker in 2026, and that AI—quote, unquote—accountability will hit full stride this year, and that manufacturers will not tolerate AI that’s not tied to hard metrics. In short, they won’t tolerate an AI investment that isn’t quickly tied to improved ROI.

We also were warned during the series that the manufacturing workforce—shrinking and often under-skilled as it has become—is a significant impediment to incorporating AI. As Stacey Ritchie of SmartCat, who has the interesting title there of Global VP of People, wrote for our January 12 roundup, many organizations are seeing a gap between the speed of AI deployment and employees’ ability to use it effectively.

Companies will perhaps anticipate that gap by phasing in agentic AI to perform some high-volume and very repetitive tasks, as we wrote about from the IFS Unleashed X conference in New York in October.

So let’s get into a lot more of the particulars from the series. Our repeat contributor Tim Gaus of Deloitte started the Crystal Ball series on December 26, writing about how AI would move from promising to production in 2026 and firmly out of the pilot stage, and that human workforce transformation as a result of AI adoption for real production would shift to the forefront of organizations’ strategic planning operations.

Tim also wrote that digital transformation, of which AI is often a part—and a topic we at Smart Industry cover just about every other day—is no longer optional for manufacturers. He also predicted that the rise of physical AI—advanced artificial intelligence married to physical systems like robots, vehicles, sensors, and other automation technology—is another early trend that is accelerating quickly and, beyond 2026, may be a crucial part of the workforce equation.

Our series with ARC Advisory Group friends also touched on a lot of the same things. And by the way, there’s that word workforce again. I will say it’s going to be a prominent part of our coverage in 2026—how the labor force is profoundly affected by AI.

Next in the series, before New Year’s on December 29, our frequent collaborator on cybersecurity, Frank Malonis of Kiteworks, delved into an interesting notion that AI copilots will recommend and sometimes even enforce cybersecurity policies. Frank pointed out that 75% of organizations moderately or extensively use AI—that’s quite a high number—and that 39% are adopting internal copilots, or, in other words, AI-powered virtual assistants that work alongside humans to boost productivity, automate tasks, and provide real-time insights by understanding context, analyzing data, and generating responses.

Frank, of course, put his cybersecurity hat on by writing that the next step is plugging AI copilots into an MCP server that would function as a centralized cybersecurity policy control pane. In other words, AI controls your IT and OT system security. That’s a bold statement from Frank.

Christopher C.J. Combs of Columbus kept going December 30 with a Crystal Ball lesson on why iterative AI is the path for enterprise success. For those who are not familiar, iterative AI is the continuous, cyclical process of building, testing, and refining AI models or outputs involving rapid feedback loops to improve performance—much like creating drafts of a document, but with AI.

C.J. made the point with us that although AI has been on boardroom agendas since as early as 2022, and even before that, a lot of those rushed AI initiatives were tried, then found to be quickly underperforming, and then got shelved. Let’s ask why. Because too many AI adoptions treat the technology as a one-shot miracle and not as a disciplined, iterative capability with the ability to adapt and evolve to a specific operation.

C.J. advises that companies should not think of AI as how it can solve every problem, but that they have to, quote-unquote, problem first and target the technology at a specific painful problem that matters most to their businesses. C.J. added that AI can reduce cycle times, improve quality, shorten queues, and free up human talent for other, more important and less repetitive tasks.

There we are in the workforce again—evolving it and not eliminating it. Many still perhaps fear that AI will become a job killer. We’re still getting that comment from a lot of people that we interview at Smart Industry.

Not shifting gears too much from the workforce theme, for a Crystal Ball piece on New Year’s Eve, David Vitek checked in with a snapshot of how the human-machine factory, as he put it, could look by upskilling employees and deploying AI at scale. David’s piece is on point because he links the absolute necessity of training programs with smart software.

David writes that by upskilling workers, measurable payoffs emerge, such as faster AI adoption, fewer implementation struggles, and higher employee retention. And as our ARC Advisory Group friend Colin Masson did in his own piece for us—Roadmap to Physically Intelligent Operations—that premiered before the Crystal Ball report started, David Vitek also draws distinctions between the laggards and the pacers with adoption.

After we rang in the new year, Chaz Spahn checked in January 2 with useful advice on how manufacturers, given their heavy reliance on vulnerable legacy systems, can prepare for when AI is weaponized. In other words, when bad actors use this technology to make cyberattacks more efficient than they already are.

Chaz also goes into great detail about how AI can supercharge cyberattacks. He really gets into the weeds, which is helpful, including how threat actors are leveraging AI models to instantly develop exploit scripts tailored to their desired targets and vulnerabilities.

This year, Smart Industry will be covering both sides of this coin with AI—its value to manufacturers and the workforce, but also as a menace to their systems and their inadequate and outdated defenses. Last year, I think it might have been Frank Malonis or another cybersecurity SME we featured who advised manufacturers to do complete cybersecurity readiness drills, including now against the threat of weaponized AI. That’s good advice for 2026 and beyond as well.

We’ll pivot to January 5. On our site, Ross Meyercord of Propel Software took on B2B’s use of generative AI in vendor selection for manufacturers—that is, not pilot programs, but real production deployments that are changing how buyers are researching and evaluating vendors.

We closed out the Crystal Ball series on January 6 with a variety of roundups, with partial predictions from experts at Kiteworks, ConfigIt, Gurucell, NordLayer, more from MainPoint, Catera, more from SmartCat, Emerson’s Aspen Technology, Operant, LeaseWeb, and HighPoint.

Just some of the highlights, from Aaron Sempley at HighByte: In 2026, manufacturing leaders will need to get their hands dirty, so to speak, and try out new technologies to determine which deliver the most value to their companies. Good advice going into any year, really. Aaron also writes that new tech tryouts will be key to taking advantage of the productivity increases AI does provide, and that making frontline employees work more efficiently is preferable to looking to AI as a completely autonomous solution. There it is—see the workforce enhancement again, not replacement.

This from Ron Thomas of SmartCat: Speed to market will become the most accurate indicator of ROI with AI, the test of whether AI is delivering real value. He adds, “This impact does not show up in abstract efficiency metrics, but in whether teams can prepare customer-facing materials, adapt them for multiple regions, and launch on schedule.”

And this from Nate Powery of MainPoint: Data governance will emerge as the new bottleneck breaker. As AI ubiquity grows, trusted data becomes the linchpin, he wrote. He also added that some manufacturers’ investments will shift toward cleaning, integrating, and securing data sets over time—flashy algorithms aside. Companies with reliable data foundations will outpace those bogged down by inconsistencies.

And finally, this from a group at ConfigIt, including prior contributor Heinrich Holgaard, Dan Joe Berry, and Laura Beckwith. They write, “Data interoperability will be a competitive advantage and will become a real differentiator as manufacturers realize that aligning systems across the product lifecycle is key to scaling AI effectively.” Reliable data foundations will become, as they put it, the new competitive edge.

You know, I’ll conclude by saying we had a lot of fun putting together the Crystal Ball series for 2026, and we hope our audience gained a lot of insights into the new year and beyond from it. It really was a good launching pad to what we’re going to cover in the new year at Smart Industry. And this series will surely return for 2027.

And with that, thank you for listening to this episode of Great Question, a manufacturing podcast. From Smart Industry, I’ll say goodbye to our listeners, and say have a great rest of your day and a happy new year.

About the Podcast
Great Question: A Manufacturing Podcast offers news and information for the people who make, store and move things and those who manage and maintain the facilities where that work gets done. Manufacturers from chemical producers to automakers to machine shops can listen for critical insights into the technologies, economic conditions and best practices that can influence how to best run facilities to reach operational excellence.

Listen to another episode and subscribe on your favorite podcast app