Podcast: Why connectivity is the missing piece to drive manufacturing performance with AI
Key Highlights
-
68% of manufacturers deploy AI, but only 19% report mature use—showing infrastructure, security, and skills gaps still limit large-scale industrial AI adoption.
-
Reliable connectivity is critical for AI-driven factories; nearly half cite network performance, edge computing, and bandwidth as top infrastructure needs.
-
Poor wireless reliability disrupts operations for 56% of manufacturers, highlighting the need for stronger connectivity for AGVs, robots, and mobile assets.
-
Weak IT–OT collaboration remains a barrier: 43% report little teamwork, slowing AI deployment and digital transformation across industrial operations.
In this episode of Great Question: A Manufacturing Podcast,
Samuel Pasquier, head of product management for Cisco System's Industrial IoT Connectivity Portfolio, says manufacturers want to use more AI in factories to drive efficiency and growth, but many programs are stuck in trial and failing to produce expected gains. New research from the networking technology company shows that two major factors are getting in the way: computing power and the network bandwidth to access that compute.
In this conversation with Smart Industry's Scott Achelpohl, Pasquier discusses what challenges still face AI-driven digital transformation efforts and what steps manufacturers can take to set themselves up for success.
Below is an except from the podcast:
SA: Samuel, my read of the new Cisco report where industry industrial AI is concerned is that the findings place outside emphasis on modernization of technical and network infrastructure in most manufacturing operations. For AI to be utilized at scale and on the need for IT and OT staff and systems to converge and collaborate. I guess I'll open it to to you to talk about the report. Is this your interpretation of the study? And I'd like to add, what else would you like to add?
SP: Yeah, sure, sure. So thank you, Scott again. So, you know, maybe for the audience, everyone know about Cisco as a IT company building network. We helped to build the internet for the last 40 years, but we also have been building network for industrial network for 20 years. So, helping our customer in manufacturing environment to build their infrastructure, to connect their machine, connect their plant, connect the factory floor, and two years ago we wanted to get a little bit of a state of industrial network, and we did a report similar report, and what was very interesting for us. is at that time, the majority of the respondents told us that AI will have the biggest impact on industrial network over the next five years. So we are down three years and we thought like, you know what, let's double check. Let's understand what's happening with AI in industrial environment and what are the impact that we see on industrial network. So let's learn directly from the practitioner. So we talked to 350 manufacturing customer really an operational leader in manufacturing environment to try to understand what's happening and that's really what we are delivering in this report and what we can go through a little bit and kind of talk about today.
SA: Okay, Samuel. So let's get into it some more. We have some questions, as you might imagine. Samuel, what do you make of the disconnect identified in the report between AI deployments, as we mentioned, 68%, and the much lower percentage, 19%, that regard their deployments as mature?
SP: The key things about that, and you know, we have to think about maybe the obstacle, right, to be able to scale AI. So we see a lot of people We talked about it in the previous part that we've done together, but really what is very clear out of the report is there's a few things that are hindering the deployment at scale, infrastructure limitation. We talked last time about the usage of machine vision, to do quality inspection, to do those kind of things. The reality is once you want to have a more global view of that around your entire infrastructure, then you need to have more performance, you need to have more bandwidth, you need to be able to store more data, to be able to have more correlation between the different information. That is one of the limiting factor.
The second thing that have been an obstacle is really around security as you connect more and more smart assets, which mean assets that are talking, which means they are connected to the network, you are increasing to some degree your attack surface. So how do customers can reduce the blast radius so they can get smart things that can talk? while at the same time not increasing their exposition to threat, right? And the last, which is really linked to this one, is you need to connect more things. You need to care about security, but there is a fundamental skill gap. You need to have people who understand the industrial environment, to understand how to, what you need to do with your machine. You know, we think about manufacturing, industrial automation. But at the same time, you need to have people who have this security mindset and understand what needs to be done in security. And this skill gap, having the large number of people at scale to be able to do that, I think that's one of the things that is limiting the massive adoption or the scale of some of those AI use cases, right?
SA: You've previewed a lot of what I'm going to ask about. The new report, Samuel, also has notable findings about how key wireless connectivity is to deploying AI at scale. And we talked some about that in our prior conversation, if you recall. The report says 56% of the manufacturing decision makers are saying that unreliable wireless disrupts their operations. What can be done to improve connectivity with AI on the doorstep as it is?
SP: Yeah, so, you know, maybe to maybe echo, you know, what we said last time we talked, you have more and more AGVs, AMR, and soon robots, and maybe Ubanoid robots as well, that will need to move around your infrastructure. And for those, they're going to be connected with the wires. They need to be wireless. And the RF environment in a factory or in those kind of settings is very challenging. You have a lot of metal, a lot of things that can do interference. And the key is, as much as Wi-Fi is fantastic, you know, if you have something that is moving around with a lot of interference, you may need to go a little bit beyond Wi-Fi. And that's really what we see. And talking about Cisco, you know, what customer can do, we have seen adoption of our technology, Cisco, reliable wireless backhaul technology that we integrated into Wi-Fi 7 so we can bring reliability to the connectivity of those machine and AGVs and AMR, which means those equipment can move around and the wireless connectivity can actually make a connection to the next access point before breaking the previous one. So pretty much what effectively it does, it allows you to move around and keep the network reliable and keep the connectivity on. And that is extremely critical. And that's why this technology now we see this adoption and we see that being standardized in some of those AGVs and AMR manufacturers, right?
SA: Samuel, we talk about this topic a lot. I think we maybe talked about it last time too. What makes IT and OT convergence and collaboration? so important for deploying AI at scale. Some context from the report, a larger percentage, 43% in the Cisco SAPIA research, said their organizations feature little to no IT and OT collaboration. A third of the respondents, 34%, also cite lack of collaboration between IT and OT teams as a major talent that is limiting AI-enabled operations at their organizations. Why must those silos be broken, specifically where AI is concerned?
SP: Yeah, I mean, that's a very good question. And, you know, I like to joke when I talk to people is you need to have people who are very strong on the OT side. So once again, you know, think about on the plan floor, people who understand initial automation, understand automation as like a manufacturing environment. But at the same time, you want them to be networking expert and security expert and while There are people like that in the world. It's a little bit like unicorn. You can find them if you have them in your team, keep them. Those are very precious. But if you want to scale and if you have a global presence, you're going to have to go and find a way to have your IT and your team working together. We at Cisco, we love to talk about collaboration, partnership, versus convergence. In convergence, it make feel like there's only one left at the end. I think you have room for both. You need to have both skill set. It's all about how do you make interaction between the IT and IT team works together. That's really what we hear from this report and what we see from our interaction with the customer.
At Cisco, we have been really focused on giving the tools that IT know, so IT know how to build network, structure network, manage network, secure network, But at the same time, give OT what they need, the right, for example, industrial protocol to have the right level of resiliency, maybe use the visibility that they will need, the troubleshooting tool that they will need, or like we just announced at Cisco Live Europe in February. It's about launching new tools dedicated for the OT personnel where they can help them to do their work, so giving more tools to the toolbox, with an agentic ops framework for them to ask questions and try to troubleshoot so they can get things connected and secure faster. That's really what it is. So, you know, to really answer your question on how you break the silos is really, I think people work together, I think people partner, and having a common tool that they can share to do things together, right?
SA: Interesting. I could talk about IT and OT convergence all day long. It's smart industry, we do that. Another factoid in the Cisco, and I don't mean to discount Sapio's role, I'll try to mention them as often as I can as well. I found interesting was another healthy percentage, 42% identified cybersecurity as the top obstacle to scaling AI. Now, you mentioned cybersecurity a little bit in passing earlier. Also, 54% said they expect returns on their AI investment within a year. My real question is, doesn't that pressure manufacturers if they want to incorporate AI much less at scale to figure out their data security posture and to do so pretty quickly.
SP: Yeah, so that's a very good question. You know, like when you think about it, when people think about security, there is a data security, right? To make sure your data is not leaking or being used or things like that. But when we talk about manufacturing, this is definitely a component. But I would say the biggest thing about security is people are more concerned about losing control of their infrastructure. So the data from the machine, you don't want that to leak, but it's not as critical as, let's say, a user data or your end user data. That is more impactful for you. But it's all about how do you secure the infrastructure so no one can take control. The plant is not going to stop. So while there is definitely a need for people to think about the data security posture from an AI standpoint, I will say today, I would say the top of mind is more about security of the process, making sure no one is going to stop it. The data security posture, I think, is still a few years out and people are starting to think about it. And I think it will come very quickly as people are going to export more and more data from their production tools.
SA: Samuel, here's my last question for you. Does the top line 68 percentage versus the 19% maturity percentage mean manufacturers for now must be experimental about AI or only phase it into smaller segments of their operations? The research does say process automation, 66%, and automated quality inspection, 54% are the most widely deployed use cases for AI in manufacturing.
SP: Yeah, I see what you mean. I think the thing is AI brings tremendous amount of value. So people do not want to waste. That's why you've seen those kind of number in process automation and quality inspection and so on. The way people are deploying AI today is more think about it at deploying it at the station. So looking at improving task station within the manufacturing process because that's I would say might be the easiest to go and implement. But the real value that is really exponential value is when you can use AI to optimize your entire process and to be able to do that, that's where the connectivity is important. That's where you need to go and go from optimizing a task to optimizing the set of tasks that communicate together and work one within the other. I think what we are seeing is just a natural evolution that you start by optimizing one station, one task, and as you connect more and more things, you can optimize the whole system. If I come back here in a few years, I think that's what we're going to see. that the system are going to be more and more connected. AI is going to be from a station to the food cell, to the full shop, to the full factory. And I think that's why the gain and the benefit were going to be exponential for the customer. So I think it's a journey and people are definitely leveraging AI where it has the biggest gain today. And as they get more and more connected, the gain will just be increasing over and over.
About the Author
Scott Achelpohl
Scott Achelpohl is the managing editor of Smart Industry. He has spent stints in business-to-business journalism covering U.S. trucking and transportation for FleetOwner, a sister website and magazine of SI’s at Endeavor Business Media, and branches of the U.S. military for Navy League of the United States. He's a graduate of the University of Kansas and the William Allen White School of Journalism with many years of media experience inside and outside B2B journalism.