“The value is moving from the metal to the code.” As manufacturers add AI and software to their products, the old one-time hardware sale is running out of runway when it comes to improving AI models for manufacturing. For Thales’s Dale Hopkinson, the real opportunity — and the real risk — sits in how that software and AI actually get monetized: not as a bolted-on price tag, but as a flexible, recurring business built to change as fast as the technology does.

In this episode of Manufacturing Matters, TECH B2B Marketing’s Winn Hardin sits down with Hopkinson, Senior Product Manager for Cloud Services at Thales, to unpack why software is decoupling from hardware, what that means for OEMs retasking and repackaging existing equipment, and why Bain projects AI, software, and data will add roughly $70 billion to the market within five years.

The conversation covers the organizational groundwork companies need before they can monetize AI — P&L ownership, dedicated staff, and roadmaps that survive personnel changes — plus the shift from subscription tiers to usage-based and hybrid pricing, the operational lift of managing renewals, entitlements, and channel relationships, and why protecting AI models and training data on the edge is now a board-level conversation.

Winn Hardin: [00:00:01] Hey everyone, and welcome to the latest episode of “Manufacturing Matters,” where we talk about the technology and the trends reshaping the global manufacturing industry. My name is Winn Hardin and I’ll be your host today. And today we’re lucky enough to look at a very interesting topic. It’s related to artificial intelligence, which is what we’ve been talking a lot about in recent months —— and even years. And we’ve got an expert in that area, in cloud services, that’s going to help us to understand how companies can map a course forward as we go forward with AI model optimization, and we leverage more and more of this into industrial applications. So today, I’m lucky enough to be joined by Dale Hopkinson, Senior Product Manager of Cloud Services at Thales. Dale, how are you doing, sir?

Dale Hopkinson: [00:00:41] I’m doing well. Winn. Good to see you again. Great to be on to support the community.

Winn Hardin: [00:00:46] Wonderful. I really appreciate it. Now, Dale, we spoke not too long ago at the Automate conference, but we were doing a very short, 15-minute Automate Live segment. And I’m really excited that you were able to come and join us back on the “Manufacturing Matters” podcast so we can have a more in-depth conversation about AI and where we’re all going from here. So thanks for making time.

Dale Hopkinson: [00:01:04] Absolutely. It’s my pleasure. Thanks for having me on.

Winn Hardin: [00:01:07] So, Dale, do us a favor. Kick us off a little bit. Tell us a bit about yourself. Tell us a little bit about Thales and how cloud services fits within the enterprise.

Dale Hopkinson: [00:01:14] Yeah, absolutely. Hey everyone. So I’m Dale Hopkinson. I lead our product management from the innovation and cloud services side. So with Thales, a large company, I have three primary things we do. We have security and defense and aerospace security, and enterprise and cyber security, which I’m representing today. As part of our group, we have a team that does software monetization. So companies, especially in this space —— think of like your Stäubli, your Hexagons, your Siemens —— as they add software and AI, and they want it to be a revenue stream for their business, they need a set of modern monetization tools. And that’s what we offer them. What I manage on our cloud services team, we essentially give them the ability to protect their software and AI, as well as to sort of control the business model in a very flexible way so they can change the price and packaging without calling the engineering teams. And not only for this space, but as I mentioned, for enterprise, right? This is a natural SaaS and enterprise software motion. And it’s just an area now that as manufacturing puts more emphasis and growth through software-defined automation, they’re now learning that these sorts of capabilities exist, and it’s an area that is part of their stack that they need to invest in to scale through software.

Winn Hardin: [00:05:03] You know, I just see this as such an important topic right now, not just within the industrial base and the manufacturing companies that we’re talking about today and that I work with so much. It’s across the board when it comes to artificial intelligence. And we’ve got new models that are being launched. There’s a lot of discussion about how do we generate ongoing revenue to pay for the processing requirements, the overhead, and model evolution, right? And continual training and optimization. So it just seems critical if we’re going to do like Bain has shown us in recent research, that software is becoming more critical for industrial applications and revenue sources —— and especially those that are AI focused —— that we really need to be thinking long term. We can’t just do a launch-and-forget scenario when it comes to model optimization, right? And somehow these OEMs, if they’re going to keep bringing us new and better solutions, they’ve got to find a way to support that development work. So I think this is a critical time. So tell me, how is the role of software changing in industrial applications?

Dale Hopkinson: [00:06:04] You know, it’s actually very exciting to see. I’ve been connected to software automation for, I don’t know, years. I think I first spoke at Automate maybe over 10 years ago. Software … The shift that’s mainly happening is, one, it’s being decoupled from hardware, right? So the value is moving from the metal to the code, as I like to say. And with that, software is becoming less of ‘features’ and more of actually full-grown products, right? Products that are creating value on top of the environment. I know everyone is talking about AI vision and digital twins, right? But the value that’s being created through software, in some cases, is increasing the lifetime value of the hardware. It’s unlocking new value through analytics and data for companies. I mean, it is really bringing not just new value from the innovation side, but as you mentioned, and as Bain mentioned, new revenue opportunities. I think within five years or so—— Bain, they had a research study that says AI software and data are going to add about $70 billion to the market. So it’s creating new markets in addition to creating new value. One of the things I look at really in automation is: There’s a very interesting parallel between what happened years ago with software defined networks, when software went from buying a physical router and software from the same company —— Cisco, right? —— to being opened up to everyone. You see a lot of startups now in industrial automation. They have AI vision solutions and other solutions for robots. So it’s very exciting. It’s creating new markets, creating new value, and really making manufacturing a very innovative industry.

Winn Hardin: [00:07:49] You know that brings something to mind, which has always been interesting to me, which is retasking. If we’re talking about a server or if we’re talking about a router, the hardware doesn’t significantly change in doing SDR modifications or whatever it might be to add new functionality. It’s not too much of a leap. But when we can leverage, when software defines the application layer in an industrial application, are we finally going to see a situation where companies can get even more use out of an automation hardware scenario because they can retask it, move it possibly, or just, you know, have more flex? Is it more about flexibility in the operations for an existing work cell, or are we giving more legs to the hardware that we’ve got in the field or that we’re putting in there now?

Dale Hopkinson: [00:08:33] Well, absolutely it’s a bit of both. I mean, I’ve seen this in networking, automotive, as well as non-manufacturing. I mean, especially now with the cost of hardware so unpredictable and the timelines of delivery, right? If you can get your hardware SKU down and have that hardware doing multiple things and retask it, as you say, or repackage it, from the monetization perspective that becomes very advantageous. For the customer, like you said, for the manufacturing floor, they can retask and they can get more done. And for the provider, you’ve actually now increased your land and expanded, right? Once you get your physical device on the floor, you sold that, you can expand that to different use cases and to different models by simply now turning new software or new AI on through monetization. This is exactly the key point of what’s happening, where I get excited because I’ve been doing this and solving for SaaS and enterprise software. And now to see this in manufacturing … I think one of the early indicators of exactly this retasking —— I forget when it happened, but maybe five, six years ago, Tesla, they had different battery cells. And if you wanted the battery with the longer lifetime, you had to buy a physical battery. Now they have the same battery. They can control the power, the battery lifetime timeline through the subscription you pay for. They actually had one use case a few years ago down in Florida —— I forget which one of the hurricane storms came in —— but for all of their customers that had the lowest, they unlocked the largest, so you can kind of get out of harm’s way. But no, you’re exactly right with that retasking ability. That’s one of the key growth opportunities as well as just operational capabilities on the floor.

Winn Hardin: [00:10:14] You know, I know that Thales has been working —— and as you mentioned, Siemens, maybe Schneider, others, right? —— and is one of the largest when it comes to AI deployment, industrial applications. But farther downstream, I’m just curious, is it too early in this process to comment? But I would love to get your opinion on this: One of the greatest growth areas for industrial hardware has been the small enterprise, small and medium enterprises who’ve been, you know, don’t always have that CapEx to be able to put bunches of money into production lines, but have labor shortages. Have capacity issues. Need new quality of precision capability. So do you see this as being detrimental? I mean, I guess an outsider might say, “well, is this going to slow hardware sales?” Or is this actually going to open up the opportunity by giving a much better ROI argument to small and medium enterprises and possibly accelerate those implementations?

Dale Hopkinson: [00:11:08] Yeah. From the growth perspective, this is where this concept, again, is … Part of my excitement when I follow this space is being able to now sell into smaller areas that that could not afford buying a physical device upfront. Think of something like, you know, they can now lease a device or you can get different business models, whether it’s a usage base or subscription terms. So while it does have the risk, of course, of disrupting hardware sales, the opportunity that opens now —— to going downstream, to selling into smaller markets, different countries, however you define regions, etc. —— that opportunity now is much greater. And the cost of setting up a pilot, a demo … So, for one example, with Siemens, right? One of their challenges that they wanted to get to was a demo model —— and I can share this, this is public on our website —— they wanted to get to sort of a demo version of their software. One of their challenges was they had to maintain two versions of software, which was quite cumbersome —— for anyone who’s felt that pain, and I’m sure there’s some listening that they’re nodding their head and putting their head down now! Siemens had the ability now, leveraging the concepts of monetization, to get to one software. And by turning things on and off and defining their time limit or terms, they can go into markets at a much lower cost. So it lowers the customer acquisition cost. You don’t have to necessarily get the full set up, a full team of engineers on site and installers and integrators to do everything.

Dale Hopkinson: [00:12:48] So this definitely has the ability to open markets. And we’re seeing a number of companies with that exact motion going downstream. And again, something we’ve seen with medical years ago —— I’ve worked with medical companies where their challenge was the small clinics. The banks didn’t know how to assess their risk for a small clinic wanting to buy such a large array of X-ray devices. So getting to the same model and getting them out of that CapEx sort of challenge where they can lease the machine and you’re making it up on the usage side, the subscription, and you’re really beginning to think now about the customer lifetime value and not just that one-time sale. So the risk for sure is there, shifting with this model, decoupling the software from the hardware, but the opportunity is much greater, right? Which is why Bain and others are saying that the revenue that’s going to be created from AI and software, and monetizing both, as well as data is very large. And we’re actually seeing that the profit pools are shifting. It’s becoming very challenging to compete on hardware. Now, there’s a little false sort of aspect in the market to compete on hardware with the war and the things that are going on. You may be able to [modify] pricing, but that’s temporary. Hopefully, you’re really going to have to learn to compete on software and digital innovation and how you monetize that.

Winn Hardin: [00:14:09] You know, it’s interesting to me, and obviously we’re at a pivot point, right? When we talk about more dependency on the software side, what were the impacts on hardware? For years, I’ve wondered, when will robot as a service, you know, when will it become more of a thing? And, we talked about bringing it into new markets. I mean, we’re not just talking about vertical industries in the United States and developed countries, but also theoretically talking about emerging markets. Folks who are who are earlier in the adoption curve of automation technology. And I think that cost compartmentalization you’re talking about, I could just see that being a major driver for many more entrants into the marketplace. So we’re definitely going to have to keep an eye on that and revisit that in a future talk. When we say, “okay, well, how is this coming out in reality?” So I look forward to reflecting back on this next year, perhaps at Automate.

Dale Hopkinson: [00:15:06] Yeah, absolutely. And just quickly on that: I was in Asia, Southeast Asia a few months ago, in Singapore. And with these countries having access to hardware and labor cheaper,  this is one of the topics that’s coming up. That they now have the ability to enter markets, global markets, with this same exact model. So, yeah, that’s something I’m tracking —— interesting you brought it up. So yeah, let’s definitely put it on our list and I look forward to revisiting it, hopefully a year from now.

Winn Hardin: [00:15:37] You make an excellent point. When you think about the Asian … We think about the per-unit cost for robotic and many automation solutions going down significantly. I mean, it’s like it’s not just AI and the retasking that poses a pivot point for hardware sales. You know, as this becomes a mature technology as robotic work cells and others, we would expect the per-unit price to come down as global competition goes up. This is the normal market evolution. So it was ultimately unavoidable. The beauty is now, possibly by using some of the things that we’re going to talk about here —— that you’re going to share with us about how to monetize it and the path forward —— that, you know, there’s a way to reclaim that revenue while continuing to better service your customer segment. And so I just find this all super exciting. So let’s go a little bit deeper. Can you give us some advice that you would give to clients about how, okay, now you’ve got this AI platform. It’s super powerful. How do you monetize it? How do you sell that to customers and make them understand?

Dale Hopkinson: [00:16:38] Yeah, absolutely. I’m laughing because this scenario … I said a few months ago as a company acquired an AI platform and they said, “yeah, this fits perfectly, but we have no idea how to monetize it.” And I’m like, “I want you to pay for it again, you know, nine figures like, okay, let’s get started.” So I think the first thing is, in this shift, it is strategic, right? It’s a strategic mindset and a strategic muscle. You have to really think of software and AI as a core asset. And what I mean by that is you want to have PNL leadership for the revenue that’s coming out of this, ideally for the entire portfolio. A few companies do this well. Some of the products may not fit in one clean portfolio. So how will you break your products down or portfolio down. You want to get PNL leadership for that. You want to have dedicated staff around that. So that’s the first thing that I’ve seen once companies are really strategically thinking this way. Anything that happens after the monetization, the projects, the execution … This is the last thing. I’ve seen the other way where companies, they’ve shifted to the execution. They’ve had great plans, put in a great tech stack, but it just wasn’t a strategic  muscle or strategic discipline. And when someone left the company, things changed. So you really want to first start there on the strategic level, have PNL leadership for the software portfolio, and the software revenue.

Winn Hardin: [00:18:09] I really want to hear … I mean, what you said: you mentioned changing personnel in different directions. If you don’t have high commitment, if it’s not part of an organization, then it’s harder to develop roadmaps and to stick to those roadmaps.

Dale Hopkinson: [00:18:21] Yes.

Winn Hardin: [00:18:22] But I’m also going to guess that there’s other implications, such as to be able to easily identify revenue from these new activities.

Dale Hopkinson: [00:18:28] Oh, absolutely.

Winn Hardin: [00:18:29] So that we can get C-suite attention. So can you tell us a little bit more about why it’s critical? I mean, you might not think that an organizational plan is such a critical component, but it is. So can you tell us a little bit more about the specific reasons?

Dale Hopkinson: [00:18:42] Yeah, absolutely. I mean, there’s two ways. If you’re a public company and you want to begin breaking out software revenue, recurring revenue, the analysts and your investors will appreciate that your stock will grow and it will go up and to the right in the good direction, right in the midst of all the challenges that you may have. So that’s certainly one way, right? You can prove the level of investments. You can prove the ROI easier. On the internal side, the other side, [where you’re] asking for investments. If you can show now by breaking out recurring revenue, you’re … I mean, of course, deferred revenue is the future growth of the business, the life of the business. The more you can attribute that internally, asking for investments and monetization projects and new innovation becomes easier to prove those business cases. And of course, in turn, like you said, the roadmap lives on. It’s not dependent on a team or individuals. It’s the roadmap and the growth of the business. As part of my role in cloud services, I actually had the opportunity —— sometimes burden and headache! —— of pivoting our business from on prem to SaaS, right? So going through exactly that, when we were the perpetual one-time sales with maintenance that looks very similar to manufacturing and now shifting the business to recurring revenue, right? So that it was easier for us reporting up. We can show what our current investment actually really means for the growth of the business, the health of the business, and why it makes sense. We got additional investments and we eventually grew and became very successful there. So those are the critical reasons you want to do this, but especially if you’re a public company, being able to show an annual recurring revenue. Wall Street and the street really likes that. So those are some of the key reasons you really want to break it out. And decide whenever you want to do reported in your earnings. But the internal benefits are tremendous.

Winn Hardin: [00:20:40] That makes all the sense in the world. You’re right. I couldn’t agree more with investors loving IRR. So I mean, you can predict project sales growth based on what you think market adoptions are going to be. Macro trends are kind of locked in unless there’s going to be a major investment or big business decision to change course.

Dale Hopkinson: [00:20:59] So absolutely, absolutely. And, actually, there’s another point there. I mean, if you’re on a certain size of M&A acquisitions, right, that can also improve valuation of the company. So IRR will be valued much higher than just one time sales.

Winn Hardin: [00:21:17] I couldn’t agree more. Now I interrupted earlier just to dive deeper into that importance of the organizational changes. But you had some more.

Dale Hopkinson: [00:21:24] Yeah. So the second thing I would see there, really digging deeper and advising companies, is once you have that level of structure, you have to rethink the business model right now. Now with software and AI, you mentioned a few things earlier. It’s recurring. You’re delivering ongoing value, right? You’re updating models, you’re optimizing models. So there’s recurring activity. You need to make sure that the business model around that does two things: It is trying to capture the recurring growth, the customer lifetime value, as we call it. You think less of the one-time sale and now,”how much revenue do I make over the course of contracts and renewals with a particular customer?” And that also helps you to deal with the economics that come in with software, the maintenance. So is it not just one version and the next version is coming with a new hardware —— you have maintenance. Now you need to have ongoing maintenance. You have bug fixes, you’ll have support cases, you’ll have new innovation, right? You want to make sure that’s not going in the product for free. Because this is where you get the cost and not the benefit. So you want to rethink those business models, the business models that typically make sense. Everyone starts with subscriptions. When it comes to subscriptions, I’ve seen it two ways. Some companies take the first step where they have all of the features, all of the benefits, and it’s a flat-rate subscription.

Dale Hopkinson: [00:22:45] I think that’s going away a lot more these days as companies add AI and there’s so many things that’re rate-able. You want to get quickly to the point of subscription with different tiers. So here’s where you can get to some of that, the retasking and entering new markets, as we mentioned earlier. When you have a subscription with different tiers —— which just means different features on and off, different usage limits, and different bundles —— you can decide now to target different markets, target different regions, target even different verticals. You can even subsegment industry. So you can get really granular on your strategy. The other typical model is usage-based. Now that everyone is adding AI, usage-based models, whether it’s prepaid or postpaid, are becoming fairly acceptable. You do also want to think about this. What in the software can be fixed-value where you have fixed subscriptions, whether subscriptions are based on a per site, per device, or just a term. Versus now what what’s rate-able. Where can we add an incremental value through usage that also makes sense to our customer base and the offer? The truth is that there still is perpetual, right? I know there are many groups who say, you know, perpetual is dead. It’s minimizing, from my point of view, of course, it’s going away because you have recurring value, recurring maintenance you’re offering.

Dale Hopkinson: [00:24:09] But in this space, manufacturing perpetual is going to be around for quite some time, but the emphasis is not there. So, in rethinking those business models, you want to make sure you can manage all of those business models, which leads to the ultimate point, of sort of advising companies, is, you know, we kind of say, “you want to fix the plumbing.” So in managing the business models and making sure there’s this level of flexibility to change, as I just mentioned, it really is a concept of modern monetization. There are three key things you want to pay attention to: how that software is licensed and how the license is being controlled. If you’re doing a usage base, how are you metering or capturing the usage for billing purposes? And then, ultimately, how you’re protecting your AI and/or software, especially if it’s on the edge. You want to make sure no one can compromise that and manipulate that. So, yes, it’s really those three key things. It’s one, making sure there’s PNL leadership, that you have the right team with a vision that’s thinking through the business models, and making sure you have the right architecture to support those decisions as they come.

Winn Hardin: [00:25:19] Do you have any more thoughts on either industry or customer profile types or application types that may have more, may want more of a flat-rate subscription model versus maybe a subscription giving you a certain amount of usage, and then more if you exceed. Are there certain methods that are better for, say, small enterprises but others that are more attractive to the large enterprises? Any more guidance there about how they would design that?

Dale Hopkinson: [00:25:47] Yeah. You know, it’s a tough question because things are changing so much. If it was three years ago, I can give better guidance. But with the introduction of AI and the speed of change, everyone is becoming comfortable with change. So I think that the biggest guidance here, honestly, is —— and we’re working through, I’m working through a few projects now where the guidance is that the best practice may change tomorrow.  And to one of your points earlier, we may have a new entrant in the market that has some economics or risk profile that allows them to have a very small price. You now need to figure out how to compete with that, right? So the demand of the day is being able to change in a very flexible and agile way without disrupting engineering. So the key becomes, “how do you make sure from the monetization, the modern monetization side, that you have the right capabilities that allows you to change the feature tiering, change the usage limits, change the business model, without requiring an engineering team?” But there are a few trends. When it comes to any kind of AI offerings that you’re doing per transaction, per scanning —— any kind of per metric where your value is delivered on a per x —— usage is usually more aligned there. In the enterprise side, they are okay with usage. They just don’t want surprise billing. And I was both an offender and a victim of this in my product management life with enterprise.

Dale Hopkinson: [00:27:19] If you do have billing or usage base, you do want to think about bundling an enterprise license agreement, giving them access to X parts of the portfolio on a certain usage base, spelling out very clearly in the contracts, what are those usage limits and making it transparent to them. And I think also with enterprise license, allowing the customer to change what part of the product they can substitute or switch, right? Because they may not know all of the use cases upfront, but being able to change from one product to another during the life of their contract. I’m seeing enterprise license agreement with software becoming very advantageous, tying that to usage. But other than pure AI and where it’s like an AI agent or AI is doing something on a per X basis, it’s hard to say now where flat rate and usage stands in. But the other trend I’m seeing, though, that’s sort of creeping up in manufacturing as well, it’s becoming very acceptable to have a hybrid price. You have both. So in a portfolio, the core platform is a subscription tier. And then you have portions with usage. This is actually one of the models that I love to see. What we call multi-access pricing. Within one tier, you have multiple levers that drive upsell, right? And not only just someone needing a new device.

Winn Hardin: [00:28:44] Is helping customers answer these types of questions, is this one of the things you do during a consultative meeting from Thales’ point of view? Or are there other recommendations, like check out the AI4 conference series or anything else where CIOs or CAIs might be able to learn more and get their head around? Because what we’re talking about, you know, depending on what your service group is, I mean, you’re talking about a fairly nice sized room of MBAs to be able to figure out the perfect tiering structures.

Dale Hopkinson: [00:29:17] Yeah.

Winn Hardin: [00:29:18] Is this something you help consult on at Thales, or would this be beyond your scope?

Dale Hopkinson: [00:29:23] So it is beyond the scope, from when it comes to the pure pricing and packaging, figuring out the value to metric to number. We typically work with a few companies there. One we typically work with is Simon-Kucher. They’re one of the fair ones in the business, I think; I’m not really a promoter for them, just saying we work with them. Myself because I’ve sat in the rooms with the MBAs so much and I’ve solved this, and I’ve owned the problem. I have some unique insight that from time to time, if it’s a customer that has already agreed to do business with us, I’ll help them based on my experience. But from this pure consultation side, it’s outside the scope. We like to stick on the side of: after this is defined, how do we take action on that with technology to make sure you’re implementing it in a very scalable way for the business?

Winn Hardin: [00:30:17] You know, you said something earlier when we were talking about the speed of change. And we talked about maintenance and we’ve talked about making sure that we have that built into a business model so that as an OEM or a provider of an AI platform, we don’t end up taking it in the shorts. Right? Because of being so successful, you don’t want to be a victim of your own success. But it seems like, while these are significant changes, especially in industrial and manufacturing environments, asking customers to consider new service-agreement levels, contracts, and how we’re going to operate like that, it also seems to bring in this reoccurring relationship usage thing. Does it open the door for us to be more partners for that AI platform provider, to be more of a partner with the customer, which isn’t just awesome PR, right? Our models are constantly changing, so it seems like we can do one- or two-year iterations on a general manufacturing software platform or hardware platform, and not be out of sync with the industry in the AI space. Those changes are happening daily, if not monthly. So it seems like this not only is a good way for AI service providers to organize themselves for revenue success and to limit liability, but also to be in constant communication about how their models need to change, how their platform can improve to capture even more, or even lead to more ideas about how to segment those service offerings to capture larger markets. So are there any other operational changes that you recommend to companies? That sounds like that resonates with you, but is there anything else that we haven’t talked about?

Dale Hopkinson: [00:32:03] No, you’re right. The operational change is on the customer success side. You now need to think about, how does your customer get and continue to be successful with your offer? When you’re in this field or this stage, your offer really begins to evolve to a platform. So, exactly as you mentioned, you’d begin to go from just a provider to a partner. You’re a solution provider, but you’re really partnering with your customer to ensure their success. So now you have to think through a few things —— and they’re not too different —— of how many companies have done it before. They’ve just relied on many integrators and other third parties to do it. And some of that you bring in-house. But customer success? You really need to think through that. How successful are your customers? How are you getting feedback directly? I’m working with many companies now as they make this shift. They want to own the customer relationship. They’re still relying on the channel and resellers to sell their offers, but they want to own the customer relationship. So one of the big things we see here shift operationally is first helping the channel really being able to manage what we call the entitlement, the commercial access. “I purchased the premium package. I should have the ability to use the premium features.” So being able to and allowing your channel to manage the initial onboarding, the initial entitlement and set up of the offer. But, of course, as you mentioned, there’s ongoing things, there’s a smaller SKU that may add additional value. Individually, your channel may not want to sell just one small thing, but for you in the platform selling one AI model on top, it creates more lock in. So companies are now looking at how do they own the customer’s relationship?

Dale Hopkinson: [00:33:53] So they need to understand not just what was sold, but who was it sold to? What’s the current state of it? How much is it being used? And wanting to see that data, typically into the CRM or close to where their sellers are, that’s one of the operational changes. So two things: their customer success and renewals. You have to think through now, how do you manage renewals? Do you have a renewal team? Is it the same salesperson that does both land and expand? That’s for sure an operational piece, a big operational piece, we didn’t touch on that. That’s probably another two sessions by itself! How do you even come to this, if you’re moving to the subscription side. What does that look like for your salespeople? How is it different from the first sale to renewal? Is that the same team? So there’s a lot of nuanced topics that you have to think through from the operational side. SKUs. I mean, there’s something even as simple as SKUs. You need to have the right SKUs in the system. Even with a subscription —— there’s a portion of a subscription that you can recognize upfront for the revenue, but the majority of it is ratable over time. And you need to define that, the difference between the maintenance and the value of the software. So, I’ve owned that. I’ve sat in those rooms and banged my head with finance, figuring out the right percentages. So I understand all of those nuances. But there are operational changes in finance, product operations, and customer success.

Winn Hardin: [00:35:21] I suspect the output of many of those meetings was a fragile truce. On one side.

Dale Hopkinson: [00:35:30] Those are the things that make good friends! And you look back now on those stories and you have a strong bond because of the big fights that you were in.

Winn Hardin: [00:35:38] Because we didn’t kill each other. Nothing brings you closer.

Dale Hopkinson: [00:35:41] Exactly. Yeah, exactly.

Winn Hardin: [00:35:45] All right. So I think we’ve talked a lot about how folks should think about the need to monetize their AI platforms, about organizational changes, first steps forward. Is there one key takeaway that you want to leave our audience with about where we are today or where we’re going?

Dale Hopkinson: [00:36:04] Yeah. So for me, where this is exciting —— I’ve seen this movie before. And I get to leverage my compound expertise, and helping and owning the problem in this new space, which has always been really exciting to me. One, this is not going anywhere but up and to the right. So there are really some key things that you have to think through about modern monetization, as I call it. I mean, if you Google “AI monetization,” it is one of the top stories right now. If you’re a company that has AI in your offer and you don’t have an AI monetization plan in your public [offering], your stock will be punished. This is a board-level conversation. So one of the key takeaways is, you have to make sure that you have the right monetization infrastructure. What I mean by “right” is one that enables you to be flexible. Which is already a tenet that software-defined automations offer anyway. And by flexible, I mean you want to be able to give your go-to-market teams the ability to change the pricing, change the packaging, change the offer without having to call engineering. This is a very fluid time in the market. It’s very early days, so you have the opportunity to gain a lot of market share by having this flexibility.

Dale Hopkinson: [00:37:23] Another key thing you want to look at is really how you protect your offers. If you have AI that you’re deploying on the edge, you have to at least have the conversation of, you know, “how are we securing the model? How do we make sure this model is going to deliver the same output that we trained it to do? Jow do we make sure our training data is not manipulated?” You want to make sure you have the appropriate conversations on protecting your AI software and model. That’s just the protection for the integrity side. There’s also the IP side. You’re investing in software, you’re investing in AI for differentiation, for competitive reasons. You want to make sure that it doesn’t become a liability. You want it to be maintained as an asset. And protecting you does just that. The honest consideration is really productizing. What what are you productizing? I’ve seen some companies who just try to jump on the bandwagon: “hey, we have software,” and they slap a price on it. You have to first make sure that what you productize is really delivering real value that your customers are willing to pay for. You don’t want to have a layer of things that they have to pay for. It makes your device not usable. That’s very annoying. And the other side of this sword is, software-defined is making it easier for switching.

Dale Hopkinson: [00:38:43] So, if you get some of these things wrong, it becomes a little easier for your customers to switch. So you really do want to pay attention to whether your product has real value to the customers. So, the first thing I’d say is, yeah, you really want to design for flexibility. Now, back to the question —— and I can’t really give a definitive answer —— what’s best for subscription or best for usage? It’s changing. I think one of the exciting things is, there’s a number of startups, a number of even incumbents who are moving to software-only or having software organizations. And, you know, they roadmap out how long they can run software as a loss leader. So they’re changing prices, changing packaging. To me that’s the need of the day. And I see that across not just manufacturing, but across everywhere. Salesforce, for example: in the past five months, past 18 to 19 months, they’ve changed their pricing about four to five times, trying to figure out the right mix of usage and flat rate, because that’s just what the market is doing. The key is to make sure you design your monetization ability, and the main flexibility is making sure you don’t need your engineering team to manage entitlements and limits hard coding in the application.

Winn Hardin: [00:40:00] Absolutely brilliant. All good advice, 100%. Maybe the next time we can talk about how customer data sets might be feeding back into models, or if there’s any options. Because  it seems like that would be part of an SLA agreement. How do customers feel about that? I mean, once upon a time, nothing went past the firewall. Well, that firewall crumbled about 12 years ago, probably. So, are customers willing to let their data be part of the enhanced solution set. In general, we can go deeper.

Dale Hopkinson: [00:40:34] So yes. So we can go deep in the future, but I don’t know  where the clear delineation is yet. But I’ve seen some where they’re okay with some data sets being part of what can be trained to improve the model. And for some customers it’s a no go. And that’s where you still typically see the edge AI being deployed, running all locally. You need to figure out how to train and improve your data and send me the new model when you do it. But it’s an area I’m looking at. I’ve seen it both ways. I’m just not fully clear yet: what’s the delineator where some companies are okay, but some aren’t. But I suspect as there’s more AI, maybe with a bit more of a regulatory touch as well, and people get a bit more comfortable, I suspect we’ll see more and more of it. But I still think there will always be cases where it’s a no go —— send me the model, I’m going to run it in my private cloud or on my physical premises.

Winn Hardin: [00:41:31] Yeah, I would see that being the norm. Absolutely. And, yeah, before you overcome those concerns on data protection and proprietary competitive advantage, I think our data, our data handling, and our regulatory environments, and our reporting is going to have to evolve.

Dale Hopkinson: [00:41:50] Oh yeah.

Winn Hardin: [00:41:52] Confidence from the customer, right?

Dale Hopkinson: [00:41:54] Yes, indeed.

Winn Hardin: [00:41:55] Well, Dale, it’s been brilliant speaking with you today. Again, I really appreciate the time.

Dale Hopkinson: [00:41:59] Absolutely.

Winn Hardin: [00:42:01] I’m sure the audience has gotten a lot out of this. And this will not be the last time we will talk about this, because AI doesn’t just appear magically. And once you make it, it never stops growing. So for folks out in our audience, if you have any questions for Dale, I would encourage you to make sure you go to Thales.com. You can reach out to Dale directly on LinkedIn. If you’ve got a question for him, feel free to go to manufacturing-matters.com to see all of our past episodes, and you can send us a question. We’ll make sure we get it to Dale and get a quick answer for you. In the meantime, thanks so much for joining us today.

Dale Hopkinson: [00:42:37] My pleasure.

Winn Hardin: [00:42:38] We’ll talk more. Safe travels, Dale. It’s been a pleasure, brother.

Dale Hopkinson: [00:42:42] Yes, indeed. Thanks, Winn.