Can electricity grids built for an earlier era support AI data centers, expanding manufacturing, electric vehicles, severe weather, and rising customer expectations at the same time?
In this episode of Tech Talks Daily, I speak with Mark Hollis, utility executive advisor at SAP, about the pressures reshaping the utility industry and the practical choices available to leaders today. Mark spent over 15 years at Duke Energy before moving to SAP, where his work gives him visibility into utility organizations across North America.
Mark describes a combination of load growth, disruption, long construction timelines, and regulation. Data centers are receiving much of the attention because of the electricity required by AI, but he argues that they are only part of the story. Manufacturing is returning to parts of North America, transport and heating are becoming increasingly electrified, and utilities must prepare for wildfires, hurricanes, winter storms, and other events that affect generation and delivery.
The obvious response is to produce additional electricity, but every option comes with physical and commercial limits. Wind and solar contribute to the generation mix, although output depends on conditions. Small modular nuclear reactors could support future demand, but commercial deployment takes time. Batteries can store electricity and release it later, but they do not generate the power they hold. Customer programs can also reduce pressure at busy periods, including arrangements that allow a utility to adjust connected thermostats by a few degrees.
This makes modernization a portfolio of decisions rather than a single bet. Utilities must decide how to divide capital among generation, transmission, resilience, customer systems, and new technology. The people who understand existing processes are often the same employees needed to design and implement replacements. At the same time, information technology and operational technology are becoming increasingly connected, forcing companies to reconsider how business functions share information and how technology decisions support outcomes across the enterprise.
AI creates another tension because it contributes to electricity demand while also offering utilities new ways to work. Mark says most utilities he meets are cautiously optimistic. Their questions include where to begin, whether the value is proven, how long adoption will take, and whether poor data must be fixed before useful work can start. He warns against choosing the hardest problem first or judging a business process by the forgiving standards people accept from consumer AI tools.
His advice begins with the business problem. Automating an inefficient process can make it more expensive and harder to correct. Utilities should define what they need to improve and why, establish connected data with the right business context, set clear quality requirements, and retain human review where errors could affect customers, safety, finance, or regulatory obligations.
Mark brings this to life with several utility AI use cases. AI could summarize customer interactions across field service and contact center systems, allowing the next employee to understand what happened previously. It could review billing exceptions during unusually hot or cold periods and support earlier customer communications when consumption is likely to produce a much higher bill.
He also discusses using AI to summarize lengthy rate-case rulings before approved changes enter billing systems. During outages, an AI system could help dispatch crews by considering skills, equipment, parts, certifications, location, safety, and customers with medical needs. A human dispatcher could then review and approve the recommendation rather than building the complete schedule manually.
The opportunity is real, but so are the limits. Utilities operate regulated infrastructure where reliability, safety, auditability, and public trust cannot be treated as optional features. Where should the industry begin, and which use case offers the right combination of low effort, meaningful impact, and manageable risk? Listen to the conversation and share your thoughts with me.
Useful Links
Connect with Mark Hollis
[00:00:03] Can power grids keep up with AI data centers, return in manufacturing, electric vehicles, severe weather and the air conditioning that many of us are relying on right now? Well, in today's episode, I'm going to be speaking with Mark Hollis, Utility Executive Advisor at SAP.
[00:00:26] And together, I want to talk about a power system that was built for a very different era and the decisions utilities must make now. Because Mark has spent 15 years at Duke Energy before joining SAP, which has given him a unique view from inside the industry and across utility organizations across the US.
[00:00:51] So today, I want to discuss why adding more power generation alone is not going to solve the problem. Maybe look at where wind, solar, nuclear, batteries and customer behavior out and why modernization competes with other large capital priorities out there right now.
[00:01:08] And Mark will also share a practical, some practical AI use cases in everything from customer service, billing, rate cases and even dispatching field crews without pretending that automation can rescue a broken process. Yeah, we got lots to talk about today. So enough from me. Let me introduce you to him right now. So thank you for joining me on the podcast today, Mark.
[00:01:36] Can you tell everyone listening a little about who you are and what you do? Sure, absolutely. No, thank you for having me. It's exciting to be here and a great time in the industry. So most of my career, I spent the majority in the utility industry. Before joining SAP, I spent a little over 15 years at Duke Energy, which is one of the largest utilities in North America, you know, around 9 million customers.
[00:02:01] And so in different leadership roles from compliance to strategy to customer programs and transformation. But today I've been at SAP five years this week, actually doing utility executive advisor work. And what that means is I get to see just about every SAP interaction in North America with the utility industry. And so it's it's exciting. It's rewarding.
[00:02:27] You get to see the journey that every utility across North America is is on. And they're all in different places, which makes it, you know, a little bit more exciting. But it's it's a truly unique time in the industry. So I'm excited. Well, it's a pleasure to have you join me today. And it must be quite a fascinating time for the utilities industry with the increase in demand. Before we started recording today, we were talking about the AC and the demands placed on that as the planet warms up.
[00:02:56] And what we didn't talk about, of course, is also AI and the rise of data centers that are increasing pressure on the grid. So I'm curious, can you give an overview of some of the big themes and challenges that you're seeing utilities facing today? What are you seeing out there? Yeah, AI is certainly a huge one. And and I'm not sure people are afraid. And I say that in quotations with the AI in the middle of that word. Or, you know, is it more excitement?
[00:03:24] And I think there's a combination of both data centers are certainly front and center globally. But they're in my opinion, they're only part of the story because yet new load is spiking tremendously because of AI. AI is going to help solve part of that problem. But it's also creating problems. But we're also talking about, especially in North America, the bringing back of major manufacturing, which people aren't you know, you don't hear a whole lot about that.
[00:03:51] But there's a lot of movement that's that's massive consumers of electricity happening, especially in the states. And so it's not just data centers, but that's a huge piece. And so utilities are really trying to plan for more electrification across the board from EVs to distributed energy. But then you throw natural disasters. There's tremendous wildfires going on right now. Hurricane season isn't far away.
[00:04:19] You know, right behind that will be winter with ice storms and some of the air. And so just this perfect storm and that no pun intended, but the perfect storm of of really what's facing the industry is the challenges. And I would really put them into, you know, for and it's the grid was built for a different time in a different world. Really, AI was not what it is today, even five years ago. And so the four that I would say is the load growth. That's a big one.
[00:04:49] Everybody's hearing about that. You know, how how in the world is enough power going to be generated to keep up with the demands coming on the grid? But then you throw the second one is disruption. It's it's not simply about adding more power. That's going to take time. You know, all these storms that I mentioned, you know, some of these utilities are doing instead of 20 year plans,
[00:05:13] they're doing 50 year plans now putting steel structures where they never had to put steel structures before. And so that leads to my third challenge, which is time. You know, it's not simply, oh, let's make more power. It takes time to to build generation capabilities. It takes time to transmit that across wires and poles. And so it's it's just the perfect storm of of too many things coming together. And then the fourth one is we cannot forget about the regulations.
[00:05:43] Yeah. Regulations always lag technology. Most things lag technology these days, but it's it's, you know, across 50 states and in Canada. It's just fascinating of how it has to keep up. So the simple answer is do more, better, faster and cheaper, of course. You make it sound so simple. Yeah, so simple. Because what would you say is the solution here to this increasing power demand? And is there one right now?
[00:06:10] Because a quick look online and as you said, some will say, why not just create more power with all the technology we have today? Others will say because of the scale, you need to go nuclear instead due to the scale of demand. And again, that can take up to a decade and longer to build. So how do you see this evolving? What's the solution? Yeah, it's it's fascinating. So there is more generation needed. It's necessary. And that's a global statement. That's, you know, we're seeing that across the world. But it's OK.
[00:06:39] If that takes time, how can I modernize today? What are some practical ways that I could that I could use technology? And so generation is certainly part of that. But every boardroom in the world is talking about AI. And the byproduct is these power quarters. Where is enough power for me to be positioned? So it's no longer interstate corridors or logistics corridors. It's power corridors. And so there's not one solution. It's a it's a myriad of solutions.
[00:07:07] But some of those are simply we were talking about our AC units before we came on. And it's some of them are just behavior based, not necessarily technology based. But how do you push those programs from a utility to the customer base so that they can, you know, I love my AC, but but I do allow my utility to control my thermostat, bump it up a few degrees when they need to relieve pressure on the grid. And so, you know, there's not one technology that's going to solve the problem.
[00:07:37] A lot of people like to think about, well, wind and solar, we've got to have wind and solar. And yeah, wind and solar is going to play a part, but it's it's limited production, limited predictability. You know, my even my late father said, you know, not not even in this context, the sun don't always shine and the wind don't always blow. And so, you know, it's it's easy to say wind and solar. It's just not reality to make that a full replacement for.
[00:08:04] And then the nuclear piece you talked about, you know, the I'm working with with several of these small modular reactor companies. And that's fascinating and going to be a huge part of the solution. But that's going to take time. You know, I was reading today. Some of them won't won't be commercially available till 2028, 2030, 2032. Well, that's still a few years away. So in the meantime, OK, batteries is another one. Right. And I laugh because people are like, well, we need more batteries.
[00:08:33] And I'm like, you do understand that batteries don't generate power, don't you? They store power. And so you still got to generate power to put onto the battery and then you got to discharge it. Then you got to generate power again to recharge the battery. And so it's all these physical limitations that that come up that that sometimes people, you know, that aren't as deep into the industry as some of us are don't fully see the big picture and don't understand.
[00:08:59] So it's it's really how can they employ today's technology to modernize what they can? And that goes back to some of these programs. So it's really. Use the technology you have that was not built for the conversations that are needed. Massive capital investments completed yesterday, of course, while continuously advancing your technology, but then trying to modernize at the same time. And so it's, again, a simple answer.
[00:09:30] It's a quite complex problem, but we'll get there. It's just a matter of time. And what are some of the biggest dilemmas and challenges that are maybe keeping utilities from modernizing their infrastructure? What's holding them back? Yeah, it's that's it's very interesting to me because, you know, I said that the technologies that that many of us are working with were not created to do what we want them now to do.
[00:09:57] And they weren't created, certainly to communicate with different technologies the way that we want them to today. And so, you know, most utilities are working with systems that have evolved over over decades. And so, you know, many of them are doing exactly what they were built to do 15 years ago.
[00:10:15] And so, you know, one of the biggest dilemmas that I'm seeing with every utility is how does my organization and I don't mean, I mean, the org chart, the people, like how do the business functions share information in the way in which we now know it needs to be shared when the solutions were not established to do that?
[00:10:39] And so that that could be a whole nother discussion to itself around how do we change org charts or org structures to keep up with technology? But, you know, the technology question is, you know, you have many hurdles. Regulatory change is a big one. The cost of capital. And this is a huge dilemma because you're competing with think about it this way. I need to create more power, generate more power.
[00:11:07] That cost of capital is a big spend, right? To generate more power, to transmit it is a big spend. Well, then you get down to technologies that help you do some of those things. The spend is not quite as big. But which one is the priority? It's chicken or egg discussion here. Do you need the technology that helps you move faster, be more accurate in your spend, prioritize better, or do you need to start building?
[00:11:35] And so there's a lot of that. There's a lot of resource constraints because you have the people that truly understand these processes are also the ones that you would have to pull out to help put the technology in place. And so, you know, as technologists, it's really about the value. How do we drive the value? And it has to outweigh the risk, but it also has to outweigh other priorities to get the spend where it needs to be. And so all of that is a huge challenge.
[00:12:06] And then one more, and then I'll stop for a second. You really have to think about the industry. And we talk about ITOT, and some people may or may not know what that is, but the information technology side, which I correlate to the transactional side of the business, and then the operational technology side of the business, which is, let's just say, moving of electrons. Those two different platforms and multiple platforms within those spaces are now converging more than we've seen.
[00:12:36] We've been talking about this for three decades, but it's now starting to really have to happen. And so technology providers are now fully engaged in those business outcomes. You know, how do we drive buzzwords like modernization and transformation? But these are journeys, not single projects. And so modernizing, it's no longer what's the simple business process problem that I need to solve?
[00:13:04] What's my technology adoption strategy? And not for one technology, for all of them across my enterprise. And it can feel, especially when we're talking about energy demand, that AI feels almost like both a solution and the problem. So what are you hearing from utilities about AI adoption? Are they all in? Are they hesitant, cautiously optimistic, pessimistic? What are you seeing here in general when the AI conversation comes up?
[00:13:34] Yeah, it's fascinating because I would say most are cautiously optimistic. Yeah. Some are much further along. Others are waiting to see. Every single one is on a different stage of their journey. But the questions that I get daily from utilities across North America is, well, where do I start? You know, that's a big one. Where's the real value for my business? Is it proven yet? And so they look at other utilities.
[00:14:02] Well, no other utility is bleeding edge either. And so, well, how long is it going to take? Or my data is terrible. I have to get my data clean and I have to get it usable. And that's just going to take too much time. So, you know, and then it's all these statistics about 95% of AI projects fail. I have utility leaders saying, I'm not going to take the risk and be in that 95% statistic. I'm going to wait until I'm in the 5% bucket.
[00:14:31] And I'm like, well, that's one way to think about it. And that is a path. The problem is the longer you wait, the more compounding effect this technology has on your waiting period. And so it's the answer to the questions. There's a lot of answers to these questions.
[00:14:48] But it's really, you know, to be in that 5% of successful AI, it's how do you get the human expectations to align to what the technology is actually capable of? Because what we use in our daily lives is AI at some form or fashion, and we see fantastic things, but it's not at a core business process, you know, functionality.
[00:15:16] And so failure in a core business process at 65% accuracy is not acceptable. Whereas for my vacations, okay, well, it got me started. So it's really fascinating right now of where they are. But many of them are, you know, they have employees using AI tools successfully. But then statistically, 40% of tools that they're buying are successful.
[00:15:43] And the reason I think goes back to what I said around the expectation of what people think should happen with the AI is just simply not happening. And a lot of that, I think, is we pick too hard of a problem first. You know, people are thinking, well, this is going to solve my hardest problems. Well, no, I wouldn't start there. I would start with something that builds a little bit of trust. You know, you got to do some proofing.
[00:16:09] And this goes back to a term I like to use, your technology adoption strategy. You know, it's not simply automating some processes anymore. It's really what's your strategy to integrate this into the infrastructure of your business. The leading issue of agentic AI in businesses right now is ensuring agents act with compliance guidelines. And Denodo applies guardrails across your entire data estate.
[00:16:38] By aligning your company's data infrastructure under one system, these guardrails perform consistently across your platform. So start scaling your business and start with Denodo. Simply visit Denodo.com to learn more. And maybe we have a few people listening in the world of utilities there. And I always like to try and give people listening a few valuable takeaways.
[00:17:02] So if we do have anyone listening and maybe they're contemplating AI use cases in utilities, any advice that you would offer there? Any pointers just to keep them on the right track? Maybe you've got some use cases of your own that would bring that to love. Yep, absolutely. Yeah, I can do some of both of that. So, you know, the hype cycle is phenomenal.
[00:17:23] And it's fascinating to watch the AI and these agents that people can build in the matter of some of them in minutes, some of them an hour, some over a weekend. But to me, it goes back to your strategy. And so you can't just dive into automating things that aren't automated today. If you automate a bad process today, what you're going to get is a more expensive, worse process. Yeah. Right.
[00:17:51] So you have to think about this, you know, start with the business challenge you're trying to solve rather than the technology itself. You know, is it even worth automating or putting AI to or is it not worth it? You've really got to have a strategy around this. And that's the first takeaway. You really have to think through, you know, investing and having trusted, connected data.
[00:18:16] At the end of the day, if your data is not, you know, satisfactory or clean enough or going to give you accurate answers, it really doesn't matter what technology you put on top of it. And so what we refer to as knowledge graphs, you know, does your business context overlay with your strategy that ensures that the answers AI is producing produces an answer within the context of your business? That's critical to what you do.
[00:18:45] Otherwise, you're going to get 65% accuracy, which is just not good enough. And so, you know, transformation means delivering substantially more value than ever before, but in a fundamentally different way. And it's not simply speeding up today's processes. To me, to get specific, you really need to define precise requirements, the what and why. Not the how yet, not the technology side, but what and why do you need to automate something?
[00:19:14] And then design your architectures where AI can effectively, and this goes back to the expectations of humans, your architectures have to be effective based on the expectations. And then reviewing and refining that AI output against stringent quality standards, because we've all heard about hallucinations and, you know, AI giving refunds that it should have never given and it didn't stop. And there was never a human in the loop. And it just, like I said before, you take a bad process and automate it.
[00:19:43] You just compounded the badness. Not that that's a good word to use, but you get my point. And then, of course, the security, the auditability, you know, the gap between the 5% that are successful is really their employees understand how to create value with this new technology. Instead of just giving them more tools to automate what they're already doing. And so, you know, it's the what and why first, not the how.
[00:20:12] And then it's, you know, it has to have a return on investment. And that's what the executives at these companies look at. If I'm going to spend the money, I really need a return. When am I going to get a return? And so you want to start with something that builds trust. And so some of those specific use cases that I'm seeing, you know, and utilities, like I said, are some of them are where do I start? And others have, you know, 90 different use cases identified. And so will they build all 90? Time will tell.
[00:20:41] Where will they start? That's a good question. But here's a few real use cases that we're seeing, you know, across the country. And it's really a lot of these are low effort, but high impact. Not strategic determination type things, but, you know, routine manual processes that can be automated. But you spend a lot of human time doing these.
[00:21:05] So think about, you know, a simple summary analysis for anyone engaging a customer premise, either virtually or in person. So you think about a field service team that came to my house. If I was an irate customer last week because my bill was high, it was hot, and I had to pay more money than I was hoping to. And the field service guy shows up in my house and I start yelling at him and he puts notes back into his tool.
[00:21:31] Well, the next time that I call the call center representative, that person should see what happened when the field service agent was on site at my home. AI could take those two different platforms and bring a summary analysis together for both sides of that coin. That's just a simple example. Another is what we call BPIMs, business process exceptions.
[00:21:55] And so these are during times like this when it's extraordinarily hot, your bill is going to be higher. It's just a matter of physics at that point. And so if it's going to be higher, it's going to kick out an exception because it's more than X percent high than last month or last year at this month. And so a person then has to manually process that exception.
[00:22:18] And so at different times in the year, those exceptions are based on rules that are not dynamic. They're very static. They might even be set by regulation. You know, how do we if everybody's bill is going to be high, is my bill really an exception? Or could it be automated to be released because everybody's bill is high this month? Right. So it's I can do those things. That one's a big one.
[00:22:46] And that's just one example of an exception. And then the preemptive or proactive communications around that exception. So if my bill is going to be high and I should know that I should know halfway through if if the utility knows halfway through the month that my bill is going to be extraordinarily higher, 20 percent than last month or last year this month. I should get a notice saying, hey, your bills might be high this year. Here's some ways that you might could reduce.
[00:23:14] And if you think about that, the context involved in sending that communication, either via text or social media or email, you have to have a bigger context than just the bill, because now you're tying back to those customer programs. So the AI could do this in a matter of seconds versus a human having to do it for every exception, which just takes way too much time. And so that's another one.
[00:23:42] Rate case outcomes is another one we've been asked about. You know, these utilities file their rate cases and they get hundreds of papers, you know, back in the report out of what happened through the rate case. What was the ruling, if you will, and how are the rates going to change? And so then they have to digest all of those papers. Somebody reads them all, builds the rates into the rates engine and then loads it, tests it. And then eventually it makes it into the production system when it's supposed to.
[00:24:12] But how can we expedite that process? That's another huge AI use case that we're seeing. And then one more I'll throw at you that's fascinating, too, is, you know, dispatchers. You think about this time of year when it's really hot or you get in the winter and it's really cold and people lose power because storms, because, you know, the grids overstrip, whatever the case may be.
[00:24:35] You know, how do we assign crews with the right skills, the right parts, the right equipment, the right tools, the right certifications, if you will, in a safe manner to actually go do the work the first time and not have to do multiple truck rolls. And then prioritize critical care customers. You think about nursing homes or hospitals, or you think about people that have medical, like just residential customers that have medical alerts on their account.
[00:25:03] How do we prioritize? AI could look at that and in a matter of seconds give you a dispatch schedule that could be used. Now, you may still want a human on the loop that can review it quickly and hit approved, but the AI can generate that schedule much quicker than a human can. And so those are some of the specific use cases we're seeing brought up routinely by many utilities across the country. Wow.
[00:25:31] So many great points and indeed actionable insights there. And for people listening, want to learn more about you, the work that you're doing and maybe continue the conversation we're starting here. Where can they find out more information? Yeah, absolutely. It's really LinkedIn is the best way to reach me. You know, I'm active on LinkedIn. I publish things periodically, but reach out to me.
[00:25:55] Happy to carry the conversation forward around your strategy, how you're thinking about this and help have those dialogues. So reach out, happy to interact. Awesome. Well, I will add links to everything that you mentioned there as well. And so many big talking points. Now, I'd love to hear back from people listening in the industry there. What are you doing? What insights have you got to share? What's working? What isn't? How are you overcoming the challenges? Please feedback to me over at techtalksnetwork.com.
[00:26:23] And Mark, it'd be great to get you back on next year and see how this story is unfolding. I think there's a lot more we can talk about too. So thanks for starting this conversation today. Yeah, it's been great. Thank you, Neil. Very much. I think Mark's argument today leaves utilities with a difficult but very useful reality. And that is there isn't a single technology waiting to rescue the grid. New generation matters, but it takes time to build and transmit.
[00:26:52] Yeah, batteries can store electricity, but somebody still has to generate it. AI can improve decisions, communications, and field operations. But only when a business starts with a worthwhile problem, connected data, realistic expectations, and human review. But what really interested me today was the need to modernize physical infrastructure, enterprise technology, organizational structures, and customer behavior.
[00:27:20] All these things need modernizing together. So thank you to Mark for sharing his experience today. You can connect with Mark on LinkedIn. But over to you. Which part of electricity system should utilities modernize first? Let me know. TechTalksNetwork.com I'll be back again tomorrow with another guest. But thank you, as always. And a big thank you to Mark for bringing this to life. Speaking with you tomorrow. Bye for now.