The Energy Gang at Distributech 2024 in Orlando

Hello and welcome to the Energy Gang with a special edition from the distributec Conference in Orlando. I'm Ed Crookes. Distributech is a big event for the electricity transmission and distribution industry. In fact, it's the biggest in North America. Being here is a fantastic opportunity to talk to many of the leading figures from companies that provide technology for moving and managing electricity, and from the companies that use that technology, including particularly utilities. Later on in this episode, you can hear my conversations with Anthony Allah, who's the head of Hitachi Energy in North America, with Tom Dietrich of Itron, with Quinn Nakayama from the utility pge, and with Zach Kass, who's a futurist who until last year was an executive at OpenAI. As you might imagine, AI has been a hot topic at this event. It's been talked about both as a source of new demand for electricity and as a tool for managing the new strains on the grid. To discuss what these latest breakthroughs in AI mean for energy, I spoke first with Hussein Shell, who's the chief technologist for aws Amazon Web Services. Hussain, thanks so much for joining us today.

Thank you for having me. I'm excited.

So AI does seem to be the hot topic already. I've just come from the plenary session, the opening keynote speeches in the main hall. AI was a consistent theme running through everything that everyone was talking about there. When you think about AWS's approach to AI in the energy industry for power and utilities, what are you offering people?

Absolutely. So it's really exciting times. It's definitely the topic of the moment and I believe, and the company really believes that this is going to be transformative technology for the next decades. Our approach really around making generative AI practical, cutting through the hype and helping our customers do the work that they want to do with it. Right. And trying to find something that is secure, flexible and capable of doing the things that they want to do. And all of those are available through our services in the top stack.

So AI is a subject we've talked about quite a bit on the Energy Gang podcast and we've had quite a few people come on and talk about the possibilities that technology opens up. And very much that's a theme here at DistribuTech. And there's a lot of excitement for the expression. Game changer, I think has been used about three or four times already in my hearing. You've just been talking about the exciting potential that you see. I have tended personally to be a bit of a skeptic. I have felt like there's A lot of excitement around sort of what you might call the latest iteration of AI in terms of large language models and so on, which have clearly got enormous implications for language based industries, essentially for journalism and the law and teaching and so on. I can see really where these new AI technologies make a massive difference to those. What I feel sceptical about is their application to energy where it seems like maybe some of the use cases are not so well demonstrated yet. And as I say, a lot of talk about the tremendous potential in principle, not so much really concrete demonstration of what's actually being done and what's really being changed on the ground right now by AI. Am I wrong to have that impression? Do you think it is changing a lot already or is this more kind of something that's still for the future and something that people are kind of speculating about and looking forward to rather than actually demonstrating right now?

I think we're in the early stages, right? As I said, generative AI as a discipline has been around for a while, but from a technology and enablement and the ability to use it at this large scale is fairly new, right? With the advances of ChatGPT that kind of put it out in the market and expose people to the possibilities. But to be honest with you, the year 2023 has been a lot of POCs and pilots and people are trying to understand what they can do with this. Just like as any new technology, it takes time a little bit to adapt to it and adopt it. And so I would say 2023 has been a significant amount of piloting and POCing and trying different things and different ideas, working with partners like Accenture, they use CodeWhisperer, which is a gen AI based code suggestion system that we have. They've seen 30% more productivity from their engineering developers. So we're seeing some of those cases. Right? Other customers are using it around data. Like you said, it's very text specific and also there's multimodular, right? With video and voice etc. And energy customers in energy industry is rich with data and documents that we don't know what to do with and where, what information and what insights are there. So some of the use cases that we're looking at are looking at this corpus of data sets and documents that have just been sitting there in drives for decades and trying to glean insights from it and see what we can do with it and what our customers can actually find value. Right? But I believe 2024 is going to see a significant scale of deployment to productions applications and solutions. 2023, mostly POC.

Right, got it. And POC then proof of concept.

Proof of concepts, yes.

Yeah. So as you say, so when people are moving, then you think in 2024 to actually deploying these applications practically. And you talk about in energy sort of going through the vast reams of text and data that are out there that nobody ever looks at at the moment. So this is what typically regulatory filings, things like that, that.

But also like in the case of oil and gas, for example, there's a lot of drilling reports, a lot of production reports, there's a lot of content that is generated that simply is forgotten after a project is done. And a lot of insights and data driven decisions are buried within these PowerPoints or PDFs or Word documents and people are starting to find it. You mentioned also law, we're seeing customers looking at warranties and looking at documents that could potentially allow them to have profit recovery use cases. And so we're seeing a lot of that. But yes, absolutely, that's very interesting.

And that's something you think then we should really be looking out for this year to see those practical implications taking effect.

Absolutely, absolutely.

So something else I often think about in this context of AI is the terminology seems to be very significant. And the terminology goes through these kind of fashions and things become buzzwords and they go into fashion and go out of fashion. So now it's fashionable to talk about AI. It feels like people have been talking about AI in the context of, of business, in general, industry in general, energy in particular, for at least a decade, maybe 15 years or so. It's often been known as machine learning there. And people have used the term AI and machine learning pretty well interchangeably. But now it's being called AI. And people don't seem to talk about machine learning so much because AI is. Exactly. And people talk about generative AI as being the kind of the new, new thing. And that's the whole thing that people are excited about. What are the real differences here? Is what we're talking about fundamentally different from the machine learning that people have been talking about for many years now, or is it just essentially the same thing being rebranded?

Yeah, well, it's really different disciplines when it comes to using data and technology to glean insights into the information that is provided. Machine learning is traditional machine learning. You feed it data, you train it, it comes back with learning information from that data set. AI is when it learns on its own. And there's different reinforcement learning techniques, there's very different disciplines, generative AI in particular is using large language models and large content model to generate new text. Simply put, right. We could go for days describing those different disciplines, but at least for today, we'll talk about that. Now, what's interesting, like you said, is that in our engagements and a lot of times when we sit down with our customers and work backwards from their problems, 40% of the problems that they're trying to solve are easily solvable by existing machine learning models and machine learning algorithms that have been used for decades. Right. It is our job as technologists to sit down with our customers and explain the differences and explain where one works better than the other. And so, you know, we get asked by many of our executive boards from customers on coming to them and explaining to them the differences, helping them kind of ideate new ideas and use cases that we can use, and then we can target each one of those use cases with the proper technology and the proper implementation of machine learning and artificial intelligence.

So for instance, one of the uses people often seem to talk about is using data to optimize the performance of a power plant. Let's say take an energy industry example. That's something that's really a machine learning job. Is it not really what you call something for generative AI, correct?

Traditionally, yes. You'd feed a lot of sensor data from a plant, you know, then any kind of machinery that generates a significant amount of inputs and data and temperatures and pressures, etc. You feed it into a model, the model learns from it, and then provides predictions and inferences on that data. Typically, yes.

And when people, for instance, talk about grid optimization as well and some of the increasingly complex problems that people are encountering with managing grids nowadays, that's also what again, that's really a machine learning.

Well, it's complicated because I think you can merge different disciplines together to create the right solution. Right. So you can use machine learning for the typical sensor data and ingestion data that is typically used in the past. Learn and educate the model and train the model to give you the insights. But you can also merge it with new data sets, synthetic data. We're seeing, you know, some of the use cases that is blending physics based model with statistical and data based models. And so you're seeing that world merging and becoming very grayish when it comes to that. But you know, I'm really, really looking forward to see what customers are going to do and how we can use all of these disciplines that are available today for our customers to come up with these solutions.

Oh, that's really interesting. Thanks very much. And that's really helpful in clarifying those terms. Thanks. So when you think longer term about the potential of AI in energy come back to distribute in 5 years or 10 years, 2029, 2034, what are people going to be talking about then? I mean, what is that longer term potential and how do you think ultimately AI will change the electricity industry in

particular, I'm hoping in five years we can see a lot of these use cases come to fruition. We see a lot of business impact that's driving better energy efficiency, better grid optimization scenarios, and really see the fruit of the experimentation that's happening now in the industry to be coming true. And we're going to see a lot of different implementations of these large language models and I'm really, really hoping that customers are going to see the benefit in our partners and continue to grow that business for them and see a difference in the lives to their customers, but also their operations, their existing processes.

So you mentioned grid integration. The other aspect of AI that's been much discussed already here at distributek is the question of the increasing load that's going to be created by all the new data centers that are going to be delivering AI applications in the future. There was an executive from Duke Energy just speaking in the main hall saying that it used to be a big deal for them when they had someone saying we'd like to add 10 megawatts or 20 megawatts of load and now they've got a number of data centers talking about we'd like to add a whole gigawatt of load. It's clearly something that is making a big difference to projections of electricity, electricity demand in the future. And I guess some of that demand is coming from you at aws. How do you manage that in particular, given that you have very demanding commitments in terms of reducing emissions, relying on renewable energy, how do you square those two things away together and manage to meet that increased demand, procure that extra power you're going to need while also reducing emissions?

I'm glad you mentioned Duke. Duke is one of our partners. We've been working with them. There's a significant partnership on some of the grid optimization work. We're helping them build cloud native technology based suite of applications to really look at things that are grid enabling technologies. Right. In the meantime, before these large upscale and updated efforts that are going on in the infrastructure. So for example, we're working with them on leveraging not only machine learning, but also traditional high performance computing technologies that was available in aws. To do simulations into power flow simulations and look at how they can optimize topology optimization, et cetera, et cetera. There's a significant amount of different solutions that are available. Part of the technologies to look at, like I said, topology optimization, advanced power flow simulation, and really trying to see where can you take already on more capacity without necessarily burdening the grid that exists today in the hope of, you know, improving in the future. And so we realize that this takes partnerships with our partners, with technology providers like ourselves, but also with regulators to provide those kind of incentives for customers to do.

Yeah, and those grid enhancing technology is obviously particularly important given how hard it is to build new infrastructure in most of the world, I guess you'd say. And power transmission in particular is one of those things that seems to be really hard to get done. And clearly, given all these new demands that are coming on the grid, both on the demand side and because of increased variable renewables, more wind and solar on the grid and so on.

Exactly.

The industry is facing a whole lot of new challenges. It needs a lot more transmission. It's hard to build that transmission. So to the extent that you can do anything to upgrade the grid, improve the network without actually adding physically more capacity, that's really important.

Yeah, it's absolutely important. And there's a lot of things that we're doing internally as well. For example, we're doing a lot of work around our water consumption and cooling in our data centers. Right.

Again. Which has been a subject a lot of people are getting increasingly concerned about. Right, yeah, right.

And with our commitment to be water positive by 2030, we'll be giving more water out to the community that we consume in our data centers. We are, as you know, one of the largest renewable energy providers. We will be 100% planned in 2025. We're way ahead. We're 95% at the moment. We can help our customers decrease their carbon footprint by using our services up to close to 80% and 95% once we're 100% renewable. So we're doing a lot of things to encourage. Our hope is that we compete in that space. We're not going to be able to solve, unfortunately, climate change by ourselves, but we want our competitors and others to also innovate with us, and we want to push that limit with what we're doing. A lot of our climate pledges are ambitious goals, net zero by 2040. And that's going to take a lot of innovation, a lot of technology.

So you would be absolutely confident then that you will be able to get to net zero emissions by 2040. You're still totally on track for that?

Absolutely, we are. We're seeing results. Like I said, our carbon intensity reduced by 7% in 23. We are on our way to be 100% renewable by 2025. We're working on across Amazon, not only AWS, for example, we've reduced package waste by close to 2 million tons by simply reducing our package size by about 41%. We've introduced a lot more EVs into our shipping and delivery system. About 9,000 and 145 million packages were delivered using these trucks. So we're definitely seeing results and we will continue to innovate. Of course, it's not going to take only us and we're hoping to see what the innovation continues across the industry

and those challenges that obviously people see in terms of being renewable 24. 7 variable renewables obviously don't get you all the way there. People look at a lot of different solutions like using biofuels, maybe biodiesel in generators, by gas. People talk about nuclear maybe as being part of the solution. How do you see the potential for getting to truly 100% renewable energy 24. 7 round the clock, given that as a you can get a lot of the way with wind and solar, but you can't get all the way there?

Well, I think technology is going to play a significant part in that, but it's also going to take regulators and it's going to take partners and customers themselves to transform in Amazon. We are dedicated to working across the energy industry all up. We work with customers to innovate what they're doing today in the renewable space, but also in the oil and gas space. We will continue to do so. We'll continue to help our customers be as efficient and as innovative as possible using our technologies. And we will continue to innovate ourselves across all of the different stacks that I mentioned, from infrastructure to services to the projects that we're doing in this space, whether it's renewable projects or cutting our own footprint ourselves.

Hussein. Jill, thank you very much indeed.

Absolutely. Thank you so much for having me.

Another of the keynote speakers at the event was a former executive, the company that's been at the heart of the latest breakthroughs in artificial intelligence. OpenAI up until last year, Zach Kass was the head of GoToMarket. Zak, thanks so much for joining us today.

Thanks for having me, Ed.

So you're not an energy specialist by background, but you're coming here talking at this energy event what's the message you want to give people? What do you think the energy industry needs to know about AI generally?

My message is that of abundance, is giving people the permission to explore a world where we might have plentiful energy, foodstuffs, water, education, healthcare. Which is my mission, right, to provide a counter narrative to a dystopian or an otherwise dystopian prevailing sentiment at an energy conference. It's particularly interesting because energy will be both a beneficiary and a benefactor of AI insofar as we're going to need a ton of it to build these machines and run them. And presumably AGI will end up producing fusion and other major energy breakthroughs, which should change how we view energy forever.

So let's unpack that a bit and talk about those two aspects of energy in turn. Firstly, on kind of the demand side, if you like. So what AI needs from energy and what its needs are in terms of energy. You talk about energy, then being a beneficiary, you could also say, obviously that's one of the big problems with AI, is the scale of the energy demand, particularly, obviously, at a time when we're trying in general to reduce energy demand or at least improve energy efficiency in the economy. And a lot of companies have goals for reducing emissions. The world as a whole is aiming for net zero emissions at some point in the not too distant future in order to avoid the worst effects of climate change. AI looks potentially like a big problem for that. How do you think about that challenge and do you have a sense of what you think the scale, scale of the increased energy demand from AI is going to be?

So first, I'll say I think it's pretty clear that AI is going to present the biggest challenge faced by the grid or by energy providers. Yet it doesn't mean that we should shy away from it. I don't think that of all the things that you could compromise on right now in terms of supplying energy, AI feels like one of the few things in this world that actually give us a path out of the climate crisis that we've put ourselves in. AGI almost certainly begets fusion, among many other really positive things. So if you were going to rate limit something, I wouldn't start with AI for the aforementioned reasons. That being said, the inside baseball suggests that the scale of energy required is going to outstrip the current supply by like an order of magnitude currently, that what we think we need is so much more than the grid currently has and way more than we know how to store and distribute. So the Answers have to be found in either major breakthroughs or massive improvements in the efficiency by which we train and run these models. And I think both will end up happening.

It's been fascinating to me, just walking around, the conversations that I've had with people so far at this conference, the kind of the two worlds colliding. So you have the utility world, which essentially inherently conservative by nature, not controversial, I think, to say that. And has very much got used to a period of essentially zero demand for growth, for power in the US Suddenly being faced with all these new demands and people coming to them and saying, we need to build all these new data centers, and each data center is going to add an extra gigawatt of load to your grid. And that really being a shock to people, and people having to kind of change mindsets and think in very different ways about the challenge they face. Is that what you're seeing as. What have you talked to people about that?

Part of what I talk about in circles like this is this idea of sort of disabusing people of this Malthusian approach to the world, which is that we somehow convinced ourselves in the 80s and 90s that the way to preserve our species or to grow our species was actually stasis, that more energy was bad, more population was bad, more consumption was bad. When in fact, what we're discovering is the opposite may be true. And in fact, the opposite almost certainly is true, that stasis is, you know, the beginning of the end. And you start to see that in how demographics play out all over the world. Energy is just a proxy to growth. And an increase in energy demand should actually signal a lot of good things. In this case, I think it signals really exciting things. How we produce that energy is the most important thing. And, you know, the word nuclear is becoming more and more prevalent and more and more positive. And it's pretty clear to most people, I think, at this point that the renewable path isn't actually sufficient to meet the demand, that we need to look beyond our current sources of energy to actually solve these problems.

And you've mentioned fusion. Do we actually need workable fusion power to kind of unlock all these possibilities and as you say, to make growth in energy demand a good thing rather than a bad thing.

I mean, we certainly need nuclear to make growth in energy demand a good thing, not a bad thing? How big of a solar field can you actually build to serve the next generation of AI models? I mean, you can't build one the size of Arizona or Texas, which is what it might require how many wind farms can you actually build before every bald eagle dies or whatever? Right? I shouldn't say whatever, that's flippant. But there are real costs to the traditional renewable energies. We have fifth generation nuclear reactors today that we know are safe and we know are efficient. It's going to take a while to build them, but it seems like this has to be the answer.

Okay, but the counterargument to that then is that the current generation of nuclear technologies that are available have been tried and essentially rejected by the market. So we saw, for instance, plans to build new SMRs in Idaho last year getting scrapped. That new scale SMR project was abandoned. We've had huge cost overruns, delays to reactors being built in Europe and the US in particular, projects going way beyond schedule and so on. And if you compare that with actually very low cost of wind and solar power right now and very low cost of natural gas generation as well, it just seems like it's really hard to economically make the case for nuclear. So how do you get around that?

I'm not an energy economist, so I'll start by prefacing that I will say the equation seems untenable. Otherwise demand is going to outstrip supply by such an incredible margin that simply saying we should build more wind and solar energy isn't sufficient. Even if natural gases are truly abundant. You can only put so much onto the grid. And ultimately, look, I'm again not an economist, but it seems like the solution has to be some outside resource.

So fusion power then does seem to be the thing that sort of unlocks all of it.

This is where I hang my hat.

Right? And if you're being cynical, then you say this is the kind of the deus ex machina. This is the magic one. You have fusion power that'll make everything all right. Sure. But you're saying you believe that AI will be the crucial tool that will help us get to viable fusion power. What do you mean by that?

It seems almost certain that AI will again be a beneficiary of fusion and a benefactor of fusion. Insofar as the experiments that we run today, in general, experiments that we run in energy are becoming much easier, much less expensive because of AI. We're just able to do a whole lot more testing. And you heard it this morning, it's talked about all over the floor, the things people are doing to actually understand better how to move energy on and off the grid. Fusion is no exception to that. AGI almost certainly solves a lot of technological breakthroughs. One of them is Going to be fusion. By the same token, fusion almost certainly solves AGI because if we can solve fusion and we can build a bunch of fusion reactors, well then suddenly we can build a bunch of these massive machines, assuming we can get the compute, which is a separate issue, and run them abundantly.

That is very interesting. Does seem to be a real harmonization of interest then between people who are interested in AI and people are interested in fusion power. A lot of overlap there.

It's not a coincidence that Sam Altman is an investor in both HELI and and the CEO of OpenAI and sort of has called this one of the major problems that we need to solve, the energy deficit problem.

And the other thing I think a lot of people might say is, well, this sounds kind of like science fiction to us. That combination in particular of AI plus fusion. These are technologies that people talk about for some kind of far distant future. How far away are they really do you think? I mean, we come back in DistribuTech 2034 in 10 years time.

I think there's a fusion reactor in 2034. The prevailing sentiment is that it takes six years to build a fusion reactor. From time of discovery, we're probably between three and five years out from discovery. I'll take the under. I'm generally an optimist. I think things move way faster on the large time scale than we appreciate. And again, I just remind everyone, try to explain the world today to someone in 1950. I mean, the things that we have discovered and the ways we have moved ourselves forward is so remarkable. Imagine the scenario in which you will struggle to understand the world in 20 years. A lot of things will have happened. Principally among them, I think abundant energy and AI.

There is that good Bill Gates quote which. So I'm, as you can probably tell, intrinsically quite skeptical about a lot of this. Well, you're British, so but yeah, exactly. It's not a country that tends to be enthusiastic about embracing the future. But Bill Gates said famously, everyone tends to overestimate what they can do in one year, underestimate what can be done in 10. I do think there is something in that.

Yeah. I mean, you're talking about a species that figured out how to fly a plane and then put people on the moon 60 years later. We are quite good at things once we give ourselves the permission to imagine their possibilities. And I think the public perception is actually critical to all of this. You know, this is why I talk about we stopped building megaprojects as a species. And I think that did us an incredible Disservice that no one appreciates. The interesting thing about megaprojects isn't simply that they stand and you can remarket them. It's that it reminds us all the incredible things we're capable of. And these are really important to do to remind ourselves how we can progress.

And certainly from a climate perspective, you could say climate change certainly is a mega challenge.

Exactly.

And it is something.

This is the ultimate mega project, or for right now anyways.

And given everything that makes it so difficult to address with all the technologies that we have today, it certainly would make a big difference if we had something like that.

Maybe the ultimate difference.

Very interesting. Well, let's hope we can both be back here in 10 years time, then I can see you here in 2034.

I would love it.

We can see whether that prediction came true.

I would love it.

Thanks a lot for joining us. Thank you. For a perspective from one of the utilities that will be making the practical decisions about how far to adopt AI. I talked to Quinn Nakayama, who's the senior Director of Grid Research, Innovation and Development GRID at pge, the California utility. Quinn, thanks very much for joining us.

Hey, thanks for having me.

One of the big issues that everyone's been talking about here at DistribuTech is AI. My kind of joke about this is it seems to be both the cause of and the solution to all the problems of the grid, you know, but it's obviously creating new demands on the grid, a lot of new load being created. But also people are talking about AI as being the way to manage the grid much more efficiently and to deal with the various new challenges, including distributed energy resources and so on, variable renewables that are creating a whole new set of challenges. What's your view on AI? I mean, how are you thinking about AI at PG& E? What do you see as the opportunity that's there and what are the potential pitfalls?

AI and ML is a very big buzz word that's going around in the industry. I think it also has a higher level word called automation. And so not all of that has to be done by aiml. There's automation that can happen without that. And so the broader question needs to be where do we need to automate? And I think from our perspective is do we understand all of the different type of use cases that we as a utility want to create further and further automation, whether it's through AI, whether it's through ML, whether it's generative AI or just pure nuts and bolts automation. Once you figure out all of Those use cases that you could particularly envision. The second question is, is this really a technology problem? I think too many utilities jump directly into AIML as if it's gonna resolve all of your related issues. And when you peel back, you might actually find that, yeah, you can do AI ML for this particular thing, but it's a process issue. Stupid, right? And it's like, well, in that case, are we just applying technology to solve a higher level issue? Once you peel that back, then it's really about, well, where are we gonna get the most bang for the buck to apply AI ML to achieve our 10 year targets on our strategy or our 30 year objectives? And we haven't gone through that exercise. What ends up happening is you get all these spot AI ML conversations. If we used AI ML in this area, it would be amazing. Or here, or here. I think we need to take a step back and ask ourselves, what are going to be the biggest benefits? Now, I can tell you, generally speaking, that certain foundational things need to happen before you do more of the advanced aiml. So I'll give you an example. AIML can really help you with identifying where your assets are and cleaning up your asset registry. Once you get there, then you can start stacking on all these other AI ML related use cases. But there are certain situations where you just got to get the basics right and then be able to go into how can AIML help you with inspections? How can AI ML help you with optimization of dispatches? How can AI ML help you do simple estimations without having to have an engineer draw up those drawings? There are definitely places that we can target, but we're going to undertake an effort to really understand what is the world of the use cases and then create a multi year view into what is going to have the biggest bang for the buck. How do we think about data? How do we think about data security? How do we think about cybersecurity? These are all things that we're going to have to tackle, right?

So as you say, when you're thinking about security, privacy, some of those issues, that's really where you start to get into a discussion about the pitfalls and the risks in AI. What do you see those as being? I mean, is there a danger in creating a system? I mean, as you say, you think about automation writ large, a system that kind of escapes from human control.

I'm not so concerned about a system that escapes from human control because I think that we're going to step into the AI and ML in a very methodical way. You know, I'll give you one example where AI ML could be very detrimental to a particular utility's business. And let's take a look at it through the inspection process. Right now, utilities have the ability, due to the cost of photos becoming significantly cheaper and visualization becoming significantly cheaper, that you can get imagery of all of your assets almost all the time. Now AI ML done incorrectly could create a whole bunch of false positives that you now need to go track down. If you don't do this in a very careful way, you can't unsee some sort of degradation on the system. However, if it's wrongfully prioritized, you put yourself at significant risk, both from a cost perspective and a liability perspective. So really, how do you make sure that you're training it, retraining it, getting the right results before you start to really roll it out across the system? And this is where I think one of the biggest things that utilities struggle with is this fast pilot, fast fail mentality where you take particular use cases, you do a quick hyper agile development on that, you test it out and if it doesn't work, you move on to the next thing. Sometimes what we do is we tend to get too invested in our ideas that we have a hard time letting go and we just pump money into a failed idea. And I think with this aiml, it's going to be a lot of that. We're going to try things, some things are going to work, some things are not going to work. And as you kind of take a stepped approach, really try to make it so that the outputs that you're getting are having the material benefit that you want for your 10 year strategy is going to be key. Now is there a world where AIML will take over? What the utilities are going to do? I mean, who knows what the 30 year future is going to look like at the current time. I can't see it. Anything's possible with this technology. Skynet may take over the world. I mean, there are endless possibilities of where this can go. I think what we're going to do is we're going to walk into this very deliberately, push the realms of what we can do. But always keep in mind the safety of our customers, the safety of our grid, that is always going to be at the forefront of what we do. And if we do that, then I believe that we can create an AIML vision and a world that's really going to benefit our customers.

Quinn Nakayama, thanks very much for joining us on the Energy again.

I appreciate it. Thank you.

For more of my conversation with Quinn Nakayama, where we talked about the new challenges for the grid created by electric vehicles. Look out for a special edition of the Energy Gang, coming soon. Now, Quinn made the good point that AI applications are only a subset, really, of the wider range of technologies for automation and digitalization of the electricity system. To discuss those, I talked to Tom Dietrich, who's chief executive of Itron, which supplies technologies for utilities and cities to manage energy, water and traffic. And I talked to him about the adoption of those technologies and what might be holding them back. Tom, good to see you again.

Ed, great to be back with you.

Absolutely. Yeah. Great to have you back on the show. So interested, first of all, perhaps in what you're hearing while you've been here as you're kind of walking around meeting people, meeting other people in the industry, meeting customers, I'm sure as well, and the conversations you're having here. What are people talking about? What has struck you as interesting?

So there are two megatrends that I think are really interesting to point out, one on the customer side, one on the technology side. I'll start with the customer part first. Customers are really, really struggling to keep up with the pace of technology that's coming at them. There's so many different ways to attack a problem. There's so many problems that they have that are unsolved yet today. But technology is coming at them much faster than they can absorb it. And the corollary to that, and the second part of my answer is on the technology side, that technology companies like Itron have realized that, that, oh, this is a real problem. It's a problem for the customer because they aren't solving the problem for their consumers, you and I, but it's also slowing down our business growth potential. So what are we doing in response to that? It is spending more time collaborating among ourselves. Let's pre integrate a solution. Let's have two companies that work side by side inside of the utility to supply solutions to them. But let's do that before we get to them. So we, we present a united front by the time we get to the customer, the utility in this particular case. A couple of quick examples of that is we had some announcements running up to this event where we were going to pre integrate our solution for Grid Edge intelligence with Schneider electric, who does ADMs, or with GE Vernova who does a similar ADMS kind of solution. But if we can. Sorry, ADMs, sorry, too many acronyms. But Advanced Distribution Management System. So Itron works close to your Dwelling close to the home, higher up in the transmission line and in the stack, higher voltages where these other companies work. If we can come and show up with the solution together, it helps the customer deal with this problem that they have to provide good and reliable service, but also deal with the pace of technological change.

That's very interesting. Yeah. So going back to the first point you were making, as you say, about the two big trends, that's something I've also been really struck by. And it's obviously it's always one of these kind of terrible cliches that people come up with about, oh, there's so much change and things are, you know, we live at a revolutionary time, etc. But actually in this industry, for utilities in particular, who are crucially your customer base, it does really seem to be true. There is something that's very different about the world we live in. And the things I think of are, I suppose, two crucial ones. One is we're on the demand side, is that after a long period of essentially static demand in the United States for power, suddenly that's really kind of taking off various trends going on here. Data centers for AI being one big one. The revival of manufacturing industry, new plants springing up, particularly chip plants coming, has been a big thing. But also obviously demands for electric heating, electric vehicles becoming very significant. And then on the supply side, obviously the well known challenges of managing variable renewables on the grid, which does creates a completely different environment for utilities. Is that how you're seeing it as well?

Absolutely. I identify with everything that you said there. I had a customer event yesterday, so picture this. A room with about 100 to 125 customers inside of it. And they were just talking among themselves and one customer said, I see my demand doubling or perhaps tripling over the next 10 years. So all we have to do on the generation side is build what we built over over the last 100 years in the next decade. Oh, that doesn't sound hard at all. And what was driving that demand increase was exactly the things you were talking about. It is manufacturing, it is AI driven data centers, and it is electric vehicles, which are wildly different loads, especially EVs, than what the utility is used to dealing with.

And so in terms of tackling those challenges, what can technology do to help and what are the crucial innovations? And as you say, so there's one particular thing you've just been talking about in terms of suppliers getting together, forming alliances and partnerships to provide kind of packaged solutions to utilities. But what are the things that really Kind of make a difference and will help them deal with these as like what is genuinely, I think, a very new world.

Yeah.

I will focus my answer on the distribution grid, meaning that the stuff that's close to where you and I live, close to our houses. I think all of these distributed energy resources, electric vehicles, rooftop solar batteries, behind meter, all of those resources themselves are starting to pop up in an uncontrolled way from the utility's perspective. So first and foremost, how do you provide visibility to the utility where those things are that is truly, truly helpful for the utility? If I know where they are, then I can reach out to that customer and say, hey, would you like to enroll in my program to control charging Secondarily is understanding what capabilities you have as a utility to deal with that. Hey, will that person allow me to control when their vehicle is charged, be able to control it, turn it on and off when you need power on the grid and finally optimize, please charge this car when the sun is shining on that roof. So this whole progression of visibility, capability, control is the enhancement that's in front of us.

And the fact that utilities don't have that at the moment, I think is often shocking to people when they think about it. And if you come from another industry, anything else, where you're in the business of serving the consumer, the amount that utilities just don't know about what their customers are doing, both in terms of demand and actually then increasingly because of distributed energy, supplying power back to the grid as well is amazing.

The grid was designed as a one way push. It was always assumed we'll just push power down and it'll be there and suddenly the world changes. The way that isn't really economically possible, possible or feasible. And now how do you deal with that from a consumer standpoint, it is shocking, the service imbalance. When you procure another service in your life as a consumer, you ordered Uber Eats for delivery as an example. You knew in real time when the driver was leaving the restaurant and his or her name and how much it was going to cost and when they were showing up outside your flat. The last time you dealt with your power company may have been lights are off and you're ringing them up saying, when are my lights coming back on? And they may not have even known the lights are out. That service imbalance is a huge barrier for the utilities to overcome and something that I think technology like Edge intelligence that I talked about can really help.

So it's not really a surprise that these changes are coming, right? I mean, people have seen this at least kind of 20 years in the past, 30 years in the past, people could tell that this was the way the grid was likely to evolve. And people had that sense that the grid of 100 years ago was not going to be fit for purpose in the 21st century. How rapid has progress been towards adopting these new technologies? It feels to me like actually the industry has been often much slower than it should have been and now things which have been seen sort of as looming future possibilities heading towards us are now actually very real and are really affecting people and the industry is just not really prepared in time. Is that fair or is that unfair?

I think that's sadly extremely fair. I think grid modernization now needs to happen in years, not decades. And we're starting from a disadvantaged position. We are behind. There's lots of reasons the regulatory models that are at play, how traditional buying patterns have worked. Utilities are used to buying things and putting them in the field and depreciating them over 10 or 20 years. But. But if you don't know what happens next month, let alone five years from now, how do you plan for that? So there's many things that need to change about the buying model, the regulatory model and really the technology itself to be able to help solve the problems.

And what more could be done to help expedite this change and to get these new technologies deployed across the US Industry. Is this something that policy needs to help with more? Is it something that needs to change in the system of regulation? I mean, what kind of can really be done to sort of accelerate this and as you say, to get this change accomplished in years, which is what we need to do.

My off the cuff reaction is can I choose all of the above as the answer? I think that there are policy changes so governments can help drive policy in a way that makes it easier for utilities and for technology companies to move faster and drive better solutions? One, I think regulatory policy can be adapted to make it easier to deploy new technologies. Today it is very difficult to deploy SaaS solutions software as a service based solutions because it doesn't cleanly fit into the model. Most utilities have a change in the regulatory structure to allow capitalization of software as a service as an example is something that's possible that would also be helpful to move the technology forward itself. And then on the provider side, I think we have to continue, continue to push the envelope and do a lot of pre integration, make it easier for our technologies to be consumed.

And as you meet people, walk the corridors here, listen to the conversations that are going on. Has it left you more optimistic that the progress is going to be made to update the grid the way we're going to need to?

I would say that I'm certainly more optimistic than I was 48 hours ago when I set foot on the ground. Not because we've got the answer in hand today, but I see a lot of like minds thinking and talking, talking about the same things and that really is step one to solving the problem.

Well, Tom Drisik, thank you very much indeed. Thanks for talking to us about that today.

Yeah, I appreciate your time. Thanks for having me.

Finally I talked to Anthony Allah, who's the head of Hitachi Energy in North America. Hitachi is an important supplier of transmission equipment. It was a pioneer of high voltage direct current transmission lines all the way back in 1954. Distributec is a great opportunity for him to meet customers that use Hitachi's technology and I asked him what he'd been talking to them about. Thanks very much for joining us on the energy gang.

Nice to meet you. Good to be here.

So I wanted to start off talking a bit about DistribuTech and your presence here. So Hitachi Energy is a very important company in the transmission industry. You were the pioneers of hvdc, high voltage direct current transmission lines. That's I think it's 70 years old this year. Right. 1954, the first one was built. And so this is supposedly the leading event for transmission and distribution in the power sector, certainly in North America. So it's obviously a great opportunity for the industry to get together people to have these kinds of conversations. What are you really wanting to talk to people about while you're here? What is going to be the focus of the conversations you're having at distributed?

It's a good question. There's so many topics so I'll try to zoom on a few. I think if I reflect on the number of conversations we've had since we've been here with customers, I'd say there are a couple of themes. One is still very high for good reasons in the mind of the customers is the supply chain constraints. We know now for a number of years we've seen lead times for equipment going up. We've seen inflation impacting also the price of deploying those solutions. So those conversations are still needed because historically we've seen that those challenges were imparting certain part of the technologies, providers like us. But it's also now starting to catch up on other parts, other equipment. So kind of the second wave, third wave is coming in so it's important for us to really share what we see globally, what we see in North America with our customers so that they can then anticipate and get ready and get ahead of the curve. So that's one topic. Another topic is really on digital solutions, which is historically distributed, has been a little bit more on the digital side and remains, I think, heavy for good reason on the digital side. So for us also, it's an important topic. We know that the grid today needs to be more flexible, more resilient, needs to be able to accommodate more renewable energy. So digital is an enabler for all of this. So that's also for us, a key topic that we've been talking about with customers. So I'd say those would be two that have been really front and center since we've been here.

Thanks very much. I want to take those in turn then thinking about supply chain challenges first. So you're saying that things are getting worse still in supply chains, delays are getting longer. I mean, it feels like this is something that people have been talking about really since, I guess, the pandemic ended and since investment activities started to pick up and you might have hoped perhaps that some of the issues in the supply chain would have started to get fixed by now.

But no.

Yes or no? So the yes is there's hope. So I would say the first type of equipment which have been impacted by lead time, I think if you ask around, what will come to the mind will be transformers. Right? So here now, fast forward now two to three years now, what we see is we've seen a number of announcements in terms of investment in additional capacity. We have ourselves made a couple of investment here in North America. So here that additional capacity, some is already online, some of ours is already online, some of the industry is already online, but some is not. So I think the market sees that we're making progress, but still the demand still is much higher than the supply. Right. So we see improvement, but we still see a bottleneck, clearly.

And just particularly on that issue of transformers, there's been a lot of discussion about that. Obviously, people have become very kind of fixated on the question of bottlenecks and the transformer supply chain. But there then it sounds like you've got a reasonably positive, hopeful message for people. You think that as you say, that new capacity is getting built, we are going to see those bottlenecks start to ease?

I wouldn't go that far. We see capacity coming in, but still the new capacity is still below the demand. Right. So we still expect to see lead time at a much higher level compared to what they were, for example, three, four years ago. Right. So we see investment. This is encouraging, I think, for all of our customers. That's what they want to see. That's what we want to see as an industry. But still the demand is higher than the supply. So we're still going to see an imbalance for some time. So that's the. For the. Yes, it's getting better. The, no, it's not getting better is now what we see is other type of equipment where now we see the lead time increasing. And here, this is where we have a lot of conversation with the customers about that to make sure that they understand so that we can work with them on the planning, on the anticipation

and what types of equipment that are affected by that.

For example, circuit breaker is one of them.

And again, are we going to ultimately see, do you think, more investment in capacity?

I would assume we'll see the same dynamic.

And it also just takes time, similarly for those kind of things to work through.

It takes time, yes, because if you take the example of a transformer factory, okay, if you want to build a new one, or if you want to expand the capacity, you need to find the people, you need to be able to find the machine that you need, a winding machine or another machine that supply chain itself or also has its own challenges. Right. So it's a tricky situation. And on top of that, you have the question of the access to the raw material also.

So pick up on the other big topic you said that everyone's talking about and that you're talking about here, which is digitalization. It has become clear, talking to people that I've had conversations with just in the past day or so here that, as you say, enormous interest in digitalization as part of the solution to this huge new set of challenges that the utilities, the power sector is facing. Probably don't need to rehearse them all here, but coping with increased proportion of variable renewables on the grid, increased distributed energy resources, new kinds of demand, new load, particularly in North America, from data centers for AI, new manufacturing facilities, and so on, a whole lot of pressures hitting the industry simultaneously. When you talk about digitalization as being the thing that people are looking at, what does that mean specifically and how does it help with those challenges?

Yes, I guess the question is, what problem are you trying to solve with digital solutions?

Right.

So I think there are a couple of problems that we see the industry trying to solve. One of them is resiliency. We also see an increase in extreme weather events. So how do you restore faster? How do you anticipate? How do you bring more flexibility into your grid to be more resilient? Right. So this is a type of solutions of problem that looks for a solution and that's where digital connected product and intelligent software come in to try to provide a solution. Then when you talk about renewable integration, right. How do you make sure that in real time you able to keep your supply demand balance on your grade when you know that the sun and the wind change so fast? So that's another problem that can find a solution into a digital solution. And the list goes on and on and on. Those are kind of two examples of where a digital solution can support those problems.

Right. And though beyond that it does seem very clear that there are some problems for which digitalization and the kind of technologies you've just been talking about are not a complete solution. And I think no one would deny that. When you go around here, obviously a lot of people here at DistribuTech with various kinds of technologies, digital solutions that they are offering for different aspects of the power grid, what you will hear all of them say is this can take you so far. At the end of the day, we do still need to build more physical infrastructure. Digitalization cannot be a complete solution. That's presumably something you'd agree with. Providing that physical infrastructure is very much the business that Hitachi Energy is in. What are the issues that are really holding back? That investment in physical infrastructure does seem increasing, incredibly difficult in so much of the developed world in particular. And it's true in the United States, but it's true in many other countries around the world. It's certainly true in Europe, it's true very definitely in the uk. It's just very, very difficult to get more transmission lines put in. Why is that?

That's a difficult question. I think there are a couple of things. I think the first one that we don't see in the United States is a national transmission planning mechanism compared to maybe what we see in other part of the world. Right. So this is something that there's a lot of talk about it and we certainly believe that this is the right thing to do. So here we hope that we will move into this direction. It's number one. Another challenge is around permitting, right. So here also this administration has been trying to address that topic and we all know it's a. It's a complex topic. So again we command what they've done so far and we really hope that collectively we can find a solution to this, but then there is another element which is the speed at which we also need to deploy. Assuming that we have all the permitting in the world, all the planning in the world, then you still need to execute on those projects. And I think here, if you think about it, there's a number of statistics that say we need to double, triple the grid. But let's assume that for the sake of the argument, we need to double the grid in the next 15, 20 years. It has taken us almost more than 100 years, right? The grade was first created in other part of the world at the end of the 19th century. So here, now we're talking about rebuilding the same grade in the next 15, 20 years or so. So we need to think differently. The technology is here, so we have the chance to have the technology, but we just need to think about it differently. We cannot reproduce the same scheme and assume that we can go at a much faster pace. So here we really need to think about new business models. As an industry, it's not only one or two stakeholders, the entire industry or the utilities, the technology suppliers, the government, all the stakeholders need to come around the table and find different ways. Right? And we see different ways in all over the world, right. We see in Europe, for example, a different type of business model where some utilities have decided to sign long term frame agreement with companies, for example, like us, on a specific type of technology. The positive result of this is that that gives them security of supply for the next ten years or so. And for us, that allows us to know that we have a security of demand and in return allows us to accelerate our investment. Right? So I think this is an example of new type of business model that I think we need to look at more closely in North America. And yes, the regulatory construct is a little bit different in this part of the world. So cannot do it copy paste of what we see in the world. But we need to look at it and try to see what are we learning from that. One example I get from Europe and how could we tweak that to apply it to the US in a way that works with the US regulatory construct?

That is really interesting. How optimistic are you that that kind of new thinking in business models can be made to stick? I mean, just because the reason I raise that question is that it feels like this is a very conservative industry. The utility industry doesn't change much. It's essentially you were talking about the physical infrastructure being the same as it was in the late 19th century. Actually, the business model is also the same as it was in the late 19th century has not evolved much and for good reasons. People are risk averse and conservative and don't want to shake things up in ways that might lead to the lights going off. So how can you persuade people that they do need to think differently?

To your question, I am fairly optimistic for two reasons. I think number one, this country has demonstrated that when there is a problem, there is a solution and at some point all the stakeholders get together and really find a solution. But also we are having very different conversations today with our customers compared to the conversation we had maybe four years ago. Right. And that's really I commend our customers here in terms of acknowledging that we need collectively to work differently and the steps that we are collectively making. So I think here I think this is very encouraging. We still have a ways to go but I'm optimistic that we'll get there.

So you think we come to distribute it again in five years time, let's say distributed 20, 29, you think things are going to seem pretty different then?

I would hope, I will hope that we will see new business model, that we see a regulation which has evolved and enables faster investment and I'm sure also we will see a lot of AI solutions on the floor like we already start to see this year. But I'm sure in five years from now there will be not a buzzword, there will be a reality.

And just to take that step back then to think about the broader context of this, it's now one of those cliches, the buzz phrases that goes around a lot. People say no transition without transmission, that building new grid capacity is absolutely essential to shifting the electricity system over to low carbon technologies, renewables and other kinds. So this is something that if we're serious about addressing climate change we really do have to make progress on, right?

Absolutely. There is no transition without transmission. I fully agree and I think now what's very encouraging is again going back in time. Three years ago people in the industry knew about it but today this is public knowledge. This is something that you see in major newspapers and that's really the prove that now it's really a fight that is well accepted and understood but now we need to get to work.

Anthony Dello, thanks very much for joining us.

Thank you.

So that's it for us. From the first full day of distributec I found it fascinating to talk to all these people about the problems they're facing and the solutions they're advancing. When you stand back and think about the energy transition at the highest level I think it's very easy for the debate to get pretty abstract, but what we have at this event is there's thousands of people who are working on the challenge of decarbonising the energy system, and they're creating real change in the shape of metal and concrete and software. It's actually a really inspiring thing to see, I think. Many thanks to all of our guests who've been talking to us about what they're doing. Many thanks to our producers, Dan Cotterell, Toby Pickens, Gilchrist and Sam Nash. And above all, of course, many thanks to all of you for listening. Please do keep your feedback coming. Look out for the rest of my conversation with Quinn Nakayama of pge. You won't want to miss that. I think it is really a fascinating discussion, and we'll be back soon with all the latest news and views on the energy transition. Until then, goodbye.