Can AI revolutionize grid operations?

Everyone loves to talk about the “broken grid.” But if you look inside a utility’s planning and operations functions, a different picture emerges: infrastructure that is remarkably reliable, built and run by engineers solving impossibly complex problems with often outdated tools. In this episode, Shayle Kann sits down with Josh Wong, CEO and founder of ThinkLabs AI, a company building physics-informed AI for utility planning and operations. They break down how utility studies get built today, the data cleanup headaches, the silos between teams, and the “worst case” mentality that drives an overbuild of infrastructure. Shayle and Josh discuss: Why utility planning is split into silos across transmission, distribution, and DER planning. The gap between studies that identify grid problems and the work of actually solving them. What “physics-informed AI” actually means. The shift from AI that identifies problems to AI that actually solves them. The disconnect between utility planning and real-time operations. Should data center microgrids be treated as micro utilities? Resources: Latitude Media: GE spinout ThinkLabs has a new copilot offering for the grid Latitude Media: Do microgrids make sense for data centers? Latitude Media: The case against relying on behind-the-meter power for data centers Latitude Media: Unpacking DOE's proposal to transform data center interconnection Credits: Hosted by Shayle Kann. Produced and edited by Max Savage Levenson. Original music and engineering by Sean Marquand. Stephen Lacey is our executive editor. This episode of Catalyst is brought to you by ENGIE, the smarter energy supplier. ENGIE doesn't just provide the power to run your business — they supply the energy to move it forward, with reliable, flexible solutions built for what's next. Learn more at engieresources.com. Catalyst is brought to you by EnergyHub. Peak season puts every grid to the test — and the utilities that pass are the ones that built flexible capacity before they needed it. EnergyHub works with more than 170 utilities to coordinate 2.5 million devices and 3.4 gigawatts of dispatchable flexibility through a single platform designed to perform when it counts most. See what that looks like at EnergyHub.com. Catalyst is brought to you by Bloom Energy. Bloom Energy fuel cells deliver affordable, ultra-reliable onsite power for hospitals, utilities, and data centers – at speed and at scale. Learn more by visiting BloomEnergy.com.

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:02Speaker 00

  Latitude Media, covering the new frontiers of the energy transition.

• 0:07Shayle Kann?

  I'm Shayle Khan. I invest in early-stage companies at Energy Impact Partners. Welcome to Catalyst. All right, let me just say that it's an incredibly exciting time in the energy world, and in electricity in particular, for many reasons that you already know and I don't need to explain. And it has, for me at least, been great to see so many people start paying more attention to energy over the past, I don't know, 24 or 36 months, something like that. But I will admit a little bit of frustration with a particular crowd that is constantly referencing the "broken grid" that needs fixing, knowing full well that most of these people have absolutely no idea how utilities actually work. Even setting that aside, I think very few people outside utilities know what actually goes on

• 0:51Shayle Kann?

  inside their planning and operations functions. But once you do start to see inside, I think two conclusions sort of naturally fall out. First, it's wildly impressive, and we are all fortunate to be afforded the degree of reliability and even affordability that modern society gives us with electricity. And I say that as a Californian who pays over $0.40 a kilowatt hour for power and sometimes experiences wildfire-driven proactive power shut-offs. Okay, but the second conclusion is, oh, man, what a future AI might deliver here. So I wanted to talk about that second one. I brought on Josh Wong. Josh is the CEO and founder of Think Labs, where we at EIP are investors. Think Labs is leveraging AI in both utility planning and in operations.

• 1:37Shayle Kann?

  We talked both about how things work today and how they'll change in the future. Josh is coming up after the break.

• 1:50Speaker 00

  Demand flexibility is now an essential part of the data center's speed-to-power playbook. But how do data centers actually curtail load? Grid Beyond is the end-to-end solution for flexible energy management. Grid Beyond's advanced derms and VPP platform is tailored to what data centers care about most: maximizing reliability, intelligently meeting curtailment obligations, reducing costs, and managing all your distributed resources in one place. Ready to transform your distributed energy management? Get started at gridbeyond.com. Catalyst is brought to you by Antenna Group, the communications and marketing partner behind the most influential companies in energy, climate, and infrastructure. For three decades, Antenna has helped ambitious companies stand out,

• 2:32Speaker 00

  build market leadership, and become category-defining brands. From breakthrough startups to global industry leaders, their team brings the expertise, relationships, and track record to accelerate your momentum. Learn more at antennagroup.com.

• 2:47Speaker 02

  The power plant for the modern grid isn't a building, it's a network. Energy Hub's derms platform connects batteries, EVs, thermostats, and commercial and industrial sites into virtual power plants that over 230 utilities dispatch as reliable grid capacity, day in, day out. VPPs deploy in 60 to 100 days at 40% to 60% less cost than traditional infrastructure, and they perform like conventional generation from day one. Learn about the flexibility platform for the modern grid at energyhub.com.

• 3:22Shayle Kann?

  Josh, welcome.

• 3:24Speaker 03

  Thank you, Shayle. Great to be here.

• 3:26Shayle Kann?

  It's crazy that it has taken me this long to have you on, but I'm excited to finally have you here to help explain how utilities actually work, or at least part of the utility functions. And we're going to talk about two parts that, as you pointed out to me separately, do converge, but I want to talk about them somewhat separately. The first being planning, how utilities plan, and the second being operations. What are they doing day-to-day to maintain the grid? So let's start with planning. What is, in the modern day, but prior to revolutionization by AI, what does utility resource planning actually look like inside the utility?

• 4:04Speaker 03

  Yeah, there's many parts of planning. When you say resource, it might be integrated resource on the bulk power system or with centralized generation, there's transmission planning, there's distribution planning, then there is DER planning. Each one of these are on silos. And recently, just talking to a very large East Coast IOU, even transmission planning, there are like 12 departments doing their own thing. And each time a study is done, it's completely separate and siloed. So each use case is like a study. You want to connect a generator, it's a study. If you want to connect a load, it's a different type of study. But the underlying model is still the same. So studies today, I would say, half, if not more, of the time,

• 4:47Speaker 03

  especially for distribution, is on cleaning up data. So data is in a ton of different places. Data quality is not to par, and a lot of manual effort is required to pull data together. Then you run the analysis, which is really tuned for a worst-case scenario planning. What's the five hours of the year that's going to be worst-case for the next 10 years? And you plan to that level of standard. Now, people are improving, like looking at 96 hours per year, 288 hours per year, 576 hours per year. So each of those are like high best-case and low cases, weeks of the season or month over the years.

• 5:32Speaker 03

  And I would say the gold standard today is probably 8760, which is like 24 hours by 375. So every hour of the year for profile analysis, great for battery storage, which you need to charge and discharge, and also really necessary if you start looking at things like load flexibility. But people are not there yet. So I think by and large, it's still worst-case scenario planning. And this process just repeats over and over again. We call these re-studies. When I used to be a planning engineer when I started my career, we say, "Planning engineers are always right because we can always blame it on the assumptions. But we are always wrong because the assumptions are never correct."

• 6:14Speaker 03

  And so it becomes this vicious cycle of blaming data, et cetera. And now I think the greatest place where we stuck, though, is all the studies tells you problems, not solutions. So you run a study through, let's say, a decades-old tool for distribution and transmission planning, and it says, "Hey, you have all these congestion and these locations, whether in under contingency or not, and you have voltage violations here and there. So red light or yellow light, you cannot connect to the grid. Restrict access or build more lines." But the actual solutioning depends on the planner, and it's completely by trial and error, or, in other words, experience.

• 6:58Speaker 03

  And so each time the engineer sees, "Okay, I have a sea of red here, all these problems. How do I solve it?" In the past, it's fine because it's all upgrade. Build more substations, add more wires, more transformers, that's fine. But that's not going to cut it right now because of affordability and supply chain-ish lags, et cetera. So now we have a ton of other options. Should we flex, like curtail, curtail generation or curtail load? Should we add a battery? Where? How big? When do we charge and discharge? That's an operational issue, again, instance to operations. Or should we switch the grid around? Oh, that's operations. Planners shouldn't care about switching. But it is getting more important.

• 7:42Speaker 03

  And then we do have to upgrade our lines. I think FERC recently just said, "We do need to reinforce the backbone of our infrastructure." But how much? Should we do it after giving precedence to flexibility? Or should we bring your own capacity? And where? And so there's all these millions of lines and quintillions of combinations of lines we can build, but where and how much? And trial and error is not going to cut it. So I think we're at a time where we are completely overwhelmed by these manual processes, re-studies, and just, I would say, shot-in-the-dark type of trial and error on what type of solutions we can use to mitigate the grid. And that's where we're getting stuck.

• 8:24Shayle Kann?

  There's an extent to which, when you describe it that way, like the studies and re-studies and the judgment required by the planners, your former colleagues, and so on. There's an extent to which my reaction is like, it's remarkably impressive that we have the reliability that we do have, right? Like, we actually have quite good reliability, right? And it's worked for quite a long time. Do you think that that is because, as you said, basically, the planning is built around, "Let's make sure we solve for the worst-case scenario"? And so is the net result of that that we have high reliability but a lot of overbuild, basically?

• 9:02Speaker 03

  I do think the grid is, as we call it, the human's most complex and largest machine, is an act of miracle. Like, it is impressive how much reliability gives us and how much we all depend on it for all livelihood, social, economic development, security, et cetera. And it's been keeping doing well for what we need it to be. And I think redundancy is one thing, but it is an engineering marvel for its time. I just think it sort of failed to keep up with the digitalization and the modernization process, especially for the last two decades or so.

• 9:42Speaker 03

  But we're very good at building infrastructure, like bare metal, coppers and wires, oil in the ground, et cetera. But running a digital grid as a software company, as a modern company, I think we are just in the beginnings of that phase.

• 10:00Shayle Kann?

  Back to what actually happened inside the utility, just to make it more tangible, right? I think it's easy to imagine the bulk power system. Somebody says, "I want to connect a new generator," or "I want to connect a load. I want to connect a data center." Okay, so somebody inside the utility then has to go run a study and say, "Okay, you're trying to connect 300 megawatts of load. Do we have capacity for that? What happens if X, Y, and Z happens, et cetera?" So that all makes more sense, I think intuitively, than the distribution side, where it's less obvious to me what would trigger a study. Is it just that utilities are cycling through studies of individual feeders all the time? Or is it like, "Hey, we noticed that there's a bunch of new load on this feeder.

• 10:45Shayle Kann?

  That neighborhood has a bunch of EVs that just suddenly showed up. We should do a study of whether we need to upgrade a substation." Like, what does the distribution side look like?

• 10:54Speaker 03

  Well, I think it's actually quite common across both, but the characteristics are different. Transmission system naturally is one big thing, so every single time you study into entire region. I think for a cluster study or large interconnection study, the grid impact and solution side, the base, I think most utilities are talking about six to nine months to perform one of those studies. And it's a one time, right? So risk of restudies will compound that. And it does cost internal utility typically a hundred, a couple of hundred thousand dollars. The typical benchmark is about quarter million dollars per study. So it is a long time and it is very expensive. But you study the whole system at once and it's completely reactive and ad hoc.

• 11:38Speaker 03

  So you basically respond to an interconnection request and maybe now we can cluster it. But I think clustering is sort of a shortcut because the queue changes all the time. So even if you cluster it, I think it doesn't add any more certainty to the queue. Distribution, you have simpler studies, but there are places where it is actually more complicated. As I mentioned, data quality suffers so much more because the level of coordination between distribution utilities and transmission and bulk power and the market participants are not there. So you have less checks and balances and audit trails around distribution data.

• 12:19Speaker 03

  And you have a lot more nodes or customers on distribution than on transmission systems. But the good thing is it can be completely paralyzed. So I can study one feeder at a time, unless you want to study around switching, like load transfers, et cetera. And so, but the volume is also key. I think when we, a public study that we did with SCE, I think they are projecting like up to 10,000 of these energization requests per month. And each one of them takes like 30 to 45 days. So if you do it manually and expect different results, like do more of the same, it is going to be a massive issue.

• 13:00Speaker 03

  And so that's where the scalability, data quality, automation is really key for distribution even more.

• 13:06Shayle Kann?

  Okay. So you've been alluding to this, but let's talk about it directly, which is what changes in a world of AI here? You've described a bunch of challenges, right? Like you've got the timeline of all these studies that take a long time. That's kind of an outcome. The challenges that lead to that are data quality, particularly at the distribution level, the, you know, all the different parts of the system being siloed and all the studies being siloed and so on. Like what does it look like in a native AI world? And what's the fundamental challenge to get to that?

• 13:41Speaker 03

  So AI can be ambiguous. So I think let's look at general AI and where we have been focusing on. But I think the origins of Think Labs is to ask a big question, right? Can we automate the grid, whether it's planning or operations similar to how we can drive towards autonomous driving? So can AI help drive towards that efficiency and autonomy to handle the complexities

• 14:06Speaker 03

  with human interloop? So human centered AI is what we're looking at. So, but AI can be, has been historically used for things like data cleansing, pattern recognition, right? Really good for data cleansing. Forecasting is probably the dominant use case in AI. And most recently, there's a lot of general AI around, like there's massive amounts of data. Can we retrieve all the records? Can we learn from manuals? Can we learn from previous rate filings? That's great. For us, we specifically ask, can AI help us plan the grid? Meaning actually to run a system study. How does generators and loads and battery storage impact the grid?

• 14:48Speaker 03

  If we need to invest in lines or connections, reinforcements, where do we do it? How do we do it? How much flexibility versus how much storage versus how much wires? That's a really complex problem. So can AI help us do that? And so for us, what AI is, is physics informed AI. So can we teach the AI how the grid works? How power flows? How voltage traverses through the line? How loads behave in contingencies? Where should we put new lines? Where do we have capacity? So, so we need to make sure AI is trustworthy.

• 15:30Speaker 03

  And to, for AI to be trustworthy, we need AI to learn the physics of the grid. And so can we have physics informed AI? So teaching the AI how the grid works so that the utility, which is, which is the actual entity to authorize the interconnection can actually believe in the results and drive towards actions of the results. So that's where our, I would say specialized AI work in terms of learning power flow, learning system interconnection impacts and learning how to actually plan the grid.

• 16:05Shayle Kann?

  How analogous is that to an LLM? Like I think of the, you know, maybe the analogy would be you train an LLM or let's say like a coding, an LLM that's good at coding, right? You have to like train it, it learns how to code. And then when you're doing inference, you are, you know, using that training or using the, the model that has been trained to, to execute tasks. Is it a similar thing in the sense that what you have to do is basically train a model on the physics of the grid? And then once it has been trained, then the idea is you can query it like you would query an LLM and get something that is driven by that physics?

• 16:47Speaker 03

  That's right. I think, one, we're not an LLM, but I think the training inferencing process is similar. An LLM is more meant for generalization. It learns a lot of stuff and you can have natural language type of interfaces, but it is prone to hallucination. So for us, since we're trying to really bring the world of engineering and AI together, we really lean into machine learning, so deep machine learning models. But we are really playing at the intersection of physics equations and the AI models, the machine learning models themselves. But what that drives towards is a very deterministic model.

• 17:29Speaker 03

  So our models actually can't hallucinate. It's actually deterministic. You ask it the same thing, it will give you the same answer all the time. The other thing is you have that type of efficiency, like the unit economics of AI models. Like we're very familiar now with frontier LLMs that will cost billions of dollars to train and months, if not years to train as well. It learns a lot of stuff, but it is a huge hefty investment. And you have to generalize it towards everybody. So it is a jack of all trades and getting smarter and smarter all the way. For us, there's a few things. One is we need to establish security.

• 18:10Speaker 03

  So we do models that are exclusive for that utility. We're not getting one utility's data and trying to cross train with another utility's data because that would violate the security of the, of the utilities. So we actually bring all our AI models into utility and basically train their own quote unquote foundation model. And now you said, you might ask, that's a lot of work to train a foundation model based on each utility's own proprietary data. But as a case in point, and we have done that for multiple large IOUs now, a couple of thousand buses, it takes us about 10 minutes per training run.

• 18:50Speaker 03

  For a power flow model, the size of a state typically. And that 10 minute training run costs maybe five bucks of compute. And we were able to get powerful models that are over 99.9% accurate across all your quote unquote system states, so your voltages, congestions, like line flows, voltages, et cetera. And so thus, these are small, deep, deterministic and highly, highly efficient models. But now what you get though is like super speed. So studies that would take previously nine months, as I mentioned, interconnection studies, now it takes us only a matter of couple of minutes, like 10 minutes or less.

• 19:35Speaker 03

  And it's also becomes hyper parallelizable. I can completely scale up with more GPU, so I'm not bounded by CPU compute because traditional mathematical like iteration, like iterative techniques, like Newton-Rapson techniques are very much CPU bound. Now, we are working with Nvidia to see how we can run GPU native type of calculations, but that's, that's still evolving. And then the last thing, which is, I would say, mind blowing for us as well, is that AI can be generative. It doesn't just give you the problems, right? In the past, grid simulator only give you problems. Now we can actually say, hey, we've learned all these combinations, but now we can generate solutions. And to your point, this is where the training comes in.

• 20:17Speaker 03

  We have been so used to doing all these studies, one use case at a time, one study at a time. And there is no knowledge retention inside the system itself, the process itself. We just keep doing things over and over and over again with different inputs. But here we can pre-train the AI on the grid where we could say, hey, let's train from zero to 50% low growth by 2030. That's pre-trained from half a gig to 20 gigawatts of data centers. Let's pre-train from zero to 100% renewal penetration by 2040.

• 20:57Speaker 03

  So we can bring all these scenarios and prepare the AI for all these complexities. So when we get into inferencing, it's actually lightning fast. We're talking about like sub second for an 8760 power flow. We're talking about, you know, in a recent example with a large utility, around 15 minutes to generate about over 10,000 line comp lines that can solve all of the system operators region for low growth. And so we can generate the right solutions to the problems and that's really where the power of AI is.

• 21:40Speaker 00

  For data centers, connecting to the grid quickly means two things: building distributed energy and having a plan for demand flexibility. Grid Beyond helps with both. Grid Beyond's DRMs and VPP platform unlocks the potential of distributed assets and dispatches flexibility into the right market at the right time. And this is more than software. Grid Beyond supports developers every step of the way. From modeling your highest ROI energy mix to orchestrating the full set of assets around a data center, the track record speaks for itself. Grid Beyond manages over 1,400 sites and more than 3.7 gigawatts of load across Europe, the UK, and the US. Learn more at gridbeyond.com or click the link in the show notes. If you work in energy or climate, you already know the company's Antenna Group has helped

• 22:25Speaker 00

  build. For three decades, they've been the communications and marketing partner behind the industry's breakout companies, category creators, and market leaders. Today, their team is working across the most consequential areas of the energy economy, from data centers and grid modernization to nuclear, geothermal, storage, critical materials, and next-generation technologies. Antenna helps ambitious companies become the names everyone in the market knows, trusts, and wants to work with. They build reputation, elevate executives, drive demand, and turn momentum into market leadership. If your company is ready to lead its market, Antenna is ready to help you get there. Visit antennagroup.com.

• 23:05Speaker 02

  When you harness the power of millions of DERs in a VPP and put artificial intelligence behind the dispatch, the outcomes exceed conventional generation. Energy Hub's DRMs platform has been trained on years of real dispatch data from over 2.6 million DERs, so utilities can plan with confidence. That means consistent load reduction across a four-hour event window, even during the widest, most uncertain peak periods. Utilities target, shape, and verify capacity down to the meter. Unlock flexible capacity that costs 40% to 60% less to deploy than traditional infrastructure in a fraction of the time it takes you to break ground on a peaker plant. Learn more at energyhub.com.

• 23:49Shayle Kann?

  Okay, so we've been talking about the planning side. And there, I think, you know, we painted a reasonably clear picture of like what the past looks like, or really the present most utilities looks like, which is a lot of planners running a lot of studies manually that take a long time and then need to be re-studied and re-studied and re-studied. And people understand how at least part of this results in on the bulk power system, long interconnection queues and so on, but also on the distribution system, why it's so, so hard and slow to get new stuff built. And then, of course, this new world that you're describing, where all this stuff is essentially automated and there's a human in the loop, but it's, you can generate them very quickly. Let's switch to the operations side. I'm going to paint for you a picture that I think is what most people imagine when

• 24:32Shayle Kann?

  they think of what utility operations looks like to a first order, and then you can tell me what it really looks like. I, my guess is that most people think, first of all, it's mostly automated already. And second of all, to the extent it's not automated, that there are, you know, there's like a bunch of people in a, in a NOC with a bunch of big screens who are like watching the screen light up red when there's an outage somewhere, and then they have to go dispatch somebody to go solve it. But that generally, the grid is already fairly automated from an operations standpoint. What is it actually like today?

• 25:06Speaker 03

  I think that's a fair statement. It's, I think the automation today is, is there thanks to field automation. So I think we, traditionally, we have been very concerned about automation inside the control room, but the grid itself in a substation, at a generator site, it's pretty much completely autonomous. And that's, that's why it has been so reliable and so fast, and it balances itself so well. And it's, it's too fast for human in the loop control. So the human in the loop side for operations, it's really monitoring and responding to the abnormalities. So it's alarm management. What happens when things go wrong and the, the operator has to diagnose it.

• 25:50Speaker 03

  It's a lot of rules as well. So the, the, so there's an operational planning phase. Let's, let's traverse the timeline. So operations typically look at, look at maybe a week ahead, two weeks ahead to look at switching procedures. So which lines do I have to take out of service to do planned work? If we can't do planned work, we cannot upgrade the grid. And now here's the sort of the constraint during that time period. The grid is getting so congested that if we do worst-case scenario all the time, that we, we are, utilities are beginning to struggle to find time to take the grid into contingency for plant work. So if we too, if we overload the grid too much, we, we can't work on it.

• 26:35Speaker 03

  That's a big problem. And the other is the grid is changing so quickly, like our, our short-term forecasts are no longer predictable. There's a huge like call for can we start moving towards probabilistic analysis? So instead of, hey, here's what I think the single crystal ball future looks like, or can we prepare for an ensemble of potential futures? And how do we risk adjust so risk inform our decisions? Now utilities are not there yet, but that's within the today's operational planning timeline, which is do one-time studies, switching, that's it. Then you go into real-time side. Real-time, there's not much that they do, but mostly it's around dispatch. So which crews go where?

• 27:20Speaker 03

  Preparing for supplies, lock and tag type of deals. So it's very manual operation, work procedure, real things. So more rules and work coordination than actually balancing the system or recovering the system. And then when things go wrong, then I think Flizzr can kick in, so self-healing switching can kick in. But the control room more or less is mostly alarm management. Now there's a flood of alarms, and that's another emerging practice, which is how can we guide the operator through the flood of alarms coming in? Like, I'm jumping forward to the AI world, but we can learn from the patterns of alarms to help root cause it.

• 28:02Speaker 03

  But today, it's very much a bit of ADMS rules to help see what's important and what's not, and a lot of like human intuition that over, still overwhelms the operator. And then you have the major event side. When a big storm comes, when a wildfire hits, when there's a winter storm. And so how do we do like really emergency type of supplies, dispatch, et cetera, to prepare for those storms?

• 28:29Shayle Kann?

  Yeah, I mean, just based on your description of it, it sounds like maybe there is less to be gained from AI in operations versus in planning. In other words, like if it's already more automated today, setting aside alarm management, extreme events, maybe there's, there's less of a, the ceiling is lower for improvement. Is that true in your mind, or is it just that it's different?

• 28:56Speaker 03

  I think it is the, the complete opposite. I think the potential is even higher when you align. Like, I think today we care so much about connecting data centers and large loads and even generators that we, we put a lot of emphasis on the business case of planning because you are increasing, it's the growth engine. But before this whole large, big electrification, low-growth process, I think the biggest value has been reliability, and there's a lot more to do with that. And then workforce efficiency, workforce efficiencies. So in the, in the operational side, the role of AI is really to bring a lot of those system studies in the past. That's one time, ad hoc, or episodic, or periodic, or reactive, or responsive to customer

• 29:41Speaker 03

  interconnection requests into a continuous, real-time analysis engine. So what is the true like autopilot? What is the true co-pilot? I would expect that from a car, of course, a plane and a spaceship, but can I now expect that from the grid? Can the grid study, read the roads, analyze scenarios, assess risks all the time? And then that's one layer. The one, the first layer, I would say, foundational layer would be the grid analysis and simulation intelligence layer. But the layer on top is there's so much to be gained by having AI learn how the human processes can be improved, how dispatch can be improved, how alarm management can be

• 30:26Speaker 03

  improved. And that the tail end, which we're seeing a lot these days, is can we learn from each day's events? Currently, there is no closed-loop feedback until you have a post-modem. You have a, an event that drives an investigation that needs to be reported back up to FERC.

• 30:48Shayle Kann?

  It takes a year to do or something, anyway.

• 30:51Speaker 03

  And the records are not good. They're not reconciled. They have many sources of truths. None of the recordings from SCADA or PMUs or work orders, they never really sort of paint the common picture. You can't reconcile the history of events from these disparate records, but that's where AI can really come in as well. To reconstruct the daily events, to replay it, and to start putting a closed and continuous improvement loop on how can we execute better, dispatch better, and even plan better. Like, how much is that cable actually being utilized historically? Nobody can really tell you that answer today in planning more operations, especially in distribution.

• 31:36Speaker 03

  And so I think if we can drive that level of closed-loop analytics each day, I think it's going to be a huge game changer.

• 31:43Shayle Kann?

  I think you allude to this, but one thing I've always wondered is the degree to which the planning side or the studies that occur on the planning side and the operations side are talking to each other. Like, as an example, let's say you did a study, a distribution study, let's just say. And that study, and as you said, that study is planning for, okay, what happens in kind of a worst-case scenario? Well, the worst-case scenario, let's say the study says, okay, the worst-case scenario is it's a hot day, and X, Y, and Z happens, and there's a possibility we're going to overload this feeder or something like that, and then we'll have an outage as a result of that. That occurs in the study phase today.

• 32:26Shayle Kann?

  But does that then feed anything into the actual operations phase? Right? Because what you would imagine is that in operations, then what should happen is that the grid should basically be looking out for that scenario, because somebody else inside the utility studied and determined that that's the scenario you need to watch out for. Does that happen historically?

• 32:45Speaker 03

  Of course, each utility might be different, but that is a very, very rare scenario that these study assumptions and results actually feed into operations because they're measured by different results and KPIs and all that kind of stuff. So really, this seems, what happens, it seems at best, I would say, is a set of Excel spreadsheets, as always, that captures the results of the planning study. And then the operational planning, like a week, two week ahead planning, would use those as fixed criteria to inform their models. So these silos between planning and operations, it's not just between these two gigantic departments, which I would call the heart of the utility,

• 33:30Speaker 03

  which is not just connecting people and customer service and all that, but it's really maintaining the capacity adequacy, resiliency of the grid, reliability of the grid. But even planning, as I mentioned, there's like three, four different timeframes that they should look at. There's the long-term planning, like 25 to 40-year-ahead planning. There's a medium-term time, which is great for like generation planning, bulk power integrated planning. Then there is the medium term, which is like five to seven, sometimes up to 10 years for like transmission expansions, right? Because even buying transformers would take that long. And then there is like short-term planning, which is anywhere from, I would say, months ahead to two years, three years ahead,

• 34:13Speaker 03

  which has been traditionally what the key for distribution planning and somewhat interconnections as well. So each time you have the seams, they don't really connect. The forecasts don't connect. The decisions don't really connect. And so I think, and I will say the key point is not really just blaming it on people in the process, but you don't even have an analytical platform that can consume all these different timeframes and trying to piece them together in a coherent timeline. It will sort of overwhelm any traditional software engine. And so in some ways, I would say the utilities are actually adapting their people and processes around the limitations

• 34:58Speaker 03

  of the analytics intelligence and decision-making frameworks that they have. So I think that has to be addressed.

• 35:04Shayle Kann?

  All right, Josh. The last question I want to ask you is a little different, and it's like a hot topic at the moment, hotly debated, on which I think you'll have an opinion, even if it's not exactly in your wheelhouse or what you're doing at Think Labs, which is there are tens of gigawatts of data centers now that are being planned with behind-the-meter generation. And increasingly, because of challenges in grid connection and all the moratoria and stuff like that, that behind-the-meter generation is acting as a bridge to a long-term interconnect. And so these are essentially, they're operating these massive microgrids, or they're planning to operate these massive microgrids, right? And it's comprised usually of some kind of gas generation

• 35:48Shayle Kann?

  or a fuel cell, maybe. There's a UPS, maybe there's a battery. But that's basically it. But they're really, they're operating a microgrid, right? And we've been talking about all of the complexities of operating the macro grid. And I wonder, you know, the debate that I think is raging right now is like, is that harder than it seems, basically? Is it going to be operationally very difficult to run one of these very large loads that probably doesn't, that has like a spiky load profile off of this gigantic microgrid? From what you know about operations of a macro grid, do you think it's going to be easy or hard? I mean, or, you know, what's the nuance here?

• 36:33Speaker 03

  Well, it's a bit of both. But I think we were building microgrids on the distribution side since the 90s and 2000s, right? We understand that they can add a bit of local resiliency, but let's remember why we have a grid in the first place, which is so that we don't become individual islands. And individual islands are always more expensive than sharing the resources that we have. And so, yeah, I think they can use it to protect themselves and to advance their time speed to power, their own capacity requirements, et cetera. But it's not going to sort of serve the greater good by sharing, having centralized generators planned appropriately and serving a variety.

• 37:18Speaker 03

  So it is really being an island. So technically, I think good thing is, yes, you can mitigate some of the immediate capacity constraints, but I think there are existing latent capacities. So let me just make a statement here, which is, I believe the grid has enough existing latent capacity to connect the majority, if not all, of the data centers today. We just need to find where they are and improve the overall processes of planning and operations for the utility. So if we are to build all these generators behind the meter, why not just use those to really upgrade the infrastructure itself to plan better, to operate better? But in the meantime, while the utility is modernizing,

• 38:01Speaker 03

  yes, generators can connect faster if it's a hard capacity that they don't have on that location, that the grid doesn't have in the location. But bringing these additional generators and batteries and fuel cells, et cetera, will complex the other side of the equation, which are the non-steady states. All your transients, all your electromagnetic transients will get way more complicated with all these behind-the-meter resources. So is it really good or bad? Yeah, you might alleviate some constraints, but you might create new ones. So it's not that easy of an answer. And those studies, I would say, would bog down the utility even more.

• 38:42Speaker 03

  Maybe one last opinion on the whole data center and microgrid side is, I really believe that, of course, data centers are driving economic growth, et cetera, for both the data center and the utility, and hopefully the community. But that scale, you're basically building new cities and towns. And so in that case, I really think these data centers should be seen as, I would say, rather than microgrids, I would rather say microutilities. They have a responsibility and they have a capability to reinforce the grid and benefit all those people around them. So can we have similar updated, modern planning and operational processes around these data centers

• 39:24Speaker 03

  as a microutility and not just as a singular customer?

• 39:29Shayle Kann?

  All right, Josh. It took too long to have you on, but glad we did it. Thank you for the time.

• 39:35Speaker 03

  No, thank you, thank you.

• 39:38Shayle Kann?

  Josh Wong is the CEO and founder of Think Labs. This show is a production of Latitude Media. You can head over to latitudemedia.com for links to today's topics. This episode is produced by Max Savage-Levenson. Mixing and theme song by Sean Markwan. Anne Bailey edits the video version of the show. Stephen Lacey is our executive editor. All of our episodes are now on YouTube. Subscribe to Latitude Media for episodes of this show and Open Circuit. You can also find the audio version of this show anywhere you get your podcasts. I'm Shayle Khan, and this is Catalyst.