WEBVTT 1 00:00:00.111 --> 00:00:02.813 It's really hard to make predictions, especially about the future. 2 00:00:03.093 --> 00:00:09.598 It might take 18 to 26 months to build, but new grid infrastructure, we're looking at five to ten years these days. 3 00:00:09.838 --> 00:00:13.881 A couple of months ago, we saw PJM drop 3.8 gigawatts of load. 4 00:00:13.921 --> 00:00:18.304 But a dose of realism when it comes to the physics of building infrastructure always helps. 5 00:00:23.375 --> 00:00:25.076 Can data centers lower your power bill? 6 00:00:25.116 --> 00:00:26.958 Can they really flex when the grid is stressed? 7 00:00:27.138 --> 00:00:31.381 Can utilities develop the grid infrastructure needed for tech that's built in 18 months? 8 00:00:31.642 --> 00:00:34.364 Everyone has a pitch, but few have the actual playbook. 9 00:00:34.464 --> 00:00:39.388 Today, we're joined by Chris Sturgill, grid modernization and clean energy partner at Scott Madden. 10 00:00:39.528 --> 00:00:44.052 And together, we pressure test the hype and give utility leaders the insights that they can actually use. 11 00:00:44.292 --> 00:00:46.974 And we want to give a shout out to Scott Madden for sponsoring this episode. 12 00:00:46.994 --> 00:00:48.975 We'll talk more about them as this conversation goes on. 13 00:00:49.255 --> 00:00:53.696 So Chris, utilities are getting flooded with huge data center load requests right now. 14 00:00:54.156 --> 00:01:01.119 Your data center playbook report mentions developers may show up before they have land, permits, financing, or even an end user. 15 00:01:01.659 --> 00:01:06.801 So in this environment, how do you separate serious projects from those that are just pure speculation? 16 00:01:06.881 --> 00:01:07.821 Thanks, Matt, and you're right. 17 00:01:08.101 --> 00:01:08.762 Utilities are, 18 00:01:10.222 --> 00:01:11.963 beset by all these applications. 19 00:01:12.003 --> 00:01:14.804 And in almost no time, the paradigm really flipped. 20 00:01:14.864 --> 00:01:23.808 So it went from utilities looking to go out and attract new load, because that's a way to show growth, to now being beset by too many applications to even review. 21 00:01:24.748 --> 00:01:36.273 And what utilities are doing to help with this is setting up structured intake processes to vet these projects, both from a commercial perspective and from an operational feasibility perspective. 22 00:01:37.033 --> 00:01:52.617 So in terms of what looks like a good project from a commercial perspective, those that are serious would have an identified project sponsor with a track record of performance, a financing plan with upfront deposits, and a willingness to commit financial resources as the project progresses. 23 00:01:52.857 --> 00:01:54.538 And then looking at it from the technical side, 24 00:01:55.178 --> 00:02:12.675 Those projects that have demonstrated site control, siting with proximity to existing resources, credible load estimates with options for phasing or ramping in power, a schedule with realistic timelines around things like siting, permitting, design, procurement, and all of that. 25 00:02:13.416 --> 00:02:17.099 and other infrastructure coordination when they're looking for water or fiber connections. 26 00:02:17.539 --> 00:02:17.840 Excellent. 27 00:02:17.860 --> 00:02:18.080 Yeah. 28 00:02:18.260 --> 00:02:28.648 And so if you're seeing from a developer, all those things, you know, obviously there's always the question of, you know, how much background is being put into it, how much they can come into fruition. 29 00:02:28.968 --> 00:02:39.837 I'm curious from your experience, are there almost under the radar, red flags or on the other side, green lights that planners can watch for across the lifecycle to really, you know, 30 00:02:40.237 --> 00:02:42.640 get under the hood about whether these things are coming or not. 31 00:02:42.840 --> 00:02:49.446 Yeah, I think the red flags would be those things that are kind of flying in the face of those commercially or technically serious projects. 32 00:02:49.466 --> 00:03:02.579 So it's it's kind of urgency without clarity around the schedule or it's opaqueness or lack of visibility into kind of who the project is, where it will be cited and that they demonstrated site control and all of those things. 33 00:03:02.639 --> 00:03:02.959 So it's 34 00:03:03.580 --> 00:03:10.102 So it's a bit of a diligence exercise in making sure that the developers have dotted their I's and crossed their T's. 35 00:03:10.402 --> 00:03:13.803 So for the utilities involved here, there's obviously a lot at stake. 36 00:03:13.863 --> 00:03:19.325 And I'm curious if you can talk about what are some of the biggest operational risks, financial risks that they're taking on. 37 00:03:20.205 --> 00:03:21.206 Yeah, thanks, Matt. 38 00:03:21.286 --> 00:03:23.288 And they are taking on both those risks. 39 00:03:23.368 --> 00:03:33.737 So on the financial side, it's the Goldilocks question of managing overbuild versus underbuild and missing the opportunity for growth. 40 00:03:34.137 --> 00:03:35.658 On the overbuild side... 41 00:03:36.519 --> 00:03:38.842 You're looking at things like stranded asset risk. 42 00:03:39.262 --> 00:03:44.408 And thus far, that hasn't really borne out because the markets have been so short of power. 43 00:03:45.549 --> 00:03:51.636 But at some point, if everyone keeps building, there may be a tipping point where there's an oversupply. 44 00:03:52.437 --> 00:03:55.200 And if you have counted on off takers of that, 45 00:03:56.301 --> 00:03:59.723 you may end up being long power when you didn't mean to be. 46 00:03:59.943 --> 00:04:11.508 There's also counterparty risk on the financial side, which is that we're seeing a lot of long-term contracts being set up, but if that entity isn't around, you can't sue nobody. 47 00:04:11.728 --> 00:04:19.452 So there's counterparty risk to manage as well, and that's why that commercial vetting that we talked about on the project intake becomes so key. 48 00:04:20.652 --> 00:04:27.736 And then there's also potential for cash flow crunches where utilities are having to commit to infrastructure projects up front. 49 00:04:27.997 --> 00:04:32.639 They get built and then you start collecting revenue from the customer once operational. 50 00:04:32.679 --> 00:04:34.100 And that can be years in the making. 51 00:04:34.180 --> 00:04:42.906 So we're seeing a lot of deposits and kayak and things like that that would help with the cash flow management. 52 00:04:43.233 --> 00:04:43.453 Yeah. 53 00:04:43.733 --> 00:04:52.896 Are there any real world cautionary tales that you can share with us where you've seen things go sideways, whether or not you can actually name who is involved? 54 00:04:53.116 --> 00:04:56.597 Not really on the financial side, but you mentioned operational risk as well. 55 00:04:56.637 --> 00:05:03.259 And I think that's where you've seen a lot in the in the press around data center trips offline. 56 00:05:03.279 --> 00:05:07.260 And then all of a sudden you're seeing gigawatt scale drops in load and. 57 00:05:08.460 --> 00:05:16.523 over frequency over voltage to, um, as a result in, in utilities and ISOs, RTOs and, um, 58 00:05:17.385 --> 00:05:19.626 NERC are looking at standards to address that. 59 00:05:19.666 --> 00:05:28.728 I think a couple months ago, we saw a PJM drop 3.8 gigawatts of load, and that's got a lot of people's attention. 60 00:05:29.308 --> 00:05:40.391 You mentioned sort of the Goldilocks situation where key to all of this is reliably and accurately, as much as possible, predicting future demand. 61 00:05:40.631 --> 00:05:43.172 And for the utility leaders, the cost of 62 00:05:43.832 --> 00:05:47.196 forecasting wrong is it can be astronomical. 63 00:05:47.276 --> 00:05:56.567 So how do how do leaders of the utilities balance that of the need for accurate forecasts with the inherent uncertainty that comes with them? 64 00:05:56.847 --> 00:06:01.533 Yeah, I really sympathize with utility forecasters because for years their their job was 65 00:06:02.623 --> 00:06:06.505 what it was last year, plus one or 2%, and it was pretty stable. 66 00:06:06.765 --> 00:06:12.247 And now it's kind of like the Yogi Berra quote of, it's really hard to make predictions, especially about the future. 67 00:06:12.427 --> 00:06:19.690 Because forecasting is at that front of the planning process, sharpening that process is top of mind. 68 00:06:20.990 --> 00:06:24.212 It goes back to that Goldilocks dilemma we were just talking about, and you're either... 69 00:06:24.972 --> 00:06:31.481 overbuilding with unused infrastructure or you're under building and then running into resource adequacy issues. 70 00:06:31.781 --> 00:06:37.628 And so there are a handful of kind of tactical steps that we've seen people taking to help with this. 71 00:06:38.554 --> 00:06:44.877 The first is around establishing stage-gated forecast categories that apply to the portfolio. 72 00:06:44.897 --> 00:06:55.883 So if a project is kind of just at the outset, has just been taken in, it really doesn't factor into the load forecast in the same way that a more mature project that is under construction would. 73 00:06:56.844 --> 00:07:02.407 And so you're almost qualifying the contribution to the forecast as projects mature. 74 00:07:04.100 --> 00:07:06.566 we're seeing more granular and more frequent forecasting. 75 00:07:06.606 --> 00:07:09.955 So if it wasn't an annual cycle, it might now be quarterly. 76 00:07:12.057 --> 00:07:26.982 and maintaining a consistent source of record or source of truth for the data around these projects so that forecasters and planners and the finance folks and the customer development groups are all kind of singing from the same sheet of music there. 77 00:07:27.882 --> 00:07:40.026 And then finally gets into a feedback loop of making sure that if we have assumed certain load shapes or certain phasing of consumption, that that matches what developers are doing in practice. 78 00:07:40.606 --> 00:07:53.275 One of the reasons that we're seeing data centers and energy really capture a lot of headlines is the intense focus on rates by everybody these days in the political sphere and at the utilities, customers themselves. 79 00:07:53.976 --> 00:07:57.538 And at least within the data center sphere, there's a popular narrative that... 80 00:08:00.320 --> 00:08:05.602 building these data centers can lower power bills by spreading fixed costs over huge sales. 81 00:08:06.122 --> 00:08:12.764 But in your report, you say that high load factor usage alone does not mean a project will necessarily lower rates. 82 00:08:13.364 --> 00:08:20.447 So if that's the case, what actually determines if regular rate payers benefit or get burned from these data centers being built in their backyard? 83 00:08:20.667 --> 00:08:28.950 It's a subject of much debate and really a way that you see the headlines impacting dinner time conversations that you didn't really expect where... 84 00:08:29.870 --> 00:08:38.094 Folks know that I work in the energy industry, so I get a lot of inbound questions about data centers and costs because that's the natural connection to draw. 85 00:08:38.954 --> 00:08:49.619 I think the correlation is there where data centers are a relatively new topic of conversation in the last five years and energy costs have gone up in the last five years. 86 00:08:49.659 --> 00:08:51.580 So kind of thus it must be so. 87 00:08:53.257 --> 00:09:12.197 And I think if you can take it on the simplest ends of this, which is if there was a big data center project coming on that was able to use kind of all existing available capacity, so no incremental costs, it would in fact spread the same fixed costs over more megawatt hours and 88 00:09:13.678 --> 00:09:15.379 put downward pressure on rates for ratepayers. 89 00:09:15.979 --> 00:09:18.520 But I think that capacity has been consumed. 90 00:09:18.560 --> 00:09:24.242 So now we're in the era of incremental infrastructure to support incremental demand. 91 00:09:24.642 --> 00:09:32.305 And the factors that go into that are kind of the headroom, the way the tariff is structured, the consumption pattern of the data center. 92 00:09:33.245 --> 00:09:36.286 and the phasing of how that load comes online. 93 00:09:36.986 --> 00:09:49.810 And that's where utility cost of service studies come in and the need to set up the tariffs in a way that cover all those costs that are directly attributable to the data center and then whatever they are consuming of 94 00:09:51.384 --> 00:09:52.385 Incremental capacity. 95 00:09:52.405 --> 00:10:03.654 And we've seen some interesting data center tariffs come out of Evergy and AEP that try to assign those incremental values and get that math right. 96 00:10:04.794 --> 00:10:12.020 I don't know that we will see the math play out on that in a definitive way for several more years because we're going to have a lot of data. 97 00:10:13.677 --> 00:10:23.191 new assets come on, are getting built, they're hitting the CapEx budgets, and will go into future rates. 98 00:10:23.251 --> 00:10:29.440 So it's almost too early to tell, but that's the balancing act on the math and where utilities are spending a lot of time. 99 00:10:30.101 --> 00:10:32.223 on cost of service studies and tariff design. 100 00:10:32.403 --> 00:10:45.217 And you mentioned, you know, too early to tell, but at the same time, all sorts of stakeholders are looking to build in protections now for consumers, whether that's politicians, consumer advocacy groups, regulators themselves. 101 00:10:46.358 --> 00:10:54.465 And we're seeing a lot of everything from voluntary pledges to mandates being debated, bans coming into certain localities. 102 00:10:55.305 --> 00:11:03.692 I'm curious what concrete protections you think that those consumer advocates should demand upfront to actually ensure that those benefits come to be. 103 00:11:03.892 --> 00:11:06.154 Yeah, it gets back into that tariff design. 104 00:11:06.214 --> 00:11:12.159 So you want provisions that deal with delayed projects or canceled projects, 105 00:11:12.879 --> 00:11:21.661 or where consumption doesn't match what was stated, and so you've overbuilt infrastructure for this project that's now getting spread elsewhere, and early termination, early exit. 106 00:11:21.801 --> 00:11:23.821 So you want to have protections against all of those. 107 00:11:24.341 --> 00:11:35.784 The cost of service piece of it wants to look at both the incremental, almost directly ascribed costs, but also what portion of existing capacity are these projects consuming that – 108 00:11:36.984 --> 00:11:39.948 that they should pay for, that they should, they're the causer of those costs. 109 00:11:39.968 --> 00:11:43.392 So those, and then transparency and reporting and tracking. 110 00:11:43.452 --> 00:11:47.657 So, you know, project said this, that was seven years ago, where are we now? 111 00:11:47.997 --> 00:11:55.546 And understanding if the benefits have, have borne out or if I think, as you said, the rate payers have been burned. 112 00:11:56.086 --> 00:12:01.709 And one of the other sources of consistent tension in this space is timeline. 113 00:12:01.769 --> 00:12:09.934 Data center, it might take 18 to 26 months to build, but new grid infrastructure, we're looking at 5 to 10 years these days. 114 00:12:10.014 --> 00:12:15.577 So how do utilities look to close that timeline gap when it's so wide? 115 00:12:15.777 --> 00:12:19.318 It is a wide gap, and it's probably not one they'll ever close fully. 116 00:12:19.459 --> 00:12:21.360 I think that the answer to... 117 00:12:22.240 --> 00:12:26.722 when would you like this power demand was, or when, and how much would you like this power demand is all of it. 118 00:12:26.802 --> 00:12:37.086 And two years ago, you know, if you think about all the steps we've been talking about so far around forecasting and planning and project vetting and contracting, all of that gets you to a point. 119 00:12:37.166 --> 00:12:43.609 And then you have to actually go design and procure all the material for and build the infrastructure to serve that. 120 00:12:44.409 --> 00:12:45.970 And, and that's non-trivial. 121 00:12:46.410 --> 00:12:46.510 Um, 122 00:12:48.055 --> 00:13:15.655 What we see utilities doing to try to close that gap is looking at things like standardized designs that mitigate challenges with kind of one-offs and a lot of back and forth on the design process between utility and developer, looking at breaking it out, breaking out the work and the capacity delivery in phases so that as they are able to build infrastructure and bring more power to bear, the data center can start ramping up its load. 123 00:13:17.316 --> 00:13:38.873 looking at contracting arrangements for long lead time equipment, sometimes polling, sometimes kind of strategic sourcing agreements, looking at addressing labor shortages, both on the design side and the build side with EPC relationships, and looking at kind of change control procedures so that 124 00:13:40.104 --> 00:13:50.740 As the design matures and if it changes, you can quickly determine whether or not that's going to result in a timeline shift because new equipment's needed or a new component has been added. 125 00:13:50.860 --> 00:13:53.043 So there's lots that go on there. 126 00:13:54.050 --> 00:13:59.251 And again, none of that probably brings you to a point of 18 months where it takes it for the data center. 127 00:13:59.791 --> 00:14:04.492 But I think that can be, it all helps. 128 00:14:04.652 --> 00:14:11.654 And then you're working on the relationship with the customer and helping them know what to expect at every point in the process. 129 00:14:11.674 --> 00:14:13.414 So they don't expect magic. 130 00:14:13.454 --> 00:14:16.175 They're expecting reasonable build of the project. 131 00:14:16.235 --> 00:14:26.237 Yeah, that's what I was going to ask is part of the onus on the tech companies who are wanting the data centers to have more realistic expectations and maybe meet those utilities halfway. 132 00:14:26.457 --> 00:14:26.997 I think so. 133 00:14:27.157 --> 00:14:33.618 They may get creative and decide that that's not working fast enough and they're going to look for kind of macro grids and building it themselves. 134 00:14:33.698 --> 00:14:36.678 And they've certainly got the balance sheets to do that. 135 00:14:36.779 --> 00:14:43.020 But a dose of realism when it comes to the physics of building infrastructure always helps. 136 00:14:43.340 --> 00:14:43.820 Yeah, yeah. 137 00:14:43.860 --> 00:14:44.840 Physics aren't flexible. 138 00:14:47.089 --> 00:14:58.015 The other story we often hear as kind of pushback to the data center criticism as ways that they can be good stewards of the grid comes from flexible load. 139 00:14:58.095 --> 00:15:08.222 The idea that as a data center can shift their compute power time-wise, they can run backup batteries, they can curtail completely if needed when the grid is strained. 140 00:15:09.042 --> 00:15:12.926 How much of that is actually real and being used today? 141 00:15:12.946 --> 00:15:16.309 And how much of it is maybe what we can call pitch deck optimism? 142 00:15:16.710 --> 00:15:19.432 Here in Raleigh, North Carolina, I'm just down the road from Duke University. 143 00:15:19.452 --> 00:15:21.875 I put out a really interesting study on this. 144 00:15:21.915 --> 00:15:27.980 And there are potentially gigawatts of kind of available capacity if you can start shifting around load. 145 00:15:28.761 --> 00:15:35.871 And I think that's attractive for kind of obvious reasons in terms of avoiding the cost and time of building all of that. 146 00:15:36.212 --> 00:15:38.535 But I so the idea is sound. 147 00:15:38.555 --> 00:15:39.216 I think the. 148 00:15:40.947 --> 00:15:59.236 to date it has largely been kind of a paper exercise and pitch deck optimism is as you might have said it certainly has drawn interest in groups trying to figure this out like EPRI's DC flex and and working consortiums I think one important thing to keep in mind here is the 149 00:16:02.437 --> 00:16:17.332 an operational time horizon, which is things like demand response, that utility operators have the purview and the mandate to kind of control things in the moment and call upon demand response resources or curtail loads if they need to. 150 00:16:18.273 --> 00:16:27.476 But that's very different than the planning time horizon where utility planners are saying, I know that this resource will be here as stated in the future. 151 00:16:27.996 --> 00:16:41.341 And that's where I think the longest gap is in terms of getting operational flexibility stood up in a meaningful way is that planners will probably not count on that kind of capacity until there's... 152 00:16:43.279 --> 00:16:50.645 some performance track record of those resources showing up in that way, that there are some contractual mechanisms to make sure they do or penalties if they don't. 153 00:16:51.526 --> 00:16:58.632 And that gets back to grid stability and grid safety and probably not something they're going to budge on too readily. 154 00:16:58.973 --> 00:17:03.036 And then the data center side of that is, well, I don't know how often you're going to call it. 155 00:17:03.156 --> 00:17:04.297 I don't know for how many hours. 156 00:17:04.498 --> 00:17:06.980 I've got a significant opportunity cost when you do. 157 00:17:08.621 --> 00:17:11.562 if I've got to disrupt my computational workloads. 158 00:17:11.642 --> 00:17:15.423 So I think both parties are pretty wary about that as a real solution. 159 00:17:15.683 --> 00:17:19.504 And it'll take some, I think, successful proof of concept to get there. 160 00:17:19.684 --> 00:17:20.384 Yeah, that makes sense. 161 00:17:20.444 --> 00:17:31.847 And you mentioned there's different time horizons, but I imagine there's also some variation in how applicable it is based on geographic region or the local grid mix or even the type of data center. 162 00:17:32.427 --> 00:17:32.927 It could. 163 00:17:33.047 --> 00:17:36.488 The type of data center and what kind of loads they are willing to shift or... 164 00:17:37.688 --> 00:17:44.072 or can't because it's a long training run kind of workload, that certainly plays into it. 165 00:17:44.832 --> 00:17:49.315 The ramp of your generation mix and how quickly you can ramp that up or slow it down. 166 00:17:49.635 --> 00:17:49.995 Excellent. 167 00:17:50.035 --> 00:17:54.418 So Chris, we want to end the conversation on an actionable point. 168 00:17:54.458 --> 00:18:03.943 So if you were in the C-suite, you're steering a utility through this wave over the next five years, what's the one biggest mistake you'd be trying to avoid right now? 169 00:18:04.323 --> 00:18:05.904 I think it's... 170 00:18:06.938 --> 00:18:08.679 locking into something prematurely. 171 00:18:08.739 --> 00:18:13.523 So the biggest thing right now is just these great cones of uncertainty and where things are going to go. 172 00:18:14.143 --> 00:18:29.774 And so I'd want to maintain kind of nimbleness in my processes, probably most of all, that link between the capital planning and budgeting forecast and the actual operational kind of infrastructure planning processes and making sure that 173 00:18:30.735 --> 00:18:32.196 I'm staying on top of my projects. 174 00:18:32.236 --> 00:18:37.940 I'm seeing what materializes and when and how quickly I'm able to build and respond. 175 00:18:38.340 --> 00:18:45.225 And then also, if things start to pull back, not being so far kind of out over the edge on that, that you get burned financially. 176 00:18:45.265 --> 00:18:50.289 It's a lot to juggle, but it sounds like it's, you know, the map is out there for how to do it. 177 00:18:51.449 --> 00:18:52.250 It's an exciting time. 178 00:18:52.270 --> 00:18:52.630 Thank you. 179 00:18:53.251 --> 00:18:54.011 It is, it is. 180 00:18:54.172 --> 00:18:55.553 And it's a critical time. 181 00:18:55.593 --> 00:18:59.536 Again, we're talking about this a lot now, but we know it's only going to ramp up. 182 00:18:59.596 --> 00:19:05.460 So we thank you for joining us to share your insights in this fast moving space today. 183 00:19:05.821 --> 00:19:06.041 Great. 184 00:19:06.201 --> 00:19:06.621 Thank you, Matt. 185 00:19:06.661 --> 00:19:07.202 Pleasure to be here. 186 00:19:07.522 --> 00:19:10.344 And thanks as well to Scott Madden for sponsoring today's episode. 187 00:19:10.424 --> 00:19:14.966 Scott Madden is a management consulting firm serving clients across the energy utility ecosystem. 188 00:19:15.547 --> 00:19:25.372 Areas of focus include transmission and distribution, the grid edge, generation, energy markets, rates and regulations, corporate sustainability, and corporate services. 189 00:19:25.812 --> 00:19:29.014 And thanks as well to our audience for joining us today on Power Perspectives. 190 00:19:29.274 --> 00:19:30.315 We will see you next time.