Powering AI: Data Centers, the Grid, and the Ratepayer
Guest: Commissioner Jehmal Hudson, Virginia State Corporation Commission; First Vice President, NARUC
Host: Michael Santorelli
Runtime: ~36 min
Episode summary
Commissioner Jehmal Hudson joins Strategic Voices to unpack the regulatory questions raised by the rapid growth of data centers and AI infrastructure. Drawing on his vantage point in Virginia — home to the world’s largest concentration of data centers — and his incoming presidency of NARUC, Hudson frames his role as bringing balance and clarity to a fast-moving debate. The throughline is a single question regulators keep returning to: will the lights stay on, and will bills stay affordable?
The conversation moves from the scale and speed of unprecedented load growth to the principle of cost causation and how states are protecting existing ratepayers through dedicated large-load rate classes, minimum demand charges, and financial assurances. Hudson and Santorelli also explore self-provisioning by hyperscalers, water and cooling demands, the fiber and broadband dimension of digital infrastructure, affordability mechanisms, the federal–state jurisdictional balance at FERC, and where the conversation is likely headed.
Show notes & timestamps
| Time | Topic |
|---|---|
| 00:12 | Introductions — Commissioner Hudson’s background and the data center debate |
| 02:15 | The regulator’s role: balance, clarity, and the public-interest lens |
| 04:56 | Why this moment is different — scale and speed of load growth |
| 07:17 | Ratemaking and cost causation: who pays for the infrastructure? |
| 09:13 | Should hyperscalers self-provision their own power? |
| 11:56 | Microsoft’s Nevada ratepayer-protection tariff; emerging state approaches |
| 15:46 | Sharing best practices across states; grid modernization and reliability |
| 19:09 | The water and cooling dimension |
| 21:38 | Broadband and fiber — the interconnected nature of digital infrastructure |
| 24:12 | Could data centers contribute to affordability programs (a “data center PIPP”)? |
| 28:07 | The federal role: FERC’s pending large-load rulemaking and state jurisdiction |
| 30:57 | Where the conversation is headed over the next year or two |
| 33:00 | Lightning round: who Hudson reads, who to talk to next, and fun |
Quotable clips
“When people hear terms like gigawatts and interconnection queues, what they really want to know is: will my lights stay on, and will my bill stay affordable?”
— Commissioner Hudson, 02:15
“The question is not whether growth is good — it’s how we structure growth responsibly.”
— Commissioner Hudson, 07:17
“Costs should generally be borne by the customers creating those costs.”
— Commissioner Hudson, 07:17
“This is the first time I’ve experienced so many issues colliding at once.”
— Commissioner Hudson, 17:18
“No single stakeholder sees the entire picture.”
— Commissioner Hudson, 34:32
Transcript
Lightly edited for clarity. Timestamps reflect the original recording.
Michael Santorelli · 00:12
Welcome to Strategic Voices. I’m your host, Michael Santorelli. Today’s guest is Jehmal Hudson. Jehmal is a commissioner on the Virginia State Corporation Commission, which regulates, among other things, the state’s electric and water utilities. Prior to being elected to the commission, Commissioner Hudson was Vice President of Government Affairs at the National Hydropower Association. Before that, he was a regulatory practitioner for ten years, managing public policy issues at the Federal Energy Regulatory Commission, or FERC.
Commissioner Hudson also currently serves as the First Vice President of the National Association of Regulatory Utility Commissioners, or NARUC, which is the national voice for state regulators. So welcome, Commissioner. Thanks for being on.
Jehmal Hudson · 00:56
Thank you very much for having me.
Michael Santorelli · 00:59
Our topic today is data centers, and that’s a big topic that seems to be increasingly controversial. On the one hand, people are embracing AI in some fashion or another — I’m using it more, and everyone I talk to seems to be using it more. On the other hand, it seems that few people want a data center in their community, and data centers are the physical manifestation of the infrastructure needed to support AI.
Your home state, Virginia, has the most data centers of any state in the country, and many more are scheduled to be built, with many clustered in Northern Virginia, or “Data Center Alley.” But as this digital infrastructure grows in other states where people are less familiar with it, there’s opposition, uncertainty, or a feeling of unease whenever these issues arise. There was a Gallup poll a couple of weeks ago that found seventy percent of people oppose data center construction in their communities. So, there are a lot of moving parts around this debate. What role are you playing in these important discussions?
Jehmal Hudson · 02:15
One of the things I try to do — both as a commissioner at the Virginia State Corporation Commission and as the incoming president of the National Association of Regulatory Utility Commissioners — is to help bring balance and clarity to a conversation that’s moving very fast. Data centers and AI are not hypothetical anymore. They’re already reshaping electricity infrastructure planning, utility investment decisions, and transmission needs, along with questions around affordability and reliability.
Virginia sits at the center of many of these discussions because, as you mentioned, Northern Virginia is one of the largest concentrations of data centers in the world. So, our commission is seeing these issues in real time. My role is not to decide whether AI is good or bad. My role is to ask what this means for reliability, what it means for affordability, who pays for the infrastructure, whether risks are being allocated fairly, and whether we are planning prudently for long-term growth.
I also spend a lot of time trying to translate highly technical energy issues into language the public can understand. When people hear terms like gigawatts, interconnection queues, and resource adequacy, what they really want to know is: will my lights stay on, and will my bill stay affordable? That’s ultimately the public interest lens regulators have to bring to this conversation.
Michael Santorelli · 03:55
Thank you. As you said, there are any number of technical parts to this. As laypeople and members of the public, we hear a lot about the electricity needs of data centers — that seems to be a big part of your focus — but also about how much water they need to cool their chips, and how these needs seem to be novel and much bigger than previous large-scale industrial customers.
The news cycle ebbs and flows with stories about people’s bills going up, or a community’s water being severely impacted by a local data center. From your perspective, being involved in these discussions, are these accounts true in terms of the very large needs? Or are we, as the public, just hearing the worst-case scenarios?
Jehmal Hudson · 04:56
Starting at a high level, I do think the scale and speed of projected electricity demand growth is what makes this moment different. For many years, electric load growth in much of the country was relatively flat. Utilities planned around modest increases in demand. Now, suddenly, in some regions we’re talking about extremely large loads arriving in very compressed time frames.
Data centers are part of that story, but we also can’t forget about electrification, advanced manufacturing, reshoring, and broader economic development. The challenge is that the electric system can’t be expanded overnight. Building generation, transmission lines, substations, and distribution infrastructure takes a lot of time, a lot of capital, permitting, and coordination.
So, regulators across the country are asking several big questions: Can the grid keep up? How do we maintain reliability? Are we overbuilding, and can we avoid that — or are we underbuilding? But the most important question is how we ensure residential customers are not unfairly subsidizing infrastructure built for extremely large users. That balancing act is becoming one of the defining regulatory questions of this era.
Michael Santorelli · 06:32
On that affordability point — another thread running through these discussions — having followed a lot of the regulatory conferences you’ve spoken at, it seems data centers are always on the agenda, and so is affordability. This goes to the fundamental nature of ratemaking as it’s been for more than a hundred years. Could you give an overview of the interplay between the investment needed, the use by the data centers, and how that might impact customers under the current framework if things didn’t change?
Jehmal Hudson · 07:17
One of the things I want to emphasize is that economic development and ratepayer protection are not mutually exclusive goals. I always say the question is not whether growth is good — it’s how we structure growth responsibly. At the State Corporation Commission, we carefully look at cost allocation, infrastructure timing, interconnection requirements, reliability impacts, load forecasting, and tariff structures specifically for large customers.
One important principle I always stress is cost causation — meaning costs should generally be borne by the customers creating those costs. So if major infrastructure upgrades are required primarily to serve large load, regulators have to ask: who pays for these upgrades? How much risk should existing customers absorb? And what protections should be in place if the projected load never materializes?
Those are practical questions regulators are wrestling with across the country. At the same time, these facilities can bring real benefits: increased tax revenue, broader economic development, and potentially better utilization of the infrastructure. In some cases, a large customer base over which fixed costs can be spread can be helpful too. One of the things I’m learning is that this is not a one-sided conversation — the challenge is finding the right balance.
Michael Santorelli · 09:13
Following these conversations, especially at NARUC — where, as you said, you’re the incoming president and currently First Vice President — you’re involved in many of these discussions. It’s been encouraging to see all stakeholders at the table: regulators from different states, the tech companies funding much of the construction, and the electric and water utilities providing service.
Building on cost causation, one potential solution coming up more and more is to let these hyperscalers and tech companies self-provision their own power — to have their own power plants on site. This seems to be a growing trend in some states. Is it a good idea? Is it the extreme of the cost-causation principle? Could allowing hyperscalers to self-provision insulate a normal ratepayer like me from increased bills, or are there downsides to that approach?
Jehmal Hudson · 10:48
Thank you for that question. Regulators are open to discussing all options, but the details really matter. There are arguments that self-supply or “bring your own generation” models could reduce pressure on existing systems and potentially insulate traditional ratepayers from certain costs.
But there are questions regulators have to ask: How reliable are those resources? Are they available during peak system stress? Who pays for the backup service? And what happens if those facilities still rely heavily on the grid? It’s always about fairness, and how we maintain fairness across customer classes. I don’t think there’s a one-size-fits-all answer. What regulators generally want is transparency, clarity around cost responsibility, and reliability protections for the broader system.
Michael Santorelli · 11:56
Right. Last week I saw that in Nevada, Microsoft had filed a new kind of tariff. I don’t follow this as closely as you do, but it seemed novel — a ratepayer-protection tariff that would, generalizing here, let them self-provision but also pay to offset their impacts on the grid, or to cover their share of their use of the grid. Have you seen that, or heard about that general concept? Is it a potentially good balance?
Jehmal Hudson · 12:36
Not only in Nevada — you have a lot of other states, such as Georgia, Indiana, Pennsylvania, and Virginia. I won’t call them best practices, but there are a lot of commonalities emerging as we approve large-load tariffs. One thing I’m seeing is the creation of a separate rate class for large-load customers, and it’s customer-agnostic. It’s not geared solely toward data centers — it’s for any large-load customer, say with more than twenty-five megawatts of demand and a load factor above seventy-five percent. That encompasses some industrial customers as well.
In addition, we’re seeing dedicated large-load tariffs put in place, and minimum demand charges. In Virginia, as an example, we’re asking the customer — particularly the data center — to pay for eighty-five percent of transmission and distribution costs and sixty percent of generation costs, regardless of usage. This isn’t in every state, but it is in Virginia. We’re also seeing more financial assurances, such as long-term contracts of maybe fourteen or fifteen years, with withdrawal penalties and exit fees if they leave before then.
The most important piece is making sure there’s customer protection on the infrastructure itself — no cross-subsidization between residential customers, existing customers, and large-load customers. For Virginia and the other states I mentioned, we’re going down the right path. But we also have to be sensitive to the fact that different regions are handling this very differently.
Jehmal Hudson · 15:01
If you’re Virginia, Georgia, or Indiana, you’re seeing a lot of load growth. But take Vermont — they’re not experiencing it yet, but they want to prepare for it, so they’re trying to capture lessons learned. Some states are part of regional transmission organizations, and some are not. Some have vertically integrated utilities and some do not. Everyone is trying to find their right path to solve these problems. The most important thing is having the opportunity to solve it the best way you can and not being forced into a one-size-fits-all model.
Michael Santorelli · 15:46
Right. It seems like a big goal of the NARUC conversations is to get everyone together and share best practices. As you said, what works in Virginia might not work in Texas or California, for any number of reasons. But it seems there’s an opportunity, at a higher level, to talk about grid modernization and reliability. We hear all the time about the grid’s age showing, and the pace at which data centers have arrived seems like another example of the strain on the current grid. Storms and other disasters have shown that, while the grid isn’t unreliable, there are instances where its age shows. Has the notion of modernizing the grid been woven into these discussions about ramping up generation?
Jehmal Hudson · 17:18
Absolutely — I think it already is. One reason these conversations are so important is that they reveal the broader stresses that already exist: aging infrastructure, long permitting timelines, transmission constraints, and interconnection delays. Under our previous president, Erin Dull, we always championed looking into extreme weather risk. And of course, there are the recent questions about affordability. Data centers didn’t create these challenges, but they are accelerating the attention around them.
Regulators are increasingly discussing how to define customer value beyond simply the lowest possible price. Customers do care about reliability, resilience, speed of service, power quality, economic opportunity, and — increasingly — infrastructure readiness. That doesn’t mean affordability becomes less important; it means the conversation is becoming more multidimensional. When I started this job six years ago as a judge, we dealt with issues in waves — one challenge, then another. This is the first time I’ve experienced so many issues colliding at once, and it’s our job to do our best to solve these multidimensional challenges.
Michael Santorelli · 19:09
It seems to be coming all at once, and, as you said, data centers aren’t causing it — they’re just caught up in this huge wave. Another issue that’s part of this but gets less focus is water. A lot of the discussions at NARUC and elsewhere are about electricity, which is of paramount importance, but water is also implicated, because many data centers require significant water to cool chips that get very hot running constantly. Speaking from the NARUC perspective, how involved has water been in these conversations? Is it on equal footing with electricity, or is it more of a secondary part of the conversation?
Jehmal Hudson · 20:25
That’s a great question. The concerns about water usage are very real, but nuance matters. Some projections are extraordinarily large, and in certain cases you have proposed loads unlike anything utilities have historically seen or planned for. But not every data center is identical. Different facilities use different cooling technologies, different chip architectures, different operating models, and even different levels of energy intensity.
Sometimes the public discussion jumps immediately to the worst-case scenarios. That said, the real reason regulators take these projections seriously is that utilities have to plan for reliability — we can’t assume future efficiency breakthroughs will arrive on schedule. So, when it comes to water, we plan based on what we know today while remaining flexible about how we adapt as the technology evolves.
Michael Santorelli · 21:38
Another issue implicated by the rise of data centers and AI is broadband, which has been a big focus of mine for a long time. These data centers need a tremendous amount of fiber-optic capacity to transport traffic to and from them. State-level oversight of broadband is significantly different than it is for electricity and water — it’s very light-touch, or nonexistent, in many states. Have you thought about broadband or fiber-related issues in your work, or is it all energy and a little water all the time on data center stuff?
Jehmal Hudson · 22:25
It’s certainly an important issue we have to follow, because all of these issues are interconnected. Data centers aren’t just electricity-intensive; they’re also heavily dependent on fiber infrastructure and communications networks. At a recent event, everyone was asking how Virginia became Data Center Alley and the data center capital of the world. It goes back to the nineteen-nineties — and I may be dating myself here — when there was an internet company, now a web portal, called America Online. It was headquartered in Vienna, Virginia, in what is now Tysons Corner. That led to a large concentration of fiber optics there.
More broadly, as utility regulators we increasingly have to think across all sectors: energy, telecommunications, water, and also cybersecurity and infrastructure resilience. At NARUC especially, we’ve had more conversations about the convergence of infrastructure systems, because modern digital infrastructure depends on all of those systems working together. So while energy often dominates the headlines, broadband and communications infrastructure absolutely matter in this conversation.
Michael Santorelli · 24:12
I remember AOL — I remember the CDs in the mail — but I’d forgotten about that tie-in with Northern Virginia. That makes a lot of sense. What’s interesting about the broadband–data center–energy overlap is that some have talked about — and this ties to the notion of hyperscalers footing the bill for their impacts, self-provisioning, and so on — having tech companies contribute to something like a universal service fund to address affordability in broadband, given the amount of traffic they pass over networks, especially in the age of AI. That’s been an ongoing conversation as part of the broader effort to reform the Universal Service Fund in telecom. But I’ve seen others try to extend that into the energy space — not only looking to hyperscalers to cover their grid impacts, but also to contribute to affordability in the utility space. Has that come up at all?
Jehmal Hudson · 25:52
To your point, there have been more and more conversations about how data centers can help with affordability, and I’ll say this on two points. By the sheer nature of data centers, there are arguments that they can put downward pressure on rates, which helps spread costs among all customers. While I haven’t seen anything official, there have been conversations about whether data centers can do more.
In Virginia — and I believe Ohio has this as well — we have a Percentage of Income Payment Plan, or PIPP. If you’re below a particular poverty threshold, you only pay a set percentage of your bill. To do that, other ratepayers have to cover the difference. So, there’s been talk of whether there could be a data center version, where they pay a large portion of that. But going back to my earlier point, as regulators we always make sure cost causation is part of our decision-making. If data centers and hyperscalers are driving the infrastructure, we want to make sure they pay a great deal of that cost. We’re stepping in the right direction with separate rate classes and making sure they pay a good portion. On affordability specifically, while I’m hearing things, I haven’t seen anything formal yet. It remains to be seen — it’s still very early.
Michael Santorelli · 28:07
You’re a state regulator working in Virginia, and a lot of energy regulation happens at the state level. But the federal government also has a role — we see it in the administration setting AI policy and goals, and Congress theoretically has a role on so many of these issues. And at FERC, where you used to work, FERC looks more at interstate issues. I’ve heard FERC might be getting ready to issue a ruling on large-load issues. What are you expecting from that, and from your perspective as a state commissioner and as part of NARUC, are there things you’re looking for, or potential bad or good outcomes?
Jehmal Hudson · 29:08
To your point, the Federal Energy Regulatory Commission recently indicated at one of its commission meetings that there should be a decision on the advance notice of proposed rulemaking that was originally supposed to be announced back in April. So everyone is waiting with bated breath to see what that final decision will be.
I commend FERC, because they are facing very difficult questions — large-load growth is increasingly affecting regional transmission planning, interconnection processes, and wholesale market operations. NARUC has been very active in these discussions, particularly around preserving the proper balance between federal and state jurisdiction. One issue state regulators really care about is maintaining state authority over retail ratemaking, distribution systems, integrated resource planning, and resource decisions, which traditionally have been state-level responsibilities.
Most state regulators recognize that coordination with FERC is essential. But states also want to ensure federal actions don’t unintentionally override traditional state responsibilities. So there’s a strong interest in collaboration, while also paying close attention to the jurisdictional boundaries.
Michael Santorelli · 30:57
Looking ahead, these issues seem to be a permanent part of the conversation now — they don’t seem like they’re going away anytime soon. Do you see a point where we’ll solve these issues, or is it an evolving conversation that continues as things change? Projections are all over the place depending on which chip you’re using. I’ve seen in the news that Microsoft has a more efficient cooling system, and Google’s chips use less energy, so things might trend downward over time. But as AI usage goes up, we might need more anyway. Where do you see the conversation going over the next year or two?
Jehmal Hudson · 31:53
I do think these issues are becoming a permanent part of utility regulation. The specific technologies may change — today it’s AI and data centers; tomorrow it could be something we don’t expect. But the broader themes are here to stay: infrastructure investment, reliability, affordability, technology change, resource adequacy, and balancing innovation with consumer protection.
What’s going to change is the pace. Regulators are increasingly being asked to make decisions in environments with more uncertainty, faster technological change, and larger capital requirements than we’ve historically seen. So, adaptability is going to become a critically important trait.
Michael Santorelli · 33:00
As we wrap up, I wanted to go through a quick lightning round. On data center issues, who do you read or consult when you’re thinking through all of this?
Jehmal Hudson · 33:15
I call them the bread-and-butter places to get information: NARUC, of course, and FERC. Because Virginia is part of the PJM region, anything from PJM. For reliability, I look to the North American Electric Reliability Corporation, or NERC. And one last piece that’s been very helpful, though not directly related, is the Energy Information Administration, the EIA.
I’d also encourage people to read utility commission orders. I try to read what’s happening in other states — the kinds of decisions they’ve made — as well as technical conference materials. They may not always be exciting to read, but that’s where many of the real policy debates and decisions are happening right now.
Michael Santorelli · 34:19
It’s also a great input for AI to summarize, potentially. Who should I talk to next about these issues, do you think?
Jehmal Hudson · 34:32
I have a whole list of people. Grid operators are a great place to start. So are consumer advocates — the National Association of State Utility Consumer Advocates, NASUCA — along with utility planners, technology companies, and transmission experts. I’m honored to speak on behalf of all the state regulators, but talking to regulators from different regions — New England, the Southeast, the Midwest, the West — is really helpful. And reliability organizations: not just NERC overall, but NERC’s several regional entities. The important thing about this issue is that no single stakeholder sees the entire picture. So reaching out to those people would be a great place to start.
Michael Santorelli · 35:37
And the final question. It seems like all you do these days is talk about data centers — what do you do for fun when you’re not talking about data centers?
Jehmal Hudson · 35:46
That’s the million-dollar question. Right now, what I’m doing for fun is watching the New York Knicks do a great job getting to the NBA Finals. At least for the next couple of weeks, that’s my number one hobby. It’s the first time since nineteen ninety-nine that they’ve been to the Finals, and if they win, it’ll be the first time since nineteen seventy-four. It’s fun — but it may not be so much fun if they don’t win.
Michael Santorelli · 36:15
Well, as someone based in New York, that’s a great answer and a great way to end the podcast. Thank you very much, Commissioner Hudson, for coming on.
Jehmal Hudson · 36:25
Thank you very much. It’s truly been a pleasure. Thanks again.