Powering Progress Episode 1 – Texas at the Edge of the Grid: Powering the Age of AI

Aubrey Bishai [00:00:13]

Welcome back to Powering Progress, a podcast series exploring the infrastructure, energy, and market dynamics shaping the next generation of data centers and AI development. I’m your host, Aubrey Bishai, Chief Innovation Officer at Vinson & Elkins.

As AI-driven demand for power accelerates, Texas has emerged as one of the most important markets to watch. With a unique regulatory structure and the ERCOT grid at its center, the state is effectively becoming a testing ground for how to meet large-scale, power-intensive demand in real time.

In this episode, we’re joined by Jaren Taylor, who works closely on Texas energy policy and power market dynamics with a focus on how the grid is evolving to support growing demand from data centers and other large loads.

Together, we’ll explore what’s happening on the ground in Texas and what developers, investors, and operators across the country can learn from it.

Welcome, Jaren. Let’s dive right in and talk about Texas. So Texas is at the center of data center growth right now. What is it about ERCOT and the Texas market structure that makes it so attractive for these projects?

Jaren Taylor [00:01:48]

There’s probably several things about Texas and ERCOT that make it attractive to data center development. I guess you would have to start with the fact that ERCOT has its own electric grid, which inherently makes it unique from every other spot in the United States.

With its own electric grid, it has local control of how entities are able to interconnect to it, how they’re able to be served, and you kind of avoid the federal politics that are involved in other areas. That really provides, I guess, a couple benefits to the state in that one, you’re able to interconnect faster and two, those entities and when changes need to happen, like are happening right now for data centers, the state’s able to be a little more nimble and adapt policies and regulations to fit the needs of any particular business.

I guess another thing that’s unique to ERCOT is ERCOT has an abundance of very low-cost power. The state is blessed to have immense wind resources and immense solar resources. And while ERCOT would tell you that it would love to have additional dispatchable generation, if you look at the ERCOT web page, basically all but maybe five days out of the year, you’ll see that ERCOT actually has substantial unused capacity — and when I’m referring to capacity here, I’m talking about generation capacity.

There’s substantial additional megawatts that can be conserved in Texas that are not being conserved in Texas. And most of those are very low cost, variable resources like wind and solar. So that’s probably a second key issue to the ERCOT grid. Another thing that makes Texas unique from the development standpoint is ERCOT can build electric infrastructure faster than anywhere else in the nation. It just takes a quick look at what utilities have built the most transmission over the last several years. And you’ll find that the Texas utilities are at the very top of that list. That comes down to, again, local control.

Transmission and its development in Texas is regulated by Texas for Texas businesses. There are no federal interconnection issues. There’s no FERC overlay to those things. And Texas generally is a state that wants to encourage business and it wants to have growth and it wants to have development. And when those policies and regulations are working in harmony together, you can get things done quickly. The last thing that probably makes Texas a massive market for data centers is its geography and fiber resources. And here we’re talking more telecommunications type fiber, what allows data transmissions.

South Dallas, for example, is a good case study in this. South Dallas is probably the hottest area in the entire state for data center development, and that is primarily due to its access to the fiber optic networks that are in place. This reduces latency for data centers. It allows them to both receive and transfer data very quickly. It’s an added element that exists to help these industries serve their end use clients, which are primarily large tech companies.

Aubrey Bishai [00:05:00]

Certainly one of the biggest issues we’re hearing about is speed to power, as you’ve just alluded to. How is Texas addressing that and where are the bottlenecks today?

Jaren Taylor [00:05:10]

I’d say we’re working on it. We definitely don’t have it solved at this point, but there’s a lot in the works in the state where we are trying to address that speed to power concern. I mean, frankly, when it comes to data centers, speed to power is the entire ballgame. It is how much power can you get them and how fast can you give it to them? Because power is the mechanism by which they do their business. They need to consume power to provide their product to their end user.

There are many bottlenecks today in this process that we could discuss, but maybe before really diving into bottlenecks, let’s talk about what makes data centers such a difficult puzzle for electric grids to solve.

One of the problems we have is everyone wants to be a data center because they’re incredibly profitable businesses if you can get them up and off the ground. This leads to a clog in the interconnection queue, or too many entities who are seeking to interconnect that have bogged down the existing system.

The problem that the utilities have and that ERCOT has in that process is how do you as a utility, someone who is going to be providing services to the data center industry, how do I know which of these data centers or which of these potential projects is real? Who can obtain the financing? Who can actually deliver and land an end-use client? How do we determine whom we should spend time and money studying to bring into the electric grid? That is one of the fundamental bottlenecks.

The data center loads are massive. They consume a ton of power. For example, I worked on one a couple weeks ago that is going to require more power than the city of Arlington. That’s where the Cowboys play. That’s where the Texas Rangers play. And also about half a million Texans. That one building that will be contained within that four slab walls will consume more power than that entire city consumes. That’s a problem that makes the elements on the electric grid that are going to serve them be heavily taxed. And those are study complications.

Other reasons why data centers are a difficult puzzle for the electric grid is the data centers can be built very fast. In Texas, we use a lot of slab wall construction. You pop 4 walls up, you pop a bunch of servers up, you put some cooling systems in, and all of a sudden you can operate a data center. So the data centers can be built and up and operational much faster than you can build the electrical infrastructure is necessary to serve them. So you kind of have a misalignment of timing that can create problems. And the last is the electric characteristics of a data center in general.

One of the problems with large electric loads like data centers is that they can interact with each other. So if you have a large data center on the left-hand side of the road that’s consuming 200 or 300 megawatts, and you have a data center on the right-hand side of the road that wants to pop up and consume an equal number of megawatts, you can now have interactions between those two loads. They can affect the ability of the system to serve each other, your neighbor can have an impact on your service — that electrical complication is very, very complex and requires some very detailed electric schemes to protect the system from those type of interactions.

So those are several of the issues that make data centers a really complex problem from an electrical standpoint. Your question though was what was the bottleneck? And there are several of them, but I would say the primary bottleneck is how do we study data centers and how do we allocate capacity to those data centers or what we would call loads? So how do you determine what needs to be studied for a data center to go in service? And then how much power can that data center consume at a specific point? And trying to answer those questions is the fundamental problem that is plaguing a lot of the United States on how do we handle data center development.

ERCOT’s tried to go about that in a couple ways and tackle that fundamental bottleneck. One was what’s called the LILIS process, or Large Load Interconnection Study or Services. That is designed to kind of standardize what studies have to be done for data centers in the state of Texas so that they can be reliably studied and integrated into the electric system. I would say results have been mixed at best on that process. There’s good and bad that has come from it. Lots of folks have complained that the benchposts are moving, that ERCOT is requiring continually different studies and more studies. And even after a study is finished, it potentially has to be redone. Those are some of the complaints with that process, but it has fundamentally been an effort to try and standardize and streamline.

But ERCOT is kind of at the bleeding edge of the sword on this issue. Again, back to the fact that it’s its own electric grid. It has more data centers seeking interconnection inside of ERCOT than anywhere else in the United States. So it’s having to grapple with these really complex problems of how do we bring this development in, and how do we do so in a way that’s safe and reliable for our electric grid and for all of the other people who rely on that electric grid on a day in, day out basis.

Other elements that have been put in place to try and deal with those bottlenecks on the study process — one was SB6 that was passed by Texas last year. This was a very ambitious endeavor and it’s too early to tell exactly how successful it is going to be. The Public Utility Commission of Texas is implementing the provisions of SB6 through five or six different rulemakings. The industry and stakeholders are heavily involved with that.

And what SB6 is really trying to do — it’s probably a 15-page bill, so I’m not going to do it full justice here in 30 seconds, but it’s attempting to set up industry-wide standards for how a large load interconnects in the state of Texas, step one. And then step two, it is trying to say, okay, we’re going to create financial requirements on these large loads that they have to post security. They have to operate in a specific way. They have to behave on the electric grid. Backup power, are they bringing it? What are they going to do with it? It’s setting out general parameters for what the state expects from these large loads, how we’re going to study these large loads, and how we’re going to reliably integrate them into the state’s electric grid. Like I said, there’s a lot of work still to be done on that. I think it’s promising early results. But hopefully what SB6 will do is it will lead to a much more cohesive process and much clearer signposting for large loads and how to integrate in Texas.

While these bottlenecks all exist, they are in the process of getting resolved. One thing I think that’s important also to point out in Texas is what is not a bottleneck, and that is the actual ability to get the infrastructure built. As I mentioned at the beginning, Texas has the ability to get infrastructure built faster than anywhere else in the nation. So while there are bottlenecks on the study side, one thing that Texas does enjoy — and this is one of the reasons I think you see so much of this industry flooding to Texas — is once it’s studied and once there’s a plan in place, there’s a very clearly demarcated line of how do you get a project from study phase to steel in the ground and energized, that all of these industries can rely on in Texas. That’s kind of a tried and true process.

Aubrey Bishai [00:13:09]

It certainly does sound like a complex puzzle. So what are some of the ways that ERCOT is adapting its market structure or planning processes to account for these large concentrated loads?

Jaren Taylor [00:13:27]

So there’s a lot going on there. What I would probably point you to first and foremost is the ERCOT batch process. A handful of months ago, as these complaints were coming in on loads not being able to be studied in time, not being able to be interconnected in a timely manner, and basically general angst over does the existing process that’s worked in the state for the last 20 years, does it work in the paradigm of these really, really large data center loads that kind of all want to come at once and they all want to come into the same geographic areas. How do we deal with those issues?

The solution that is being worked on right now, and this is absolutely a right now issue, is how do you do a batch study? And think about a batch of cookies, right? You put one in the oven, they cook for 8 minutes, you pull them out, you stick the next batch in. That’s kind of how this is designed to work. You take all the loads that want to interconnect within the state at a specific time. We study that on a statewide basis. And then we define what is the grid able to serve? What is the capacity that can be served? Then we’re going to allocate that capacity to the various data centers who have met the SB6 requirements to get into the batch. And then three, what transmission, what infrastructure do we have to build to be able to serve those data centers with the capacity that they need to operate their businesses?

So the batch study is designed to look at the state as a whole, look at the customers who want to integrate who have met the necessary requirements and then come up with a transmission plan of how to serve those and to do it all at once in batches that are kind of enumerated. So batch 0 is the process that the rules are being defined right now for. Batch 0 will kick off over the back half of 2026, hopefully arrive at a transmission plan and allocate capacity to various data centers towards the end of ’26 or the beginning of ’27. And then you will have the next batch start, which will be batch #1. Batch one will start out. You will have new loads that are studied, new capacity determinations that are made, and then a new transmission plan that comes out at the end of batch one. So what ERCOT is trying to do is move to this batch process.

I probably should have also addressed the problem they’re trying to solve with the batch process. The problem they’re trying to solve is the fact that not only are these data center loads massive, there’s also a lot of them, and they’re tending to geographically cluster. So they’re locating themselves in very close proximity to one another, primarily due to the fiber, primarily due to the fact that generally vacant land to build on is in the same general areas.

And so one of the problems we are having from a study perspective, as we mentioned earlier, these loads can interact with each other. It’s how do I get to a static point where I can go out and study loads A, B, and C when loads C, D, and E develop in the same general area. And so as that geographic clustering occurs of these loads, it becomes very hard to define a static number of variables that can be studied because these studies take the better part of two months to do. They’re extremely precise electrical studies on how this equipment and these facilities will interact with one another.

So how do you solve that? How do you solve the geographic conflicts that exist between these entities? How do you find a point where the status quo can be defined and no longer variable and moving, so that you can study? ERCOT’s solution right now to that problem is a batch study — is to define who’s in, study who’s in, get a result, and then start batch 2, study who’s in, and move forward. That’s, I would say, is the biggest change you would see in the ERCOT market right now. And those rules are being defined as we speak. It’s going to take a couple of months, probably the better half of this year and into early 2027, for us to really get a view as to how successful that process will be.

Another interesting thing about that process will be that it is absolutely going to create winners and losers in the data center industry. As certain loads will be accepted into that process and be allocated capacity through that process, there will also be loads that do not receive capacity through that process, and how ERCOT deals with those various chips that are going to land will be an interesting thing to watch.

Aubrey Bishai [00:18:18]

Jaren, as you have so effectively outlined, Texas is clearly a leading indicator in this space. So what would you say other regions should be learning from what’s happening here?

Jaren Taylor [00:18:32]

It’s a great question. I think that other regions should be watching and understanding the way that geography plays into this analysis. Where a load is located, as well as the size of the load, and the technical parameters of the load all impact your ability to get it in service.

One of the things I think we’ve learned in ERCOT over the last year or two years that this has been happening and we’re learning it more and more is these loads interact with each other and therefore they are very hard to study one by one. And so you’ve got to look at a certain geographic area. You’ve got to look at a certain number of facilities and then you have to stop and you have to study and you have to plan and you have to move that forward, or you keep restudying and restudying and restudying.

So maybe to summarize what’s one of the things that I think other states can learn from the ERCOT trials and errors over the last 24 months: it is that you have to define your variables. You have to go forth and study something, even if it’s not the perfect thing to study. And then you have to rely on those results, build infrastructure, and study the next tranche. If we spend so long in the pursuit of the perfect, we end up just restudying and restudying and restudying. We don’t get the infrastructure built and we don’t get the investment to actually occur because we’re stuck in a study-do loop.

The early data centers in Texas that have been interconnected were able to avoid that process and are on the grid taking service. They are taxed entities who are providing immense benefit to the areas in which they are located. And that’s a great success story. That’s the success story I think that Texas wants to continue.

Texas, I think, very much wants to have this development. Data centers can provide immense benefit from an ad valorem tax basis to counties, municipalities, and schools. But in order to get there, you have to be able to solve the electrical problems. And you have to be able to figure out how can we give them service and when can we give them service so that those companies can invest in the infrastructure necessary to allow those businesses to be a success.

Aubrey Bishai [00:20:49]

Thank you, Jaren, and thanks for joining us for Powering Progress. If you’d like to explore more insights on AI infrastructure and the legal frameworks behind it, visit velaw.com.