Tamra Reynolds: Both ERCOT and PJM face significant power supply challenges related to interconnection queues and supply chain constraints. PJM, the largest grid operator in the U.S., has more than 3,000 active projects in its interconnection queue, which is more than ERCOT, CAISO, New York ISO, and ISO-New England combined. The time between an interconnection request and an interconnection agreement is longer in PJM than in any other region. And, it used to be easy to build in Texas, but that situation is changing, as this part of the country is facing the same challenges as other regions.
Hello, I’m Tamra Reynolds, your co-host of Power Plays and a managing director here at CoBank. I’m joined for this continuing conversation with my co-host and CoBank’s energy economist, Teri Viswanath. Hey Teri.
Teri Viswanath: Hey, hello there. We are going to pick up where we left off last month with our guests, Spearmint Energy’s CEO Andrew Waranch and Spearmint’s senior vice president of development, Gina Wolf. I’ve had the pleasure of working with Andrew, and Tamra, when we discussed hosting a program to understand the challenges for developing power supplies in different regions of the country, I couldn’t imagine a better tour guide for that conversation. Tamra, you’re going to lead us off with a deep dive on ERCOT, followed by PJM.
Reynolds: We’ve covered MISO and SPP, I think, in some good detail. There’s always that one exciting outlier, I guess. ERCOT is next on our list. My team covers a lot of the south and southwest parts of the country, so ERCOT is something that we struggle to understand, I think, sometimes, but also appreciate opportunity that might exist.
With serving about 90% of Texas and having about 103 gigawatts of installed capacity, it’s always something that we’re paying attention to and trying to understand what’s coming next. At the end of January, ERCOT had almost 400 gigawatts of generation resources in its interconnection queue. It’s a state that you guys, I think, have a pretty active footprint in. Is that right?
Andrew Waranch: Yes. We’re actually one of the largest standalone storage developers in all of Texas. We currently have 150 megawatts, two-hour storage in operation, two 100-megawatt facilities under construction, and another gig plus in late-stage development which should hit NTP this year. By the summer of ‘27, we’ll be somewhere in the vicinity of two gigawatts installed, four-gigawatt hours.
The opportunity set in the market is fantastic. If you compare the demand side with the generator growth side, you still see that the problem is getting worse, faster than it’s getting better. Even if you discount a lot of the load growth, and personally, I’m not totally a believer that the green hydrogen and green ammonia demand is going to show up, even if you take that out, you’re looking at 40 gigs of demand growth by the end of the decade, or sooner.
I think the one thing holding back the demand growth, the same thing holding back the supply growth, is the time it takes for utilities to build substations to get the breakers and transformers. It makes overall very challenging to forecast both the supply and the demand side because you don’t always have the best visibility month to month, quarter to quarter on how those assets are going to be interconnected.
Overall, supply and demand balance is getting tighter, not looser. Fuel prices are up. We see natural gas for next year-- January of next year, natural gas is $5.40. That’s very expensive, and it only exacerbates the on-peak versus on-peak price differential, so when you’re charging overnight, when it’s cheaper and generating during the day, when it’s more expensive, it allows for even more revenue for batteries. Despite what looks like a very large queue of batteries, we think our projects are further advanced, better located, and more likely to come online on time to help solve that demand over the next few years.
Additionally, so many of the gas-fired generators in the queue don’t have good visibility to getting their turbines by the end of the decade, let alone their breakers and their transformers. I was in Europe last week visiting with a number of manufacturers on the equipment side. Some of the turbine manufacturers are sold out till 2030, and a lot of those turbines have been sold to big tech data centers.
Many of the big tech data centers are putting generation behind the meter just to serve their factory, so you have the grid competing with the data centers themselves for that gas-fired generation, which in some ways is why you continue to see projects fall out of the Texas Energy Fund. There’s not enough equipment, and with the rising gas price, their margins are less good than they thought they were going to be. All in all, it’s a very ripe place for storage development.
Reynolds: Yes, great points. Teri and I talk about that on a regular basis, the fact that the supply chain continues to cause challenges when you think about all the elements of what it takes to put something into service. The turbine issue is one that we’ve seen really rise up in the last 12 months or so, with projects that are getting announced but not really getting anywhere. You make great points about having storage in the right places at the right time. Probably your approach to the market was a good one when you got in there and it’s only going to get tighter, as you indicated, over the next couple of years according to what ERCOT is saying the market.
Viswanath: You’re bringing up a number of interesting points that I think we should unpack. The first is that Texas is very much aligned on the view that we can’t interconnect its way out of this problem. The state is going to have to get resources built, and they appear to be really focused on building gas-fired generation. But something you point out is that the fastest, the quickest we can deploy this technology, in really a meaningful way, will occur possibly in the early 2030s, and that’s a long way off.
As an industry, sometimes we get myopic about seeing only one type of solution for the problem. One silver bullet, if you will. It just doesn’t work that way. I’m wondering, does that mean from a battery developer standpoint, when you have policies that are probably flawed, how do you go about your business? We’ve seen this problem created on the coasts, east and west, but maybe we’re seeing this on the third coast in Texas. And possibly it represents an opportunity for bringing different types of power supply into the marketplace. But I’d like to hear your take on this.
Gina Wolf: It’s a great opportunity for batteries in the ERCOT market. As you’ve said, those timelines are so long for gas-fired generation. ERCOT in the past has been the very fastest place to get interconnected. In most cases, it still is, but there are pockets with certain transmission operators within ERCOT that are taking significantly longer times to do their studies than they had two, three years ago. They’re also running up against some of the same supply constraints.
They need to order transformers, they need to order high-voltage breakers, and they’re in the queue with all the other folks trying to buy all that same equipment. In some cases, they’re not willing to work with developers earlier in the process to be able to place orders for those earlier prior to getting the generator interconnection agreement executed. You have a stacking of timelines where they’re taking longer to do the studies. They finally get the studies completed, sign a GIA, and then they finally order equipment, which can take another two to three years to get those in some cases before you can actually build the substation and actually energize the project.
Certainly, the battery itself could be built extremely quickly. It can be built within that year, but all those other steps are starting to elongate out in ERCOT to a three, four, five-year timeframe, which is then starting to look a lot more like how long it takes MISO to do it, or an SPP.
Reynolds: The Texas Senate just passed a bill, Senate Bill 388, the House hasn’t done anything, of course, with it yet, but the principle behind it is to create a new dispatchable power credit trading program that would require utilities and generators, and electric co-ops even in ERCOT to offset new renewables and battery capacity with equal amounts of dispatchable capacity as early as 2026. The bill’s definition, at least from what I read, was that dispatchable excludes batteries and exempts power generators that exclusively operate battery storage systems from the dispatchable generation requirement. We’ve been talking about some of the challenges of building dispatchable resources. What kind of impacts do you guys see in terms of maybe not getting enough supply built?
Waranch: I’m sympathetic to what they’re trying to do in trying to incentivize more generation to hit the grid. In some ways, they believe that all the incentives they’ve given so far to storage are enough, and so now, they’re trying to give additional incentives to other generator types. The problem is, and again, we’ve seen this with all the plans dropping out of the TEF, I don’t think that in the next four or five years you will be able to achieve the type of thermal generation growth that the legislature would like to see. Batteries are not the long-term solution, but they are a part of the long-term solution. They most certainly are the short-term solution to help balance the grid.
Viswanath: We’re going to come full circle. I had planned to start our wholesale pricing journey really talking about PJM. But as it happens, we’re going to actually land and end on PJM. For our listeners, it really important to understand that PJM is the largest of the ISOs from a capacity standpoint. It’s really been the darling of the power development community for a long time. The operator controls 183 gigawatts of power generation capacity, serving 65 million people with about 88,000 miles of transmission. It’s spread over a somewhat compact 368,000 square miles.
As we think about this, and I’m going to be honest about the current queue, it’s hard to measure because there are a lot of moving parts in PJM, in terms of policy. Lawrence Berkeley Labs might call it 286 gigawatts in the queue, which is more than double the current installed base. But there’s a new process in place to address those interconnection requests, so from a bean-counting perspective, who knows. But let’s get a big picture on what development looks like for PJM
Wolf: Today in the PJM queue, there’s a lot of challenges even with the reforms that have tried to occur. I was looking right before the call here, the cluster group that you can actually join right now, which is cycle one, so the first full cluster study after all the transition has occurred, the folks that have been sitting in there waiting for studies in cycle one joined the queue back in 2021. They’ve been waiting for four years for their study process to start, and that cluster hasn’t closed yet. We can enter the queue today and have the exact same position of study in the queue as folks that have been sitting there since ‘21.
If we enter today, someone in probably at least a year, probably two years that enters in 2027, will also be at the same position as those folks that started in 2021. It’s a pretty challenged environment for entering new projects into the queue right now because you could join today or you could join in a year and a half and it doesn’t really matter. We still aren’t going to even start the study process of the cluster group. As they work through the massive backlog of all the transitions, clusters, to finally get to the new cycle, it is taking, of course, longer and longer, and the timelines just keep getting pushed. It presents a huge challenge for planning and for new projects to really decide to start there with that amount of uncertainty.
Waranch: Other things that I would add in PJM, maybe in more of a macrolevel is, one, our chief development officer, Peter Rood, actually developed two of the first batteries in PJM over a decade ago, and so any developer who was part of that process, which was well-intentioned but flawed, saw most of those batteries economically fail because their revenue sources dried up. One, there is a history of some programs for storage that didn’t work.
Two, when I started developing batteries 2020, 2019, the thought process was that by now we would be far along into having the 100 gigs of onshore solar and offshore wind in PJM. I mean, I don’t think hardly any of that materialized, but that increased solar and offshore wind would’ve necessitated a lot of storage, and so you had, one, an old battery program that didn’t work. Two, the delay in adding solar and wind and the delay in retirement, all of these led to less need in the short term in storage.
Add to that, you had a multi-year decrease in capacity prices. I have been in PJM since before there was a capacity market, which makes me sound ancient, but one thing we’ve always known is that prices for capacity will come up and down, and when they get too high, people will build a lot, and when they get too low, people will stop building. What we saw in the last auction or two auctions ago when capacity was well under $100 a megawatt day, people stopped developing. What happens when you have load growth and less generation? The capacity market went up above $300. The first thing you heard was some of the politicians screaming about high-capacity prices.
That was an absolute direct response to, one, not building generation for several years, and so it’s the natural response. The politicians can’t have it both ways where they enjoy the low prices of the market, and the minute prices go higher, they complain they’re too high and you set an artificial cap on capacity.
Viswanath: Yes. What you’re talking about, Andrew, is that we just saw this agreement, Pennsylvania’s governor, Josh Shapiro, had asked in a negotiation, to get the capacity auction price cap down, get it lowered. It was over $500 per megawatt a day, and it’s down to 325 megawatts a day, and it’s recorded in their last capacity auction back in last July. That was a record-high capacity price and a lot of resource advocates pushing hard, pushing back. We’ve had this ongoing debate since the ‘90s, which is can the market provide and finance adequate investment by these capacity auctions? It sounds like we haven’t achieved the balance yet by keeping it artificially low.
Waranch: I would go back 10, 15, 20 years where you saw other northeast states have similar policies. We can look back to Connecticut in ‘05, which is 20 years ago, but it led to a significant decade or two decades of higher prices by tinkering with markets. The markets are designed to incentivize generation with high capacity prices. If you remove those and you artificially lower them, you disincentivize generation at the exact time where you need.
Especially with tariffs and inflation, and that cost of new entry just going up and up and up every year, I forecast higher capacity prices in the future, not lower, and setting artificial caps just disrupts the market and long-term, hurts citizens. I think you can go back even farther to the ‘90s when the California market was being developed, much of the California power crisis was really a result of market tinkering that was well-intentioned but went awry and led to much higher prices.
Reynolds: Andrew, something that we’re seeing in PJM and other places too, but primarily there, is the development that is circled around co-location or bespoke type development around large loads. Earlier this year, FERC launched a review of issues related to that, like data centers locating or having power plants located close to loads like that within PJM’s footprint. I guess the outcome of this review that they’re doing could have impacts or set a precedent for co-located load in markets that they oversee. What does that mean for you all with regards to possibly developing resources in this region, and what are you thinking about when it comes to that sort of thing?
Viswanath: You guys talked a little bit about where a utility has an RFP out, right? They’re shopping for assets. This is very similar, except it’s a large consumer that is attempting to do this. From a developer perspective, how do you feel about this?
Wolf: We’re very excited about new load getting added, and data center load growth is very beneficial with the kind of menu of options that FERC’s analyzing on how co-locating the large loads with the data centers, or how can the data centers do some demand response, shut down an emergency operation, some of those things to help shave at those peak times could be very helpful for the reliability of the grid, and also still allow those to come online, still allow projects like ours to be able to help in those other periods and bring that additional load. Everyone recognizes we need to have these data centers, AI and everything is not going away, the demand is just going to keep increasing, and this could be a way to effectively allow that load to get online, to allow those data centers to start operating without so many impacts to the grid, but still bring the growth.
Waranch: Right now there is such a robust demand, that data centers are willing to pay $150 a megawatt hour for short-term supply. The biggest problem I have with co-location is that for the first four to five years, you’re going to be burning really inefficient units. Really, and this is not a story about carbon emissions, this is about local emissions. SOx, NOx, particulate matter, this is things that affect real local health issues, and they’re massively inefficient. That’s the biggest problem I have with a lot of this co-located generation is it’s really bad gen.
Viswanath: Yes, and ultimately likely to be replaced when we have that ability to build what we should be building now. Right?
Waranch: That’s the goal. The goal is to put these re-sips in for four to five years while you wait for your turbine, but those four to five years, it’s a mess.
Viswanath: There’s also just the cost involved because the numbers you’re quoting are about double the low end of the power capacity development price, right? Certainly when we think about utility-scale solar and battery development projects, we’re talking about multiples when we think about temporary solutions. And it’s just not an efficient use of resources.
Hey, Gina, Andrew, I have really enjoyed our discussion. It’s been incredibly helpful to gain your insights, especially from a developer perspective, and it’s really great for our audience to be able to take this tour of the regional wholesale ISO markets. I want to thank both of you for being very generous with your insights and for being on our program today. Thanks so much.
Waranch: Thank you for having us.
Wolf: Thank you.
Viswanath: There are some relatively important takeaways from our conversation with Andrew and Gina. There are a lot of obstacles that are getting in the way of building efficient, low-cost generation. Challenges related to the interconnection queues, supply chain constraints, growing community resistance to have the iron sited in their backyard, you name it. But in a world where power delivery costs are also rising at twice the rate of inflation, there is this growing desire for all of us to do something about it. But the problem is that “market tinkering” will likely make this problem so much worse.
Reynolds: True, but I view the increased awareness of these challenges by consumers, regulators, co-ops and other utilities, along with the private sector, as a positive development. In fact, this awareness and willingness to collaborate might bring new solutions to the table, the likes of which we’ve never seen before.
I recently saw an announcement by PJM for a multi-year collaboration with Google and Tapestry to deploy AI-enhanced tools to streamline PJM’s planning process for connecting new generation. Imagine if some of these large new customers could actually help solve the grid’s known challenges. With better insights and real-time grid awareness, could we operate with the thin reserve margins that we’ve discussed today? Maybe. But there might be more at stake for this collaboration. Data centers account for about 4% of total electricity demand. What if they made the rest of us – the 96% of us – more efficient?
Viswanath: Right. I worked for a merchant generator in the early 2000s and remember kicking off the modeling software we used to forecast ages ago. I would run the software, I would go out, grab lunch, grab coffee, read my mail, and hoped that the model run would be complete by the time I returned to my desk and that I hadn’t made any mistakes. The fact that AI might be applied now — cutting processing times for reviewing those new interconnection applications, allowing large volumes of requests to be processed quickly and accurately — is really hopeful.
Reynolds: But even if we speed the planning process there are still hurdles to getting the steel in the ground -- supply chain delays, permitting blockades and the growing not-in-my-background community pushback -- leaving me to think that we will need to continue to cover power supply on future podcast episodes.
Viswanath: You’re right. But for now, I do hope all of you shave enjoyed this conversation and will join us next month as we look at a pilot co-op program that looks to bolster the grid with “virtual power plants.” Goodbye for now.
