Third Bridge专家称FERC/PJM大负荷互联与RBA框架总体利好Talen及IPP，但缺少新增硬催化

[[00:00:45]]
FERC [Federal Energy Regulatory Commission] has recently issued show-cause orders to major RTOs [regional transmission organisations] to address what it called potential inadequacies with existing large-load interconnection rules, including colocation and behind-the-meter arrangements, clarity of cost allocation between stakeholders, frameworks for flexible large loads and potential curtailment – voluntary or otherwise – and transmission debottlenecking. Depending on what PJM comes back with, what do you think this could mean for Talen Energy?

[[00:01:29]]
I think generally, it's very supportive for Talen and other IPPs. There's a big issue right now where I think most people would understand this, but all this like, let's build the box correctly is you have generation submitting for generation interconnection agreements. That's a two-year process along the traditional path. The expedited path is maybe a little bit shorter, maybe it's a year. You get a generation interconnect, agreement through PJM and then the utilities then have to go take… It takes a few years to build out the substations and whatnot. You're talking about a process that takes between four and, say, six years. That's just the physical piece and then there's the cost associated with that connection of that generation. On a parallel path, there is the load submitting a load interconnect and that is also going through a similar process at a similar timeline and is run with all the cost potential upgrades to their system. The two components are not currently studied together, and where in the Talen Energy Susquehanna deal, you had a campus of data centres that was… Let's ignore how it was electrically connected, but it's at the same physical location as the generation.

It should reduce a lot of those network upgrades and those knock-on effects of just one vs the other because they're netting out to be basically zero impact on the grid for a vast majority of the time. I know there's technical components to that of if the load goes away or if the gen goes away, what does that do to the grid? What this show cause order does, it allows or they're pushing for an ability to study both of those together. Now those companies that have interconnection agreements already, who have land, who have all that physical infrastructure that's there… If you wanted to, on the one hand, build new generation, that should be very supportive. If you wanted to build load for your generation, that's incredibly supportive. I think we're going to get into it in some other questions, but let's just preview it really quickly is that right now, under the current… As much as people say, "Oh, prices are really high," they're not high enough to actually cover new build costs.

A capacity auction that cleared, when is it, last week at USD 330 per megawatt-hour, that's not enough. The cost is over USD 600, USD 500-600 to build a new combined cycle per megawatt-day. Right now, without a more comprehensive way to do gen and load together, it's a struggle to build new generation for these IPPs, getting back to the IPPs. They want to expand their footprint. They want to add new capacity to the grid. They're not going to do it without a PPA. They need the load for that PPA. We can dig into any of those pieces, but I think it's really beneficial for Talen, it's really beneficial for IPPs, but it's also beneficial for load, for economic growth. There's a lot of knock-on effects to why this is really smart and it's helpful for all parties involved.

[[00:05:31]]
The proposed RBA [reliability backstop auction] is now expected for September 2026. That's more of a structure that might feed into what the IPPs [independent power producers] or generators want – longer-term contracting and bilateral negotiations. I think that's what this is shaping up to be. Depending on what you think the RBA will look like, what does that mean for Talen, given the various factors you mentioned?

[[00:06:01]]
I think it builds on what we talked about last time, two months ago was that nobody is going to build new generation unless they have a PPA. Nobody is going to go out and spec for a few reasons, let's just be clear. I'm very sympathetic to Talen and to other IPPs when people go after them is that they're not going to cannibalise their own market, (1). (2) The history would show that if you build power plants on spec, it's really hard to perfectly hedge those over time. Usually, the equity gets cleared out and the debt usually ends up owning these assets. This reliability backstop auction or framework, they're trying to help facilitate buyers and sellers to match new generation with existing load who has existing load interconnects. I think, again, very beneficial for Talen. It's the only way new generation gets built. I also think there's tremendous knock-on effects to this in that one of the… Something we didn't talk about last time, and I think it's because it really just has popped up recently is that there is this massive social licensing problem throughout the United States associated with data centres and that they use too much water and they're driving up energy prices and yada, yada, yada.

As a result, there's this massive backlash for data centres. Why that really affects these IPPs is that the data centres are the ones consuming the incremental amount of power or any new additional power. This reliability backstop auction, my belief is that it's helping to alleviate some of the social licence concern because what it's doing is it's allowing a mechanism for a load to buy generation and point to that generation by saying, "Yes, I'm building a data centre, but I'm not going to impact the power prices in the region because I'm bringing my own generation." I think that having that mechanism is really, really important to be able to build new data centres, which will then, as a result, be able to build new generation. Obviously, with any new load coming on board, I know it helps your existing generation that Talen has a really large fleet across PJM. Maybe we can double-click into any of that, but any questions you might have on those different pieces.

[[00:08:59]]
Regarding the RBA terms or parameters, I've read that it was up to 15-year contract terms, and it was proposing to cap the clearing prices at USD 550 per megawatt-day in terms of the expected top-end procurement cost. If that's the case, it still falls under what you said it costs for an IPP to bring forward new generation – USD 600 per megawatt-day. Are generators even incentivised to participate in the RBA as it stands?

[[00:09:49]]
I was talking to somebody about this yesterday. I think it's close, and I think that is just for capacity. You still have a whole PPA that you have to underwrite for energy on the back of that. I think that maybe how I would do it is that if you're at the cap at USD 550, but you still need incremental margin to make this really work, within your PPA, I think there's maybe room for some fixed and variable… Most of these PPAs, off of gas plants will have a fixed component and a variable component because nobody wants to underwrite fixed price gas for 15 years. Anything associated with the incremental dollar of energy will be a variable component. then you'll have a bunch of fixed pieces of fixed O&M and whatnot. I suspect that within that fixed component of the PPA, you're probably going to have to add some additional layering really just for CAPEX because, yes, it's just like the CAPEX is just… It's expensive to build these assets and then to pay for the potential line upgrades and other substation upgrades exclusive of the generation itself.

I think it's a step in the right direction, but the economics, it's still tight. There's a big hearing this morning on FERC and PJM and my general take is that you should let free markets work. If you allow free markets to work, a lot of this solves itself. I think having the data centres and the IPPs talk together, what this is somewhat trying to facilitate, I think help solve this better than pure just energy on one side and load on the other.

[[00:12:00]]
Let's say the capacity element is USD 550 per megawatt-day. Assuming that the RBA is primarily for new generation and existing generators with different economics, if you had to take the risk on energy, do you think an IPP such as Talen would be willing to take that risk and put forward on the RBA at those prices, or is it still too risky because of the energy exposure?

[[00:12:32]]
No. Just maybe to break it down again is like you're going to get your capacity, which it can be a fixed price 15-year payment. Then once you get into the PPA, there's then a fixed component to the PPA that you could do for 15 years as well, which would cover any of that additional CAPEX coverage. What you're left with is just that variable component of gas. At FID, they're not going to sign up for something that actually isn't economic. I think that if I was one of the IPPs, I'm definitely leaning into this, trying to get matched up with one of the hyperscalers or a large, large creditworthy counterpart and then figuring out what I can work through on the PPA side. I definitely think that you have to be leaning in… This is the best possible cap to get new generation built even if it is USD 550 per megawatt-day, the cap. I think there's possibly other ways to make it so that it's still worthwhile.

[[00:13:45]]
How should we think about the eligibility criteria for the RBA? How involved can Talen get? Is the RBA ultimately a net-net tailwind, neutral or a headwind for the company?

[[00:14:08]]
That's a good question. This is my perspective and my opinion, but if you look across the Talen fleet, you have some fantastic relatively new assets that they bought through M&A. They have some great assets that have a very long life. Then they have some other assets like Montour, Brunner Island and some other assets and stuff, Baltimore, that's closer to the end of life. Every IPP stack a little bit differently on the age of their units. I think what Talen has is a little bit of a mix of both. With some of those older units that have land opportunities around them, I think this really gives them an ability to create this flywheel, where they repurpose those interconnection places with interconnection points with new generation and really extend the life of those assets, even though it's different assets at the same point of ground. Overall, I think I'm pretty bullish that this is helpful to Talen.

If I would be the devil's advocate of that is… The point of this call is just to share all thoughts and information. There's always a risk that you have load and data centres who are going to go sign up for a lot of generation. Let's just say that it all gets matched up and all this generation comes on and these data centres, they run at a 60% or 70% capacity factor. You have, from a capacity perspective, a lot more generation coming online. From a load perspective of the peak day of the year, a lot more load coming online. The other 95% of the year, you end up having a lot of excess capacity sitting there, not idle, but sitting there underutilised. If you're an IPP, you're going to optimise that to the fullest extent. You'll run your new assets that you build, but so will everyone else, and it might drown the existing assets a bit in pricing in that 95% of the year, your pricing comes down, in your peak period, your pricing goes up. You have to model out what that does for revenue and margin across the grid.

Just to cap it off, it's like there's a probability where you have a lot more capacity on the grid underutilised. As a result, it creates lower prices. I think a good example of that is in Texas, where you have this very bifurcated market where you have, a vast majority of the year, extremely cheap pricing, extremely, extremely cheap pricing. Then on your PPP periods, very, very expensive pricing. Tough if you're an IPP, you got to make sure you're there during that peak period as all your revenue and margins are coming in in 5% of the year vs having a more evenly distributed pricing scenario.

[[00:17:43]]
My understanding is that eligibility for the RBA is largely for new resources that haven't cleared in the traditional capacity auction and for uprates, which similarly have not been included in that. Talen now has a net-net newer fleet, where I'd assume uprates might not be as necessary, or at least not as available, given it's a more efficient, younger fleet. How do you think the company gets involved in the RBA? Would it shift from flywheel to investing in new generation just to become part of this?

[[00:18:34]]
I think that's a good question. It's a newer fleet. A bunch of it is a newer fleet, but there's also a bunch of assets that aren't newer. As I was saying, Montour and Brunner Island, and you can research those the other units they have, they have units that are an older fleet, the old PPL generation units. I think also if I try to laterally think outside the box, if they were to really lean in on this idea of, "Let's ignore what their fleet is," and they just wanted to be in the development game, knowing that they already have all the expertise for operating plants and they have the balance sheet for trading and the balance sheet to underwrite new generation, they should be able to ignore what the existing fleet looks like and just say, "We're going to look at the map, and we're going to go find places on the grid that are underserved, and we're going to go buy land and we're going to go build new generation."

They are as positioned as well as a lot of the other IPPs, maybe even better, because there's market power issues probably with some of them like Constellation, but there's the smaller ones, probably don't have the big balance sheet that Talen has and some of the expertise of operations. I think Talen sits in this… There's a good sweet spot where they can use some of their old generation to repurpose that and build new generation on top of it. They could also just go find some greenfield sites and build new generation. The upgrades, I think you're probably right on that, it's like probably not tremendous opportunities for uprates. Honestly, I think outside of Constellation, where they theoretically could upgrade much of their nukes, there's not tons of opportunities for upgrades across the whole PJM fleet. You're really talking about new generation for the most part.

[[00:20:37]]
How do you think the RBA would impact the traditional RPM [reliability pricing model] or capacity auction? Could participation or uptake of the RBA suppress future capacity prices? All in all, could this be a net negative for Talen with whatever happens with capacity prices because of the RBA?

[[00:21:13]]
Let's walk through the mechanisms of how that could work and then we can give our… Again, it's just opinions. If you clear the reliability backstop auction or if you're part of the bilateral process, you then will be placed into the BRA at a zero-priced offer, and you'll obviously clear. The load should also come in. For any contracted generation, it should go in as a net neutral load and gen into the up-and-coming base residual auctions. That initial PPA and that initial gen, which could theoretically be a bunch of simple cycle units because they're the fastest things to come online, there is the possibility that you end up turning all those simple cycle units into combined cycles over time because the added incremental megawatts are incrementally a lot cheaper for the steam cycle than it is for the gas turbine cycle.

Out of the gate, it probably is net-net neutral to the BRA, but over time, as you upgrade these units, you could say, "I have 600MW of simple cycle units. I can add 300MW of steam turbine if I just turn all these into combined cycles, and then I'll take that 300 additional megawatts that maybe isn't in my initial PPA, but maybe I contracted into a new PPA with said counterparty and then now put that into the BRA without the load on the backside to match it up one for one," you could end up in a place where you just have excess generation coming into the BRA now. It doesn't necessarily have load one for one on it. That's the one side. I suspect and I hope that the pressure from the White House down to the governors into FERC and into PJM is that data centre load has to bring their own generation.

Any new incremental generation that an IPP is looking to put on the grid, I really think that the data centre is going to match up with that one for one. Net-net, on the BRA itself, it should be net neutral, but what I would get more nervous about is just in your day-ahead and real-time markets, you got all this extra capacity now that's underutilised and it's running potentially more. From the BRA perspective, it seems like it should be net neutral. I'm trying to think of a reason why it would be really, really bearish for the BRA, if the load… Worst-case scenario is if the load isn't able to connect because of social licensing problems, but the gen is contracted and built and the load just doesn't come on as fast enough, yes, you theoretically could have overcapacity vs load, and it could theoretically be a problem.

[[00:25:10]]
Isn't that one of the main fears or issues that investor-owned utilities have raised with the RBA? It seems as if a lot of these utilities pushed back quite a bit on the proposed form of the RBA. Is it specifically for that? Are they, as middlemen, really concerned about offtaker risk or execution risk and then being caught holding the bag? Is that risk properly allocated?

[[00:25:44]]
I think what they're concerned about is not the… I guess it is the BRA, but it's more than that. They're going to do all these upgrades on the transmission system. They're going to be upgrading substations and reconductoring lines and they're going to be spending hundreds of millions of dollars, billions of dollars upgrading the utility lines. If the generation or the load doesn't show up, then who's paying for all… Then that's where the costs get potentially distributed down to the customers, and that's a major problem. I think it's more that all the costs for the transmission upgrades, if they are not appropriately protected, that's the issue is that all those upgrade costs could get associated back on to the customer. That's why you have these huge CIAC payments that are being asked for for all load to basically say, "Because I'm concerned that you may not show up, you got to pay for everything upfront," which is an issue. It's really big dollars, this is billions of dollars.

[[00:26:56]]
In our previous Third Bridge Interview [see Talen Energy – Continued Capacity Expansion & Key Demand Risks – 2 February 2026], you mentioned how the best-case scenario for Talen would be replicating the original Cumulus data centre deal, unfettered due to regulatory shifts allowing that – so behind-the-meter colocation arrangements with long-term durations on the contract and ostensibly some premium fixed-price arrangement. Obviously, the world has changed since then, and there has been some movement. Do you have any updated views on how probable that is? It was perhaps unlikely, given that the company had already shifted the deal with AWS, but where do you think we stand with that currently?

[[00:27:56]]
Make sure I'm understanding the question right, but I think the upside that I was saying was they should replicate what they did with Susquehanna at other units. We have these colocated generation and load scenarios on the same plot of land. I'm not sure if it was before or after we last spoke, but probably prior to the last time we spoke, the Susquehanna deal now seems to be 2GW, from what I can read in the press. I think that's more or less a done deal. The Susquehanna stuff is all probably done to date. I think for them, for Talen, it's just replicating this across other parts of other assets that they have. I think that everything that we talked about so far has really leaned into that idea of colocating load and gen. If it's in front-of-the-meter, behind-the-meter, I think that's, not semantics, I don't want to be that cavalier, but a lot of that just gets passed onto the customer, whether if you're in front-of-the-meter or behind-the-meter.

One of the really big benefits of the original Talen Energy deal with Amazon was the idea that if it's going to be behind-the-meter and you're not going to have to pay for all the T&D that you historically have to pay. I think they're now publicly talking about how they're front-of-the-meter. I think that just going forward, if you're front-of-the-meter, behind-the-meter, again, it doesn't really matter because customers are going to end up probably paying for that. Talen Energy is really more concerned about is how do you take an existing unit or new unit and then colocate it directly with a load? Again, going back to what we started with a little bit is that they have a bunch of existing assets. Let's go maybe down some of these scenarios. They have some existing assets. Those existing assets have five years left or whatever. I don't know what the number is. They don't have a tremendous long life ahead of them.

To build new generation, which is what we started with was two years of paperwork and then another 3-5 years of substation work and whatnot, you have this big disconnect between how quick it takes to build data centre, which is potentially two years and how long does it take to build new assets. A new generation asset could take 5-7 years to build. Data centres are two years. What happens in the middle here? You can't bring the data centre on unless the generation comes on. I think where Talen has an interesting opportunity is if they could figure out a way where you're able to potentially bring generation on to an existing asset that's potentially shutting down. Then by bringing that new generation in behind it, you're creating a regulatory pathway where the load can come on a little bit faster, knowing that there's new generation that can be coming out of the back to support it and that existing asset that might be shutting down relatively soon, it buys a little bit of life and higher cash flow off of that unit.

[[00:31:24]]
Talen is quite clear on the messaging about how it wants to take the recently acquired assets and contract them extensively with large loads, such as data centres. At a much simpler level, why doesn't the company just do what it did in the first place at Susquehanna? Why doesn't it just build the data centre itself and then sell it with a PPA [power purchase agreement] to a hyperscaler? Why doesn't it just repeat that playbook? Is it due to regulatory factors, or is it not worth the money? How should we think about that?

[[00:32:04]]
I struggled with that, too. If I was there, that's what I'd be doing. Yes, to be very, very, very blunt. I think you can ask the same thing for NRG. You can ask the same thing for Constellation. You can ask the same thing for any of these IPPs… Yes, it was up to me. That's exactly what I would do. Maybe to defend them, and we can double-click as to why they should do it. To defend them on why it's like they need to be thoughtful about it very publicly in the press is the fact that they try to do a data centre next to Montour, and they didn't get the zoning and permitting done. That's currently, I don't think (inaudible 32.52) moving forward. At the current moment in time, I'm sure there's work they can do on a community level. There's always risk associated with that. CEOs don't want to be spending much of CAPEX to potentially build something if the zoning and the permitting doesn't come up. If you're a developer, you would try to get all the entitlements done first before you put a shovel on the ground. I think your thesis is generally correct. You should don't wait for the customer to tell you… Don't give the customer a menu of options of what do they want.

I think a much better solution is tell the customer what you can offer them and go build a campus directly next to where your existing assets are. The more that it becomes a viable path where you're showing that you have zoning and permitting done and you have a load interconnect done or a gen interconnect done and you've even started maybe the civil works and the civil grading of the site, the more work that you have to show that this is a viable, viable path forward. The hyperscalers are going to be significantly more interested in that site than others. One of the problems that the data centre industry has is that they're inundated with opportunities. They're inundated with a lot of slop and a lot of developers that don't really… They don't know what they're doing, just they don't have the right funding or it's imperfect information, like they don't have all the zoning done. The land is questionable, the generation interconnects aren't done, a lot of unknowns. If I was an IPP, that's exactly what I would do. I think that's what was the math and why it was so successful at doing this was because it was all done, like, handed on a silver platter.

For any of the IPPs who are brave enough to do that, the hyperscalers are going to come directly to you. I think you've seen that a bit in some of these… I think NextEra is the best one, if you listen to their last earnings call. They called that out, basically, they're building, I think they quoted 30 campuses, where they're basically taking a bunch of their old renewable portfolio, and they're going to try to convert those assets that have financial zoning and permitting and grid interconnects or load interconnects and then they're going to convert them into data centre like large energy campuses. That's absolutely the right path to go down. I suspect Talen should be doing something like that, maybe they're just keeping it quiet. I'm not sure. I haven't really talked to anybody over there recently to dive in, and I wouldn't be able to share if it wasn't public anyway.

[[00:35:55]]
Is there a potential risk that Talen is not competitive as a power provider for data centres, given that everyone seems to be coming forward with a solution? A lot of sponsor-backed entities have gone beyond just a hybrid generation platform. Some of them have even gone to the point where they're essentially doing what we discussed – they're going to develop both the power element and the data centre itself, basically an end-to-end platform – and then go to the hyperscalers and say, "Just park your chips, units and racks in this shell. We built everything for you." Could Talen lose this battle because it is not replicating the Cumulus deal?

[[00:36:46]]
Yes. I guess why I'm not there, as we have pushed. It's tough to argue against what Talen has done because it's had a pretty good run in their equity performance. I think that, yes, the world could pass them by. I try not to talk too negatively of my old employer. They've done a lot of really, really good things and a lot of really smart things. The momentum is that end-to-end solution, absolutely. You have to believe that they're trying to work on something like that because that's what the customer wants. On the one hand, you could definitely say that maybe they're getting passed by. On the other hand, I think if you look out there in the space of like, all right, who are the best IPPs, I think they still rank pretty high in that list. If you're a hyperscaler and I still think the Talen is pretty high in your list of people you want to call and try to figure out how to do something with them through these different mechanisms of these different auctions or just one-off deals.

All I'm saying is I don't necessarily want to go out and bet that they're going to get passed up. I think that they have some incredibly smart people there who are seeing all the news that you're seeing and I don't think they're going to be sitting idly on the sidelines and letting people just run past them. I would hope that they're just doing it maybe in a quieter way. I would hope, but yes, I can't guarantee anything.

[[00:38:42]]
Talen clearly has an existing relationship with AWS. Why haven't we seen further long-term contracts for other Talen assets similar to the Susquehanna PPA? Is it because they're not nuclear? Is it to do with uncertainty around colocation and behind-the-meter guidelines from PJM and FERC more broadly regarding these types of structures? Is it social licensing? Why aren't there more contracts coming out of this relationship?

[[00:39:15]]
I think all of those things you said are totally correct. I think that they tried with Montour. They clearly tried and they failed there with the zoning. AWS is their beast who builds a lot of stuff, but they don't want to go originate a deal. They want a deal brought to them, just because of how much work they're working on. They want everything we just talked about in the last section. They want a campus, everything put together and brought to them. I think that Montour deal was that. The Montour deal did not go through because of their zoning problems, permitting problems is one of the two. I think that is a bit frustrating that they haven't been able to do that. I think the other piece or everything you said about all the unknowns on FERC and PJM, but we've had a lot of unknowns for over a year plus now. Sometimes you just have to move through the night in the unknown.

What I would say is probably the bigger reason is that as much as Talen may want to use their existing assets, additionality is still the key, the name of the game, because of social licence. As you're taking existing electrons and using them for incremental load that equals taking megawatts off the grid that would theoretically be allocated for existing customers, which equals higher prices. I believe in free markets, I think that Talen and those like Talen should be able to market those megawatts however they see fit. The data centres and whatnot, they also have this existential threat that if you don't have a social licence to operate, you're not going to build anything. This idea of we're going to go monetise the existing assets is somewhat flawed in that if it doesn't have additionality component to it, then the data centres have a really hard time contracting because it fails the social licence test.

[[00:41:46]]
In addition to the flywheel platform, Talen mentioned that it wants to pursue a hybrid model with several of its existing sites where there are existing zoning, existing generation, presumably some slack with the interconnection where you could upsize, as well as enough land for not only a lot of additional data centre capacity, but also the ability to pursue new-build options to supplement the power or supply side as those arrangements grow over the long term. What are your thoughts on what the hybrid arrangements or PPAs could look like, especially against what the company did with Susquehanna as the benchmark for large-load contracting?

[[00:42:44]]
I don't want to repeat what I said in the past. If you don't mind, can you just ask the question, just clarify the question to me one more time?

[[00:42:52]]
Basically, Talen said that for a lot of its existing sites – I'm sure these include the Cornerstone power plants – it wants to take these sites where there's a lot of land around and obviously a large amount of existing generation, but also the ability to grow that with new builds, assuming that they can support data centres, to a hybrid model. How should we think about what the contracts might look like here? Are we assuming contract durations similar to what we saw with the Susquehanna deal? Would it be something similar to the RBA terms? How do we think about the setting of pricing and whether that will be indexed, etc?

[[00:43:37]]
I think that you synthetically want to put together a deal that looks like the RBA, where I think you want to do a capacity payment, fixed price component and then a variable component of the gas. The last piece, that gas component was really hard a few years ago to get the hyperscalers and the data centres to get their heads around, like they're used to getting fixed price bundled products from the utilities. You have to go sign a retail deal with an IPP that has a variable gas component, I think took a lot of time to get people comfortable with it. The hyperscalers have hired a lot of smart people dealing with all the colos. They've hired a bunch of people from the power industry who understand that type of risk. On a new site, that you described for Talen, I think that doing something outside the reliability backstop auction potentially have that option of, "Okay, I'm going to get the equivalent of a capacity payment, but above the USD 555, maybe it's a USD 600-700 per megawatt-day dollar amount that then really helps drive new generation to get built.

I think that in a structure… Again, if I was sitting in the middle of a deal, the problem with the Reliability Backstop Auction is that you think about it in like very big chunks of capacity just coming online to meet the needs of existing load with load interconnects. What you're describing what Talen is thinking about from what… Again, what you've described, not inside of knowledge or anything, is that you have an ability (audio cuts out 45.42) derivatives or smaller simple cycle units and then you build a stack over time from… We'll start with something small, we'll add in some medium-sized things and then we'll add a bigger unit on the back end. By originating this much more thoughtful campus approach instead of just a block of boom, I'm just going to hit you with a bunch of capacity. It's a much more thoughtful approach of maybe you go behind the meter just to get the units up and running and you get the data centres up and running. This is what I would do. It's like you do something behind the meter, near adjacent facilities, but then those adjacent facilities that have that interconnect are already there, you then work down the interconnection pathway and you take that islanded solution and that islanded asset and you connect it into the grid over time.

Speed to power and speed to the market is really important. If you start with smaller units and you work with an Amazon or somebody like that who's doing this already. I guess you're not inventing anything new. The islanded campus and behind-the-meter solutions are all over the place. I think that's what they should be doing. They're already in the community. They know the communities, and then you build up something bigger than just this kind of blunt tool that PJM is trying to push.

[[00:47:18]]
The hybrid model that Talen mentioned looks very similar to what I've seen across major IPPs and other players looking at data centre campuses and providing on-site generation, colocation and all in one site. Regarding contract pricing, do you think there's the ability to simply get all-in pricing – no need to have a capacity, energy or variable component? Do you think the company could get USD 80 per megawatt-hour fixed for 15-20 years, for example? Is that a possibility?

[[00:48:03]]
Yes, the hyperscalers will sign up for it, but why would you, then you're going to assume all that risk of the gas risk. You're going to get short, if you do that, you can get short power, you're selling power and then you're going and getting short gas too. I don't know, that's not an ideal scenario. I wouldn't do it. You could do it. You could definitely do it, but you got to go buy, you have to go buy long-term gas to protect yourself against that price volatility of the gas. Then if you eliminate your price volatility, but now you have delivery volatility of the physical molecules that you have to then sell out or burn if the data centres aren't using it. I don't know, it gets a lot more complicated. 15 years is a long time to be managing a gas contract, definitely doable. I guess you could find the right producers to help you underwrite that type of deal, but there's definitely a lot of risk associated with it. There's a lot of risk, and the risk is a lot higher than what the cost of just building a data centre is. If you're not going to build that, then I don't know, it feels like a stretch to go do fixed price power.

[[00:49:43]]
Generally, for the hybrid PPA situation where you're essentially building a generation campus with land for data centres and new-build optionality, how do we think about the risk-return spectrum vs Talen's bread and butter or just straight up contracting an existing asset, such as what it has bought recently from ECP [Energy Capital Partners]? Management mentioned that the company is looking for at least high-teens returns on these hybrid campuses. Does that sound realistic to you, based on your understanding of the business and your history with the Cumulus deal?

[[00:50:21]]
I think it is realistic because the data centres are short electrons. It's not maybe you're priced. I don't want anybody quote me on that. I think that they're in a place of leverage where they… I think they have an ability to have pretty decent returns. Again, they can offer a bigger solution than most of these smaller developers who just want to do… There's a lot of developers out there who just want to go do build 500MW, 300MW of recips or 300MW of smaller turbines and do maybe, maybe don't want to manage all the associated risk of the delivery risk of the electrons and/or the gas molecules, like Talen can provide the whole solution. If I was in the customer's shoes, I would still really think about calling on them because they can provide a better solution than the vast majority of who's out there in the market. As a result, they should be able to charge a little bit higher returns than others.

[[00:51:43]]
All in all, do you think there's any risk that Talen over-indexes to large loads, whether data centres or anyone else, given that its flywheel strategy or even the hybrid model more recently all imply contracting with net-net larger customers? Would there be some material offtaker risk moving forwards? Do you think it is a pro or con that the company contracts increasingly more of large loads?

[[00:52:19]]
Again, there's always the risk of over-indexing into these large loads, but that's the trend, I guess just that they're paying over market pricing. I think about big risk in general to the market is that you have this bifurcated, I think we talked about this a little bit last time was there's this bifurcated world where new gas generation is charging USD 125 per megawatt-hour, something like that. Let's just call it like over USD 100. Wholesale power in PJM is USD 55-60 and capacity is another USD 15. Then yes, you have T&D and whatnot… Ignore T&D, just think about incremental molecules or electrons. There's a bifurcating world of two worlds of like what's the cost of incremental gen and what's the cost of existing gen. The cost of existing gen is, let me say, USD 65-75, maybe USD 75 on the high end, USD 80 on the high end and new gen is over USD 100.

I think the risk is that those worlds don't converge. I think the upside is that those worlds do converge and underlying pricing moves upwards. If underlying pricing moves upwards, does Talen want to be completely hedged by selling all their generation already and not riding that wave upwards. I think that they're long in perpetuity and they have a lot of generation now. I don't think that could be really a big concern because they just have plenty of generation and they should contract some of it. You theoretically could over-index in these large loads, and that is the direction of travel right now.

[[00:54:24]]
You mentioned distributed generation. Are you concerned about the players that have emerged and are offering data centres a more modular distributed on-site energy solution – whether reciprocals, retrofitted, aeroderivatives, smaller simple-cycle units or even fuel cells with Bloom – given that they do trump at the speed-to-power element, which is important? Is this a risk to Talen?

[[00:54:53]]
Of course, it is. It 100% is. All of this, we talked about is, yes, all these things are risks to Talen if they don't mitigate them. Other developers stepping in and building the small stuff. The end-to-end solution of data centres and generation all packaged into one. 100%, all these things are… In the small solutions, these small developers who do speed to power who put orders down and have a 1GW or 2GW of recips sitting on their books is a risk to Talen. Talen has the operational expertise, the balance sheet, the gas and trading availability, that can't be a reason why they don't succeed. That can't be a blind spot. They have to know that's there, and they have to be either teaming up with somebody or going and doing it themselves. Personally, I think that you just go team up with one of those groups and say, "All right, you guys build the first 200-300MW and then the next 700MW that comes online, we own." It facilitates speed to market but brings a customer in for those extra 700MW. It is a risk, but my goodness, I really would hope that they would be able to figure that out.

[[00:56:19]]
Do you think what we discussed could even be a longer-term indirect risk, whether regarding potential competition for gas feed or it becoming latent capacity that re-enters the market in 10 years when some of the bridge contracts lapse and then the data centres are ostensibly connected to the grid with larger generators?

[[00:56:44]]
I think, you just picked up on something that I haven't heard many people talk about. That's really interesting. Data centres, they're going to buy the incremental molecule of fuel to run at 99.999% availability… 99.9% availability. There's only so much gas on the pipeline, so much transport available. If wholesale power prices are USD 70 and incremental power prices are at USD 120, the data centres are going to pay for that incremental molecule. I think the risk is that Talen gets priced out, it's not Talen, it's like the IPPs, they get priced out on the gas side. Then on that peak day that I talked about. All the margin comes in on that one peak day, and there's not enough gas to run their units. They're short on the worst day of the year, they don't make any money and they might even be short, so they could be losing money. I think that's a risk. They should be positioned to try to manage, but it's something costly that you got to manage.

I think there's a few big takeaways that we've talked about today. I think that's something that's really been under (inaudible 58.10) really it's like not how much gas there is in the United States for swing gas, but how much takeaway capacity is there on some of these pipelines? Does that takeaway capacity get constrained? If it's constrained, who gets the molecules? Is it the existing gen or is it the new gen? If the existing gen gets squeezed, you can't get those molecules, then what does that do to the underlying businesses of those companies?

[[00:58:40]]
I'm just going to call them bridge power for now – I'm not saying that they actually are and that they won't get long-term contract. I'm just going to call them that just because they're like alternatives to grid-scale generators such as Talen. For these bridge power providers, if they're getting contract prices where it's a fixed price per megawatt-hour and it's really high because, as you said, data centres pay really high for the incremental megawatt, is that a self-fulfilling loop where those same bridge power providers then also have the funds to bid up gas prices and secure supply, whereas the traditional generators don't?

[[00:59:21]]
Yes. Yes, that's my point. I'm theoretically concerned about. The best capitalised (inaudible 59.34) could protect themselves against that. That's an expensive thing to hedge or expensive item to make sure you're taking out capacity for all days of the year, but you only really need it five days of the year. It's something they're going to have to… I'm sure they're all working on.

[[00:59:56]]
Is there any diversification play for Talen that is not talked about? Given that the company is a pure-play PJM IPP, do you ever see it looking at other markets, such as ERCOT [Electric Reliability Council of Texas]? Do you see it venturing into retail, for example, similar to larger IPP comparables such as Vistra, NRG or even Constellation?

[[01:00:24]]
I think they probably go… I don't know, you could theoretically go into MISO or they can come back into ERCOT. They sold all the ERCOT assets. They manage those assets. There's a lot of generation getting built in Texas. I think that's probably the best opportunity because without knowing any proprietary information, but you know that the major IPPs are all working in Texas because there's 100GW that have to be built there. It sounds like it's an outrageous number, but whatever, it's the number that's being just thrown around. You know the NextEras and the Constellations and the Vistras and the NRGs and everyone else is going to be building generation down there. Why wouldn't they want to get back involved? Yes, I'm not sure. If I was them, I would be getting back involved in Texas. Picking their place where they want to enter with the right customer and maybe they come in with Amazon. They have a close relationship there.

I think you're thinking about things the right way, that's the outside of the box play that they should be thinking about. I don't think it's building data centres. I think the yields on costs are lower than what they want. I think building that power-to-land solution is also something they should be doing, which is what Cumulus was, the Susquehanna deal. They should try to get back into that. We'll see.

[[01:02:01]]
All right. It was a pleasure talking to you. I really appreciate your time. Thanks for joining us.

[[01:02:06]]
Absolutely. Thanks so much.

[[01:02:09]]
Thanks so much. Bye now. Have a good one.

Transcription ends at 01:02:12 of the recorded material