Our first question will come from the line of Mike Halloran of Baird.

Aaron, can you just talk about the capacity conversation? That's the tripling of capacity at least having line of sight to that is a pretty big number. And so maybe talk about how that works out from your perspective. Is this more what you're doing today, which is more on the assembly side? Does this contemplate some sort of arrangement with your engine manufacturer and doing something more joint? And how are you thinking about -- what you need to see from commitments to start reaching those large numbers?

Yes. Thanks, Mike. It's obviously a central question here. We've been focused on this $1.25 billion, $1 billion of that domestically, $250 million internationally by the end of this year and definitely on path for that. In fact, I would argue we're running higher towards even where we'll end of the year in terms of what our theoretical capacity could be. But this pathway to tripling, as I mentioned, over the next 12 months, to be clear, it's only with relation to our generator assembly and our packaging capacity. It does not include anything with engines or any other upstream components at this point. Although those things are currently, as we've talked and as we've discussed in the past, those things are on the table. But today, the tripling of capacity is strictly for our own packaging and our assembly needs. The -- when I say we are on a path to it, maybe to put a finer point on it, that path is one that we're already walking down. In fact, I would say we're running down it. We've been speeding up our efforts to outfit the Sussex facility to bring that online quicker. We were intending to bring that online at the end of this year. We're going to be online here actually in Q3. We've got our first product will be running down the line in August as we kind of prove out the line and then finish out the testing capabilities of that facility. And by the end of Q3, we'll be starting our production ramp there. So that's about a quarter ahead of where we thought we'd be, which is really encouraging. And kudos to our team that is just working like crazy to figure this out. Furthermore, the -- we've identified a number of things that we can do to increase capacity in our existing facility footprint, both domestically and internationally. So Sussex, as an example, we're going to add a second line. And we've made that decision and made those capital commitments already. That's reflected partially in the CapEx elevated -- the elevated CapEx guide from this morning, taking that up to about 4.5% of sales, which is up about, call it, $50 million over our previous guide just in raw dollars. And then there'll be some more dollars going into that next year. But again, I don't see us getting -- we're not going to get over our skis here where suddenly CapEx is going to be 10% of sales or anything like that. It's not -- that's not going to happen. I think we'll stay comfortably below kind of 5% and guiding back towards that as our top line grows, in particular, guiding back towards our historical range at some point in the future. But we've identified a way to get more out of Sussex. We've identified a way to get more out of our Oshkosh facility. We've identified ways to get more out of our international facilities. We have a facility in Asia. We have a facility in Europe. We have a facility in India where these products are manufactured. We'll actually be standing up some production capability in Brazil and in Mexico. So the combination of all of those adds plus what we're doing to add capacity for packaging, that's through the Enercon acquisition and then also this new facility that we just closed on in Belvidere, Illinois, Northern Illinois, that will help us expand our capacity for packaging. So we won't quite be able to satisfy that full tripling of assembly capacity with packaging, but we'll be somewhere, call it, 70% of that number, and then we'll have to augment that with third-party packaging beyond the assembly capacity. So it is a full court press. It's an aggressive plan. Our team is executing against that plan. So when I say we're walking that path, we're running along that path, and we're actually executing against it. Now obviously, the agreement with the second hyperscaler that we announced this morning is a big part of why we've decided to pull that forward in Sussex and to accelerate our capacity expansions. We believe that, that agreement is going to be at least as big as the first hyperscaler agreement that we announced or bigger. And we're also talking about '27 and '28 simultaneously here with that new customer. And we're just working through the final product schedules that are addendums to the master agreement that we signed in late June. So really excited about it. We've -- I think all systems are green, all systems are go, and we're pushing very hard to make that a reality.

And our next question will be coming from the line of George Gianarikas of Canaccord Genuity.

It's more of a strategic question. As your revenue mix shifts more towards commercial and data center applications relative to residential, how are you broadly managing the operational transition? Like when you're trying to optimize capital allocation, sales channel capability, supply chain commitments, how do you preserve your return on invested capital profile given the longer sales cycles and the lumpier nature of large enterprise projects?

Yes. Thanks, George. It's a great question. So obviously, the growth in the C&I business has been aggressive. It's stark. And it's going to change, I think to the point of your question, it's going to change the face of the company in terms of the -- a lot of things. We'll get to your -- the heart of your question, which is return on invested capital and kind of the profile -- financial profile of the company. Big projects obviously bring different sales cycles, as you said, longer sales cycles. We're working through how to match the customer deposits that are parts of these big projects alongside how we cash flow to our supply chain. So it does have ramifications along all of those things. I would say that initially here, the thing that we are observing is that all of the investment level that we've been making, and that's inclusive of not only the CapEx, but also some of the M&A we've done, the payback is remarkable. I mean, based on the volumes that we're projecting. It's very rapid. It's very quick. I think it's -- even with the longer sales cycles, this isn't a dramatically CapEx-intensive part of our business. It's -- there's obviously capital to be deployed, and these are bigger pieces of equipment to move around and to manufacture and to package. But in the end, they're bigger copies of what we've been doing for a long time. We know how to do that very well. And yes, the projects get bigger, the products get bigger. But I would say based on the math that we've laid out, the reward gets bigger. And the paybacks are -- look -- again, look stunning in terms of -- we used to always say when we look at our residential side of our business, when we expand capacity there, you wind the clock back a couple of years when we added the Trenton facility down in South Carolina. And that payback on that facility was inside of -- I think it was inside of 12 months. It's remarkably -- this looks really good, right?

And how good this investment would be -- payback would be inside 12 months.

Exactly. You could argue the same is true of the investments we're making today for large megawatts. So it's...

Just with our current backlog...

Just with the current backlog we have, so that's not even adding in some of the additional opportunities that we have in front of us, which are massive. So we feel good about it. Managing that, of course, maybe to kind of hit the second part of your question or maybe it's the first part. But managing all of that is we have separate teams and separate -- our organization is structured such along the lines of the 2 segments that we have for reporting -- reportable segments here, we have completely separate management teams. We have a corporate structure that serves those teams, and we've had to beef up elements of that corporate structure around our advanced manufacturing teams and others who are supporting the rapid build-out of capacity for our industrial business, our C&I business. But right now, we feel like we're pedaling fast. That business is going to grow. We set out a target to more than double our C&I business in 3 years. We set that target out at our Investor Day in late March. And all indications are we're going to crush that, and we may crush it by next year at the rate that we're headed. So we think that this is a -- as we keep calling it a generational opportunity, we believe that deeply. And we're not going to let that go by without giving it our best shot to see how many points we can put on the board here. And we are very laser-focused on maximizing not only our opportunity, but maximizing the return on that opportunity as well.

And our next question will be coming from the line of Jeff Hammond of KeyBanc Capital Markets Inc.

This is David Tarantino on for Jeff. Maybe just following up on the supply agreements. Could you just put a finer point on the scale and scope of each of these deals? And how much of the $1 billion in orders in the quarter was from the first agreement? And then maybe give us some color on the degree of visibility you have now that these deals are done.

Yes, David, thanks. So to be clear, the $1 billion of orders that we've taken in the last 90 days, included in that number is about $700 million, approximately $700 million for the first hyperscale customer that we announced, and that's all for deliveries in 2027. So just to be clear, the $1.6 billion in backlog we have, if you do the math on the $450 million of data center business we're going to do here in 2026, $250 million is going to ship in the second half out of that $1.6 billion backlog. So we'll have about $1.35 billion and all $1.35 billion right now is scheduled for deliveries in 2027. So our backlog does not contain at this point anything for 2028. Now with respect to future visibility, as I just said, the second hyperscale agreement that we're talking about is actually as big or bigger, frankly, than that -- the $700 million for the first hyperscaler. So -- and we are talking about more definitively, both 2027 deliveries as well as 2028. And I think also putting a finer point on it, that second hyperscale agreement contemplates deliveries on a global basis, for those, at least for '27 for sure, potentially for '28 as well, depending on the schedule of development for that hyperscaler. So very exciting for us. And I think it's -- based on our global footprint, we felt we were in a very good position to be able to execute that. And it's -- it gives us a lot of confidence around the capacity investments that we just talked about. But we don't have any of that second hyperscaler at all in the $1.6 billion at this point. So we'll provide an update to the market when we finalize the product-specific terms, much like we had a notice to proceed with the first hyperscaler, so we had a little bit more to go on, but the second hyperscaler doesn't -- they don't operate in that fashion. So they don't -- they hadn't given us anything that we could share with the market. So we'll provide that detail additionally when it becomes available, and we hope that, that's going to be over the next several weeks.

The question will be coming from the line of Brian Drab of William Blair.

I guess I'll just ask for -- at this point, you've been put through the paces by 2 hyperscalers. You've got a ton of data regarding the capability of the larger format product. How are you comparing with the competition in that segment of the market? And also, how are your lead times right now compared with the competition?

Yes. Thanks, Brian. It's a great question. So just from a specification standpoint, the product that we are offering in market is very competitive on all specs with markets -- with products that are in market today for our competitive set. I would even say that at certain particular nodes, our product, at least this is from the feedback we've received from our hyperscale customers who have tested the product fully and they've tested it with respect to how it performs against other competitive products on spec. And we have -- at a couple of nodes, we're actually performing better, they tell us. And that makes sense because the core engine offering that we're using there is -- frankly, it's a next-generation engine. It's the most contemporary engine design in market today relative to the competitive set. And so you would expect that you're going to get from a horsepower standpoint and a power density standpoint in terms of power output, we're getting better values than some of our competitors at certain ratings. So that's encouraging. And in fact, the ability for the product to take load inside of the 10-second start-up requirements is very good. So all signs point to we are on market or better in terms of product specifications. In terms of lead times, Brian, our lead times remain in that 40- to 45-week range for the products today in spite of the backlog that's grown here. And that really is a testament to not only our growing capacity, but also the supply chain that we've been nurturing here over the last 18 months as we've been working to build that out. We're adding secondary sources to key components where we believe there could be constraints in the future. And we're starting to take control over the packaging element, which again is that -- it's not only about adding value to improve the overall margin profile of the products, but it's also about controlling those lead times because we're seeing lead times stretch out for these overall systems 70, 80 weeks in some cases; some cases, 2 years, depending on the node, depending on the competitor. And some of that is related to packaging constraints. And so it's one of the major reasons why we're investing heavily in packaging is because we want to make sure that, that -- even if we have shorter lead times on the bare genset as assembled, that doesn't help us if third-party packagers have long lead times. That won't -- that equation won't work. So we want to control lead times overall for the products, and that includes a deepening of our packaging capabilities as we go forward.

Next question will be coming from the line of Jon Windham of UBS.

Great results. Rightfully so, there's been a lot of focus on the data center side of the business. But I was wondering if you could just comment a little bit about where you think the residential side of the business can go. Obviously, interest rates aren't being helpful right now. New housing starts are not great. You've obviously been very successful taking market share over the last decade. What are the self-help levers that you can control to maybe get that business back on a growth trajectory?

Yes, it's a great question, and I appreciate, by the way, the question on the residential business because we haven't been getting many of those questions, and it's a great business for us. It was down in the quarter, as we said in our prepared remarks, mainly on the back of our portable generator sales and also the clean energy business, really the storage business being down because of the challenging comp from last year when we had the Department of Energy program in Puerto Rico that ended last year. So take that away. And then power outages were more than 30% below the long-term baseline average in the quarter, and that really hampered our portable generator shipments near term here. Longer term, though, maybe to get to the heart of your question, what I just -- I continue to be shocked at the resilience of home standby generator sales. They're up again in the quarter in spite of the lower power outage environment. Now that you can point to -- we did have some winter storm activity in Q1 that I think was helpful for Q2 shipments of home standby. But Q2 outage environment was weak. So we thought it prudent to kind of reflect that in the back half of the year, and that was a little bit of the recalibration of bringing down the outlook. And it's a modest adjustment there, but a slight adjustment to that. But what can we do -- to answer your question, what can we do to get that category to grow? We can't wait for housing starts. We certainly can't wait for Mother Nature. We can't wait for some of these other macro things to turn when it comes to interest rates. I think what we can do is continue to pound the table on awareness. The category is still only in 6.5% of U.S. households, single-family homes. So the upside is 93.5% of the households that don't own the category today. Power outages, power outage activity, even though near term here, it's been softer than the long-term average. In fact, the last 4 or 5 quarters have been below the long-term average. So it's actually been quite soft. That will return, we believe, to the long-term average here. That's just how weather patterns work. But in the near term, we're looking very hard at a couple of things. One, what can we do to continue to promote improved financing offerings. We know that the consumer is -- that there is some fatigue with consumers. There's some concern about inflation. There's concerns about the overall economic conditions. And when it comes to large ticket purchases, we're seeing softness in other parts of the economy. I mean we -- our customers tell us this in terms of the retail customers we deal with. We see this in some of the other sectors where our contractors and our dealers provide products, we're seeing weakness. And I think it's really important for us to continue to focus on getting in front of potential buyers when they experience an outage. Even though outage activity is down, there's still a lot of outage activity. You look across the board and every single day, we track outages. We track the outage hours. And there -- at any point in time, there are 50,000 to 100,000 households in the U.S. without power. Now it's not because of a hurricane. It's not because of an ice storm. It's not because of some other large format type of outage event. It's from the thunderstorm rolling through the Midwest. It's from a winter event. It's from maybe high temperatures on a particular day that have caused equipment failure, accidents, things like that, that cause outages, they have been increasing those everyday outages. And as we become more dependent in our lives on a digital format in our households, right? We're working more from home. We're learning more from home. We're doing a lot more from home. And that -- those were trends that were long-standing. They accelerated certainly during the pandemic, and they're not nearly where they were during the pandemic, but they're certainly higher than they were prior to the pandemic. And all of those trends point to a need for continuous power. And we're seeing great interest still. We said our in-home consultations were actually up in the quarter. We're focusing our advertising and marketing efforts on -- we're sharpening our ability to go after that. We're sharpening our algorithms. We've changed -- and we talked a little bit more about this last year and at our Investor Day, but we changed the way we generate leads and the way we allocate those leads. We've changed the systems for those. We continue to invest in the platforms there and the teams there. We continue to invest in the products. Our next-generation product line that we introduced late last year is a material step forward, and I think puts us leagues ahead of other competitive product that's available in market. So we're the clear leader there. We have the responsibility to grow that market, to create that market, and we take that responsibility seriously. There are a lot of headwinds that are working against us. But in spite of that, home standby generator sales grew. And that, I think, is a testament to just the resilience and the importance of that category and why we can't lose focus on it. We have to stay focused on growing that category because it's such an important thing for us as a company.

And our next question will be coming from the line of Tanner James of Jefferies.

Congrats on the new contracting momentum. Maybe to that point, can you discuss how you're approaching the funnel for non-hyperscale data center activity and maybe how you're matching order intake to the potential capacity ramp timing given there's going to be some dedication of future capacity to the hyperscalers? And what degree you can or would wish to manage the growth versus profit mix dynamic between the 2 classes?

Yes. Thanks, Tanner. I mean those are important questions. Maybe I'll tackle the first one -- or your last question first, the last part of your question first. There's a difference between the hyperscale business and the non-hyperscale business. And it's not unlike anything in business, right? I mean if you have a customer who is going to buy in larger quantities than another customer, typically, you will see the benefit of scale, right, should read through in a sharper price point for that customer, right? Oftentimes, what ends up happening, though, is you tend to see margin profiles of larger customers that are lower than that of our non-hyperscale customers. They're not dramatically different. I will say that. Now the contract terms can be different. And of course, the timing and the scale, as I said before, is different. How -- so we're focused on having a healthy blend. I would just say that first. That said, the addition of the second hyperscaler is going to throw that kind of mix to be obviously more heavily weighted towards hyperscale business versus non-hyperscale business in the near term. But we are very focused on continuing to add new business through existing colocator customers as well as courting new ones. There are a lot of people in this space that are opportunistic for us to go after. And we're still the new kid on the block. I mean our competitors who have been in market for the last 18 months or 2 years selling these products are well established, well known. And as I said, they have lead times, which are longer than ours. So we have an advantage today to use our shorter lead times to get an opportunity to show these non-hyperscaler customers and the hyperscaler customers what we can do as a company, how we perform. And so we're moving into that phase of our relationship. Now in terms of kind of -- I don't want to use the word allocation because it makes it sound like we're at a fixed finite level of capacity, and we're saying hyperscalers are going to get x percentage of that capacity and non-hyperscalers are going to get y capacity. Today, we're growing capacity. And obviously, we're talking about tripling our capacity from the $1.25 billion to something closer to approaching $4 billion in the next year, 1.5 years. That -- we don't have that filled up today, but we are courting customers who have needs that they have projects that they want to bring online faster. In some cases, what we're hearing, in particular, from this particular customer set is that they have projects that if they could get backup power, they could turn the project on quicker because of the lead times of the backup industry today. We can help them with that. If they want to shift those orders to us, we'll give them our slots for 2027 and into 2028 to help them bring those data centers online faster. Those are the kinds of conversations we're having today. And we just -- we're getting success there. I mean we -- included in our backlog is another large colocator that we landed in the quarter, and it's new for us. It's a new colocator. So really excited about that. We continue to grow our existing colocator relationships, getting new projects there. But the funnel, if you look at the whole funnel of opportunities, it's massive. I mean it's just -- it's embarrassingly...

A lot of interest.

Yes, a lot of interest. It's embarrassingly large in terms of the total number of projects we're quoting and the total number of conversations we're having, both hyperscale and non-hyperscale.

And our next question will be coming from the line of Praneeth Satish.

I guess just high level here, when we think about -- in your conversations with hyperscale customers, how are they thinking about backup generation for behind-the-meter data centers? I mean, usually for the front-of-the-meter, grid-connected sites, it's usually a 1:1 ratio. But I guess is that still the right framework for behind-the-meter deployments? And are any of the sites that you're bidding on or contemplating shipping to, are they behind the meter? Or are you kind of primarily focused on front of the meter?

Yes, it's a great question, Praneeth. And so the front-of-the-meter, grid-connected customers, that's an easy one. They need to have backup power. Behind the meter, prime-power generated, self-generated power, it depends on the power format. I would say what we've -- what we've done is we've -- as we've talked to the addressable market, as we've defined it and as we defined it at our Investor Day, we excluded some parts of the behind-the-meter power generation that's going on there, in particular, where a customer may be deploying certain technologies like fuel cells. They believe that there's not a need, particularly for our type of product, a backup, a reciprocating engine-driven backup genset. But those projects are far and few in between today. There are other types of behind-the-meter technologies, behind-the-meter power generation technologies. You've got turbine technologies, well-known technologies. In some cases, there is backup power required there because you could just as easily lose a turbine. Now maybe you have N+1 redundancy in the way that the turbine array is constructed. So maybe you don't feel like or that developer or customer base doesn't feel like they need 100% backup or 120% backup. I think at least today, I think what's -- this is my thesis and people can take and apply this logic if they want or they can use their own logic. There are certain of these behind-the-meter technologies, turbines being one of them and where they're using recip gas engines in particular. Those technologies are bridge technologies. They're a bridge to an interconnect to a grid. They are not permanent technologies simply because from a cost of energy standpoint to produce power on site with those technologies, it's going to be more expensive than it will be to secure grid power. In particular, the risk that comes with producing our own power in terms of hedging the cost of the fuel stock, the input, the natural gas in this case and those -- most of those technologies. So my point there is that we believe that longer term, as grid interconnects catch up, and as that queue shortens instead of lengthens as it's doing today on the grid interconnect, there could be more opportunities for backup power to present. It just may happen later when the grid interconnect time line becomes clearer. So today, we -- as I said, we've kind of adjusted our total addressable market to take out certain behind-the-meter power generation capabilities where certain data centers are going to do that -- do that on their own. I do know we have a couple of projects we're involved with, where we're supplying some backup power into some sites where they are going to generate their own power on site with certain technologies. I don't have the specifics of that, but I do know there are some projects that exist today where they're taking backup generators because they still need them. You could still have a turbine go down. You could still have part of your recip engine array go down. You could lose those technologies and you need to back them up, nonetheless, even if they're bridge power.

Our next question will be coming from the line of Manish Somaiya of Cantor.

Congratulations on all the wonderful announcements. A couple of questions. One, I guess, is there -- Aaron, is there a hyperscale customer #3? And how far is that customer behind #2? That's my first question. And then just related to that, I think you talked about some of the capacity constraints. And from my perspective, I'm also a little bit more focused on the supply chain, as it relates to generator assembly, alternator, switch gears. I think you discussed the case -- the packaging part, but perhaps if you can just give us a sense of what's going on with the context that I just provided.

Yes. Thanks for the question, Manish. So maybe I'll start there with the supply chain. Yes, the supply chain capabilities, like we can talk about how we want to triple our capacity. But if our supply chain can't keep up with it, it's kind of all for naught. And if we believe that, that was going to be a real problem, it'd be difficult for us to make the case that we should be investing in tripling our capacity. So I would tell you that alongside of tripling our capacity, we have a deep road map around how we can also triple our supply chain capacity. And that -- as you get into kind of the second order componentry that you listed, some of those components around alternators, there are cooling packages, there are some of the structural elements of the generator with base tanks and frames. Some of that we're going to undertake internally. We're going to kind of make versus buy type of analysis. We're going to have to augment some of that with external purchases as well and then the canopy components that go into the packaging and some of the other elements. There are some of the after-treatment systems that go into the diesel systems where we're seeing emissions levels where you've got local -- either local codes or you've got local permitting requirements that are calling out best available technology for the after-treatment of these products. Those also require additional components. All of those areas, we have second and third sources identified, and we're in deep negotiations with those for supply starting in 2027. So at this stage of the game, we feel like we have a road map, both for our own internal consumption, our capacity, that tripling of capacity internally as well as with our supply chain. Now there's a lot of work to do, both for our -- tripling our capacity internally here, but also with our suppliers to make sure that they're going to be capable to deliver at those rates. And of course, they're being -- their own capacities are being added to and tested by the growth here in the overall market for data centers. But at this stage of the game, we feel like we're in a really good spot with our supply chain capacity. Now on your first part of your question on the third hyperscaler, we -- look, we have a lot of conversations we have with a lot of customers. We actually serve another hyperscaler with smaller backup power and have served them for a couple of years where they have edge data center infrastructure. We back that up with our smaller generators. And when I say small, these are 500-kilowatt diesel and gas generators. So they're not that small, but they're small in the context of a hyperscale requirement. We continue to talk to these customers about their needs. And as the market evolves here, we could have, I think, additional opportunities. Today, I would tell you, whereas maybe we were on -- I think we described it, it was a 100-meter dash before, we were on the 99th meter of that 100-meter dash with hyperscaler 2 with any future hyperscalers at this point, those are just discussions and they're very early if they're having -- we're having them. So I wouldn't bank on anything coming anytime soon there. Obviously, where a big focus of our pipeline is building out the non-hyperscaler portion of that as well. We want to create better balance there. And those conversations are very robust, and that funnel is very robust, and we continue to get wins there. And I would assume that a good chunk of our backlog growth in the future here is going to be from the firm orders we get from hyperscaler 2, once we are ready to announce those and then additional non-hyperscaler orders as we work through our funnel and we progress through the sales cycle on that side of our business.

And our next question will be coming from the line of Keith Housum of Northcoast Research.

Aaron, just as we think about the data centers, and obviously, it's important to everybody here on the call. As you think about the marketplace, it seems like a lot of the emphasis has been on domestic hyperscalers, understanding the second hyperscaler has some international business. But does your product appeal to international hyperscalers as well? And is that just a little further longer to develop? Or is there a difference where that market may not be as open to you guys?

Yes, it's a great question, Keith. I would say that there are areas of the world where I think we're going to be very competitive. As an example, outside the U.S. -- first of all, just to level set, the U.S. market for data centers is the biggest market in the world, far enough, right? I mean it's just -- it's orders of magnitude larger at this point than almost every other region in the world. But that said, we have really interesting opportunities in Europe that -- because we're based in Europe, products are manufactured and we have a manufacturing facility in Italy where the products are of this size range are manufactured. We're seeing opportunities within the European theater where I think we can be very competitive. Some of our first projects were in Australia, and they were delivered out of our -- we have a factory in China. They were delivered from that Chinese factory down to Australia. Head-to-head in certain markets, like let's just go there, but the Chinese market for data centers, right? The domestic Chinese market for data centers, probably going to be very difficult for us to serve that market competitively. There are a lot of producers of gensets in that range in the Chinese market. And I don't think that the things that we bring to the table for a data center owner-operator developer, such as our brand, the strength of our balance sheet, our ability to serve, our ability to have an international footprint or a global footprint. I don't think they're as well appreciated perhaps in certain markets like China. So I wouldn't expect that we're going to be announcing a bunch of Chinese data center wins. But you get outside of China, India is a growing market. We have a facility in India to serve that market. The South American continent, you get Mexico and so Central South America. We have a big facility in Mexico that serves the Mexican market and can serve the South American market. We have a facility in Brazil that can serve the Brazilian market. That footprint gives us an advantage. And for those data center developers that want to make sure that they're buying a product from a company that's going to be around to support that product. And look, people -- there's a low bar for people to bolt these things together. There's no question that like the -- and this has been the case in generators in general. It's not just large generators. It's smaller generators. What you have to wrap around that, right, is the service network. What you wrap around that is the support capabilities. What you wrap around that is the certainty of a warranty and the certainty of the ability to get somebody to a facility or on site when there's a problem, because uptime is absolutely critical. And so for developers who take uptime seriously and understand the bit of a premium that they may have to pay for that, we're going to do well. And we are doing well. We're seeing that in the domestic market. We're seeing it in Australia. We're seeing that in Europe, where we've got project wins already.

Consistent global processes, they covet that.

Yes, consistent global processes, as York was just saying, one of our hyperscale customers, they are absolutely laser fixated on how do we replicate the processes that they validated here in our U.S. factories, how do we replicate those in other factories around the world if we're going to produce product for them for other site deployments. So I just -- I think that the international market is a good opportunity. Last point, we obviously, we have an exclusive arrangement with our engine supplier here in the U.S. market. We don't have that same exclusivity outside the U.S. So that does have -- that does probably play a role as well, where we can be competing against other packagers using that same engine package outside the U.S. So that's going to have an impact on the overall market dynamics and our ability to succeed in some of those other markets. Again, the closer you get to the Chinese market, I think it's going to be more challenging for us in Asia to be successful outside of perhaps India, if you think about that as part of Southeast Asia.

And our next question will be coming from the line of Vikram Bagri of Citi.

I wanted to understand if you're seeing difference in inference data centers attaching more sort of like -- or less capacity backup capacity versus -- let me rephrase that. Our understanding is that the training data centers will have less than 1:1 backup capacity versus inference will have more than 1, if you've seen that. I was also trying to understand if the increase in hyperscale order from initial expectation of $600 million to $700 million is a function of that or you're gaining market share from another supplier to these hyperscale customers? And related to that, you currently have a significant lead time advantage over incumbents in large capacity generators. Do you think there is a premium in pricing for quick delivery right now? And if so, is there a way to frame how much that premium is for sort of speed to delivery?

Yes. Thanks, Vikram. I appreciate the questions. Maybe I'll start backwards and with your question there. Obviously, the lead time advantage we have, I would say that today, the market overall is higher on a price per kilowatt than actually than we framed the opportunity when we put the business case together to go into the large megawatt genset market. We put that business case together 3-plus years ago now, almost 4 years ago. And we were -- the market at that point was on a lower price per kilowatt basis. Now there's inflation and some other things that have driven that up, but I think it's primarily being driven by the fact that there's a supply-demand imbalance there. Structurally, we're taking advantage of that, obviously. With our shorter lead times, we're able to supply that market that is paying more for access. I don't know that I can put like a number on the premium there. I would tell you that even though we have shorter lead times, the customer base there is not willing to pay more than market, even though we can produce the products and deliver them in a shorter time frame, I don't think we're getting an above-market premium today versus our competition on a per kilowatt basis, slight, if anything. I think all that we're doing is we're able to capture some share here very quickly because of that shorter lead time. That's kind of our assessment of the situation. I think that -- could that change over time? That's probably the bigger question. Like what does the price per kilowatt do over time? My experience in markets where you have supply-demand imbalance led price escalation, oftentimes, what you'll see is the price -- the growth of price slow, but you won't see it go back down. Price rarely goes in reverse. If you're -- especially if it's a non-commodity type business -- commodity-based business, it's -- you're not going to see price go in reverse. I think you'll see the growth of price probably slow, meaning I think these products probably will stay priced in the range that they're at here for the foreseeable future, and -- which is good for us because it's well above, as I said before, kind of how we frame our business case. Now the first part of your question on inference -- inference use data centers versus non, agentic, we do see different approaches based on the use cases for the data center. So you will see a greater percentage of coverage for certain data centers where you're talking about cloud connected versus where they're doing inference or learning model type of training data centers, where maybe uptime, they have different levels of uptime requirements. They actually get very precise, depending on the customer. They can get very precise about what that uptime number is depending on the use case of the facility. And then dependent on that uptime number, that's how they size the backup requirement. So if that uptime number is greater, it's 99.999% per se or even more than 99.999%, you're going to see a greater percentage of coverage to the total potential consumption -- power consumption for the data center. If it's less on the uptime scale because the data center is perhaps doing things that can be viewed in the context of noncritical, and there are differences beyond just inference and non-inference, but that lower uptime statistic then would lead you to a lower coverage rate. So it really is data center dependent. It's customer dependent because they all have their different views on what the right uptime statistic is. But today, we're seeing a healthy coverage of at least 100% and in some cases, up to 125% to 130% of the power consumption of the facility depending on the customer and depending on the use case of the data center.

And our next question will come from the line of Christine Cho of Barclays.

I just wanted to ask about your engine supply. I know you've mentioned that the supplier has no problem growing with you and that will never be a constraint. But as we think about you tripling capacity in the U.S., would that require your engine supplier having to add or maybe diversify their manufacturing capacity in places like the U.S. or Europe? And if so, should we think that the mix shift could impact gross margins in the future if it's skewing towards more expensive manufacturing [indiscernible] with the cost of engines going up. You talk about being priced competitively, and I think you've also talked about how your cost structure is very attractive. So just curious if you think you'll be able to like raise pricing in that case and maintain margins? Or would that impact demand?

Yes. Thanks, Christine. I mean it's a great question. Our engine supplier is -- they have indicated to us that capacity is not an issue. You used the word never. I'm not sure I'm ready to say it's never going to be an issue. But they've told us that based on -- and we're sharing, by the way, our capacity. We had a team that was just over working with our engine supply team here just last week, as a matter of fact, working through some of our kind of forecasting for '27, '28 and where we think volumes could go. And all indications are that based on their current footprint and the additions they are making, by the way, so they are making investments and additions to capacity, they have, in particular, a facility in France that they are adding a lot of capacity to because we have some customers that want to have engines that are going to come out of the European theater as opposed to coming out of Asia. So they've requested that specifically. There's a bit of a premium to that, to your point, maybe around what that looks like in terms of the -- but we've been able to roll that through with higher pricing. If that's what the customer is requesting, we've made it clear that, that could mean a higher price based on -- based on that decision. Now there are maybe offsets for that. If the customer is going to locate those products in Europe, as an example, there might be lower logistics costs for them, right? So it's lower packaging costs for them. So it's -- there's kind of an overall economic kind of model that they have to run through to understand whether our price increase on the genset because they want an engine that's coming from a higher cost region of the world that, that works in their economic model overall. I suspect that, that's going to be the case where our engine supplier is going to be increasing capacity in other regions of the world, potentially the U.S. as well. We've said this before. They continue to evaluate that. And we're evaluating alongside of them to see what we can do to help them with that should they require that. But they are a very competent company. They have a lot of really great talent within their company and a very aggressive stance on their ability to produce these products wherever they need them on the planet. And so they -- right now, they are working through their kind of footprint for the future. I can't speak too detailed about that. It's their footprint. So they would really need to speak to that. But I can tell you that in the one example we have, where we have customers who might want engines out of a different region, we've said that the price is higher. So we're trying to maintain our economics. And at least that's how I think we would treat it going forward. I don't have an idea of what a cost structure will look like in the U.S. We haven't gotten that far yet. So I don't know how to assess that. But obviously, there will be a tariff offset there in that instance, right? So there's a lot of moving pieces with that, and the tariff environment continues to move as well. So we're going to continue to watch it and put together the right economic models. And if we need to adjust pricing, we will. And if our supplier there, if they -- if we need to work harder with them to make sure the cost structure stays competitive, we'll continue to do that, too. But I have no doubt that our supplier is going to be cost competitive. They are laser-focused on being a leader in their cost structure for these types of products.

And I would now like to turn the call back to Kris for closing remarks.

We want to thank everyone for joining us this morning. We look forward to discussing our third quarter earnings results in late October. Thank you again, and goodbye.

And this concludes today's conference call. Thank you for participating. You may now disconnect.