Capital Markets Day Transcript - SEDG 2026-09-10

Speaker 7

20 years ago we asked one question how do you get more from every watt the answer started at the power optimizer a dc optimized architecture panel to battery year after year we scale ohms commercial megawatt hours of production then the grid changed. Dynamic tariffs. NEM 3.0. A watt stopped being worth the same at every hour. So we gave the system a brain to know when a watt is worth most. Now the battery has a bigger job. Once just backup. Today also the power to act. SolarEdge Nexus. Groundbreaking efficiency and high and low Power, a DC coupled end-to-end system. Powering high growth opportunities for new residential customers and upselling to our millions of existing sites. But this was only the start. Everywhere, energy stopped being abundant. A watt is worth more. And nowhere is power scarcer than an AI data center, where the racks run on DC too. So we developed a solid state transformer. Grid power in, data center power out, almost nothing lost. From a roof to the most valuable market on earth, opening a new multi-billion dollar market opportunity. 20 years of DC architecture expertise, we built for this world before it arrived.

Michelle Spina, Other

Good morning everyone and welcome. I am Michelle Spina. I'm the Director of Residential Marketing for SolarEdge North America. I'm so excited to have you here today for SolarEdge Investors Day. Whether you're joining us in person or online, I just want to thank you for being here and taking the time. The video that we just saw actually serves as a reminder that for the last 20 years, SolarEdge has been driven by one simple belief, that optimization is the key to the future of energy and that every watt matters. Energy is becoming one of our most precious resources and our mission has never been more relevant. Every watt generated, stored, and managed in a more intelligent way serves an exponential value for homeowners, businesses, communities, and increasingly emerging markets. Today's story is about how SolarEdge continues to turn that belief into opportunity, innovation, and growth. Before we begin, I want to draw your attention to our safe harbor statement. So we will jump into the main agenda shortly, but a few brief stats on SolarEdge and where we're at today. SolarEdge technology powers approximately one in three homes. With more than 4.6 million monitored systems worldwide, we've shipped 65 gigawatts of technology globally, delivered more than 150 million power optimizers and can be found on 60% of Fortune 100 rooftops. Our presence spans global markets, supported by regional teams, world-class R&D centers, U.S. manufacturing facilities, and more than 96,000 certified installers worldwide. These achievements are important not just because of scale, but because they provide the foundation for what's next. And I look forward to taking you on that journey today. So let's jump into the agenda. First, you're going to hear from our CEO, Shuki Nir, who will share how we've returned to profitability and the strategic framework guiding the next phase of growth. Lironi Nav will take you through SolarEdge Nexus and explain how we're positioning the company to capture the growing opportunity in residential energy storage. You'll hear directly from a customer's perspective with a conversation featuring Prologis, offering insights into what industry leaders are thinking about the future of energy. We'll then take a short break and reconvene with one of our co-founders, Mir Adest, who will take us beyond our existing markets and into one of the most exciting opportunities ahead, AI factories. Finally, CFO Meoz Sigron will bring everything together through our financial framework, outlining the path we see towards profitable, sustained growth. We will end with a Q&A session with the leadership team to provide additional clarity on any items that are important to you. What connects all of these discussions is a single story, leveraging two decades of innovation and execution to expand the opportunities in our core markets and our emerging ones. So with that I am pleased to introduce our CEO Shuki Nier. Please join me in welcoming Shuki to the stage.

Shuki Nir, CEO

Hey, good morning everyone, and thank you for joining us today in person and online. Well, very excited to have you here for our Investor Day, in which we are going to share with you the next chapter in SolarEdge's journey. We will start by talking about what brought us here, but then we will spend most of the time talking about the huge opportunities we see ahead of us, and why we believe SolarEdge is well-positioned to benefit from them. Let's start with where we are today. At the beginning of 2025, I shared with you the four initiatives, the four pillars that are required in order to turn our business around. Ensuring financial stability, gaining market share, driving innovation, and ramping up our U.S. manufacturing. through 25 and into 2026 we made very good progress in each and every one of them we reduced our expenses we we moved into positive free cash flow in 25 and 26 we went out of non-core activities we did gain share in some of the segments we're operating we introduced products that our customers really need like commercial storage nexus and others and we ramped up our use manufacturing to the point that the vast majority of our inverters and optimizers are made domestically. What you're seeing here is the financial result of that progress. Six consecutive quarters of year-over-year revenue growth. Six consecutive quarters starting from $212 million of revenue, going all the way up to $345 million in the second quarter. Six consecutive quarters of margin expansion. And throughout this time, we continue to reduce the EBITDA loss until the last quarter in which we move to a positive EBITDA. Let's be precise about what we are seeing here. This is not a celebration of a milestone. This is not one successful quarter. This is a trend line going up and to the right. And this trend line is what we are going to build on as we go and talk to you about the future. Now, a word about the second half of 2026 before we talk about the future. All right? As you know, our guidance, our revenue guidance for the third quarter is $310 to $340 million. Today, we are reiterating this guidance. We do not guide beyond the current quarter. However, we assume that Q4 is going to be slightly below Q3 due to seasonality. For convenience, in today's presentation, you will hear us referring to estimated revenue in 2026 as $1.29 billion. Again, this is not a guidance. But we didn't invite you here to talk about 2026. We've invited you to talk about the next chapter in SolarEdge's story. We've invited you to talk about how we are going to turn strong foundations into profitable, sustainable, profitable growth. Now, for the last two decades, SolarEdge has built something that is really hard to build. Deep expertise in DC power conversion and power electronics. The ability, actually, to take that technology, make the products at scale, at gigawatt scale, and deploy them in millions of homes and businesses, this is the foundation. More than 600 patents, more than 65 gigawatts installed, more than 4.6 million sites monitored through our secured cloud. This is the foundation upon which we are going to build everything that we are going to talk to you about. And the growth opportunities are very exciting. We plan on accelerating our growth in the core markets. The core markets are changing in a direction that is good for us. And today we will share with you more details about Nexus. We will dive into our strategy in the evolving residential market. and we will share with you why we believe that we will continue winning in CNI. Then we will talk about how we plan on unlocking a multi-billion dollars opportunity in the AI factory. Okay, the market is evolving quickly. We will talk to you, we will share with you our estimate, our perspective about where the market is heading, how it's growing, and why we believe most of it is going to be 800 volt DC native. We will share with you the details of the technology behind our solution. And why we believe, and players in the ecosystem tell us, that we are very well positioned to benefit from that opportunity. Before we start going into each one of them, let's talk about the underlying capability underneath all three of them. Power has never been in higher demand. And everything SolarEdge does comes back into one, into one, single value added. Our customers are getting more usable energy from every watt that is going through our system. They get more usable energy that goes through our system, from every watt that goes through our system. It may sound simple, but it is not. It is a result of 20 years of DC expertise, deep understanding in the technology, safety, efficiency. And this is a result, it results in residential, a 10.5% more energy harvested from every home. This is 10.5% more energy harvested compared to string technology. It was validated by VDE Renewable, a third-party research firm. This is 10.5% more energy every day, every year, through the lifetime of the system. This is the reason why more than 60% of the Fortune 100 companies chose SolarEdge. More than 60% of the largest companies in the country have SolarEdge systems installed on their rooftop. And that is the reason why in AI factories, we are the only company that is about to introduce an SST, a solid state transformer, with efficiency above 99%. Three different markets, three different types of customers, one underlying capability underneath them all. Everything that you will hear today is going to tie to one of these three. And I would like to start by talking about our core markets. Our core markets are going through a real shift. It has started to happen already. the move from PV-only to PV-plus storage. Until recently, a solar system, the value of a solar system, was the value of the electricity it produced and exported. That is now changing. Greeds are more congested, tariffs are changing by the day, and export compensation is going down, and in some countries it's actually eliminated altogether. Customers are no longer interested in production only. They are interested in control. They want to control what happens with their energy. They want to make sure that they maximize the ROI, the return on investment from this system. So backup, time-shifting, grid flexibility events are the new normal. And that shift in these markets is actually creating an opportunity where the addressable market is going to grow in dollars by 2% to 3% every year in the next 3% to 4%. More importantly, storage is going to be the bigger piece of the pie. And that opportunity and that change, these dynamics have created two opportunities in the residential market for SolarEdge. The first one with new customers. When a battery is attached in a new installation, the revenue for SolarEdge more than doubles. So instead of selling only the inverter and the optimizers, we are selling the inverter, the optimizers, and the battery. So think about it. As attach rates continue to grow globally, the average revenue for SolarEdge per installation is growing. The second opportunity is the existing install base that we already have. Globally, we have more than 4 million residential sites. Think about it. It's a lot. 90% of them do not have a battery today. It doesn't mean that each and every one of them will have a battery later, but this represents tens of gigawatt hours kind of opportunity for us. In the second quarter, and we shared it with you, in the second quarter, we generated $20 million from upsell activities in Netherlands and Germany alone. In Netherlands and Germany, we have more than a million sites. In the second quarter, we generated $20 million, and this is just the beginning. So these two exciting opportunities required us to bring a platform that was designed from the ground up to address these opportunities. And earlier this year, we launched what we refer to as the most important launch in the last decade for this company. We actually brought one here, SolarEdge Nexus. Liron will take you through the details later on, but suffice to say that this product, since its introduction, has actually received many, many different accolades. You know, Forbes recognized it as changing the future of home energy, right? And it was named Green Builder Sustainable Product of the Year for 2026. But the number I'd like to draw your attention to is this number here, the $7,000. These are $7,000 of additional return to the system owner. These are $7,000 through the life of the system that an owner of a solar edge system is going to get versus an owner of a competing system. So these $7,000 are not just a line in a spec sheet. This is the incremental profit for the homeowner and those who actually appreciate it even more for the TPOs. And that's another reason why installers are going to choose Nexus. So that is about residential. Let's switch to the second segment of our core markets. In CNI, we are winning. Less than two years ago, in the first quarter of 2024, SolarEdge accounted for 29% of the installation in rooftops in CNI. Fast forward to Q2 2026, we account for 57%, moving from about a quarter to more than half. This growing market share is actually growing our install base in CNI, which when attach rates for storage are going to grow we are going to see the same opportunity that we talked about in residential storage attached to the install base and this growth in market share is not a single quarter phenomenon it is a result of a structural shift and the structural shift has two drivers behind it the first one is technology and the second one is regulation So let's start with technology. SolarEdge systems harvest more energy. And for an enterprise customer, for a large CNI customer, that means that more dollars are flowing through their bottom line. And they like it. More importantly, because think about it, they install the system on the rooftop of their businesses. Safety is their number one concern. And for the last two decades, SolarEdge has been known for the safety of its systems and enterprise customers recognize that. The second thing is around the regulation. Non-FEOC, FCC authorized, domestic content made in the U.S., none of the major C&I competitors can claim all four. We can. And the large C&I customers, the enterprise customers, they appreciate that because it delivers a ton of value to them. And that's the reason why leading enterprise and C&I customers are choosing SolarEdge. I don't know which one you drive, Mercedes or Toyota, whether you like DHL, you like Shell. These are only a few names, more than 60% of the Fortune 100 companies. And today, we are honored to host an executive from one of our strategic enterprise customers, Prologis. Alta Yen is the senior vice president, head of the Energy U.S. at Prologis, and she will share with you and with us their perspective on the market, where it's heading, and why they chose to partner with SolarEdge Global. So in both residential and CNI, we see tremendous growth opportunities ahead of us. But the reason that we are more confident in our ability to gain share in these markets is related to safe harbor transactions. So let's talk about this number, the $3.3 billion that you see here. For the last several quarters, our teams have had multiple conversations with the largest TPOs, financing companies, enterprise companies, large C&I companies. And in these conversations, we try to understand what type of safe harbor transactions they prefer. And as you may imagine, the vast majority of them, they prefer the physical work test one. No cash laid out, no inventory has to be taken ahead of when they really need the product. So these large customers, these key customers, both in Resi and CNI, in recent quarters, as of today, have signed $1.7 billion worth of contracts with us. $1.7 billion of firm commitments for delivery between 27 and mid-2030 with liquidated damages if the customer chooses to cancel. Now, most of the contracts in the residential space are for the inverters. And as you all know, our inverters need optimizers in order to work. So when you add the expected number of optimizers that is going to be added, that's another $1.6 billion, and you get to the $3.3 billion opportunity that we believe is out there for us. Now, I'd like to be careful here. Not all commitments convert. Some customers may go out of business. Some customers may decide to cancel, and liquidated damages are not at the same value as the transaction itself, obviously. But nonetheless, this is the largest forward visibility this company has ever had. And the vote of confidence from both key customers in Resi and key customers at CNI give us the confidence that we can see future market shares as well. Future market shares gains as well. Now, one last thing. We talked earlier about the transition to PV plus storage. If one assumes even 30% attach rate, a conservative 30% attach rate, this number is going well beyond $5 billion. So that concludes, or that's how we look at the growth opportunities in our core business. And now let me switch to the multi-billion dollars opportunity we see with AI factories. now for 20 years we focused on converting DC power efficiently and safely at scale now this expertise meets once in a generation power architecture revolution as you know AI factories growth is constrained by power it's constrained by because grid connections are the bottleneck and because the rack power density is expected to exceed one megawatt in 2028. AI factories realized that the laws of physics will require them to move from AC architecture to DC architecture. And what we would like to share with you is the size of the build-out, why we believe most of it is going to be DC native, 800 volt DC native, and why players in the ecosystem are telling us that we are in a leadership position. So let's start with the size. AI factories are being built at a pace this industry has never seen before and a power density this industry has never seen before. Estimates for AI factory growth in the U.S. alone are between 12 to 15 gigawatts in each of the next four years. And you can see it here in the slide. And this massive build-out is running into a wall. This is not, in our opinion, this is not going to be a GPU wall. It is going to be an energy wall. And people have started talking about it already. So again, at the same time that the industry is going through a massive build-out, NVIDIA, and more recently, OCP as well, the Open Compute Project, have realized that switching to DC architecture is a necessity. And in 2030, as you can see here, 67% of the new builds, two-thirds of the new builds are going to be 800-volt DC-based. That's two-thirds from practically zero this year. Okay? Now, Meir is going to share with you later on that there are several different ways to get to 800 volt DC. But not all solutions are created equal. On one end of the spectrum, you have the sidecar based on AC architecture. On the other end, you have the solid state transformer. The sidecar is easier to get to and it carries less risk, but it is less efficient. The SST is the holy grail. And now the question is why AI factories are going to take the risk with a new technology that is not yet proven when they are going through this massive build-up. So in order to understand it, one needs to understand the value of efficiency. You are numbers people, so you can check my math here. Let's take a 100 megawatt AI factory. If one can recover 1% of efficiency, 1% of incremental efficiency in the powertrain, that means that you have one additional megawatt available for GPUs. This one additional megawatt doesn't require new grid connection, no new substation, no new permits. This is power that has already arrived at the site and was lost on the way to the chip, to the wreck. So this one additional megawatt, in today's economics, will generate $20 million. Again, $20 million of incremental revenue for the AI factory owner every year. So if you run a net present value over seven years for $20 million a year, you get to $100 million. So a value of $100 million as a result of 1% incremental efficiency. That's the math that AI factories, hyperscalers, and other players in the ecosystem are doing. And that's the reason that the industry is moving and moving fast. Now, this is the same build-out. Again, you see between 12 to 15 gigawatts of new builds. But here we broke down the 800 volt DC into AI factories that will decide to stay with the AC infrastructure versus the AI factories that will go DC native. As you can see, in 2027, DC native is just a sliver. In 2030, it is expected to be 50% of the new build and the vast majority of the new builds that are going to 800 volt DC. The industry is changing the architecture of the powertrain and it does it within this decade. Now, how does it translate into total available market in dollars? When you put all of these estimates together, we go from $200 million of TAM in 2027 to $4.1 billion of TAM in 2030, and probably continue to grow exponentially from there. Which now leaves one more question. Why SolarEdge? The opportunity is big, but why SolarEdge should be the one to benefit from it? So, for the last 20 years, we focused on DC coupled architecture. We are DC people, right? This experience and expertise led to three key advantages that players in the ecosystem find very, very attractive. Efficiency, safety, and scale. Let's start with efficiency. SolarEdge provides a DC-native powertrain that goes from utility to REC, from 34.5 kilovolt AC to 800 volt DC, with more than 99% efficiency. Competition, the good ones, are at 98, and we've already established the value of 1% efficiency. In a 100 megawatt AI factory, it is equal to approximately $100 million. $100 million is more than the cost of the entire system. So you can do the math, AI factories can do the math, other people can do the math. The second point is around safety. And again, the transition to 800 volt DC is going to depend on whether these systems can be protected, serviced, and trusted at scale, right? That's the way the data center owners are going to think about it. So at SolarEdge, as we told you, we are building the full solution. And earlier this morning, we shared with the world that we wrote, we actually authored a white paper with NVIDIA's contribution to help advance key protection and grounding considerations, right? Within the broader industry. Now, in the many conversations that we've had with prospects and with NVIDIA, it became clear to us that solving these grounding and safety issues is a top priority for the industry. Similarly, the same expertise led to our announcement with Infineon yesterday. And again, together with Infineon, we are enabling solid-state protection for these data centers. Infineon, like NVIDIA, have realized that safety, solving the safety issue is a requirement for mass adoption. And they, because they've worked with us for a long time, they turned to SolarEdge for the DC expertise and the DC safety expertise that we've developed over the last two decades. The third element or the third advantage is around scale. Future customers are going to benefit from our supply chain, from the scale of our supply chain. We've installed more than 60 gigawatts. The reliability of the product is out there for them to see. And our U.S.-based manufacturing footprint gives them the confidence that we will address supply chain resiliency and will help them with domestic content consideration when and if it becomes applicable. So these advantages are, we are being told by both, as I said, horizontal players and prospects that these advantages really matter to them. And based on what they say, we believe that we are ahead of the competition, both in terms of the product's capabilities, as well as in terms of a timeline, where we are. So how will it translate into revenue? We shared our timeline with you. We've shared it multiple times. We expect to have a working product in our labs by the end of this year. We expect to have pilot installations in 2027 and initial revenue in 2028. This is in line with the roadmap that NVIDIA are talking about, everybody is talking about. In 2029, when there is mass adoption, we expect to generate $600 million from the AI factories revenue. 600 million dollars of revenue in 2029. So let me put together both growth engines. Our expectation is that in 2029 we will generate 2.4 billion dollars of revenue. 1.8 billion is going to come from the core business and $600 million are going to come from the AI factories. 23% CAGR. With the core business, I shared with you the opportunities that we see, and the core business expected revenue is dependent upon products that we have and customers that we know. Or not products that we know and customers that we have, but also as well. The AI factory's revenue is going to be very exponential, as I showed with you in the last slide. And it's grounded with our belief that the industry is going to move quickly to this native architecture and that we are best positioned to benefit from that opportunity. But revenue is only part of the picture. Margin expansion completes the picture. Our estimated margin in 2026, excluding HIPAA, Mars will take you through all the details, is 24%. We expect, we target, 35% in 2029. This margin expansion is going to result from scale, from mix, from nexus, from US manufacturing, from operational excellence. All of these things are under our control. Since the beginning of 25, we've proven to you that we can apply discipline into these processes that result in increased and expanded margin. And our intention is to apply the very same discipline in the coming three years, targeting the 35% gross margin. So let me recap. This is SolarEdge's next chapter. Revenue of $2.4 billion, margin of 35% in 2029. Accelerating growth in our evolving core markets, where they're evolving in a direction that is good for us. Unlocking a multi-billion dollars opportunity with AI factories based on this expertise that we've developed over the last two decades. And doing all of that while expanding our gross margins. So, when you put it all together, I hope that you see why we're excited about what's ahead of us, and we promise to keep you updated as we make progress. With that, I would like to introduce you to Lironie Nav, our Director of Residential Marketing. Throughout my career, I haven't met a person who is more knowledgeable and passionate about their product than Liron. Nexis was brought to the market by many, many people. But today you will have the privilege of hearing about its wonders from Liron. Come on, Liron.

Lironi Nav, Other

Thank you, Shuki. Good morning, everyone, and thank you for being with us today. For the past three years, I've been breathing and living the product that you see right here. This is SolarEdge Nexus, our next-generation residential energy system, and the platform we're counting on to deliver the residential growth Shuki just mentioned. Today, I'll walk you through why we build it, how we build it, what makes it so special, and why we believe it's the right platform exactly in the right time for this market. it. But before we talk about where we're going, let's start with where we stand. As of Q2, 2026, SolarEdge powers more than 4 million homes worldwide. One in three homes that has solar on the roof is powered by SolarEdge technology in the US. And we're the number one MLPE, model level power electronics provider by inverter shipments in the world. But why am I telling you this? Because you as an investor, you know, it's not just me bragging about our past success. This is our future. This is our install base. These are our channel relationships, and both of which are built-in distribution engines, just for everything that I will walk you through. Shortly, you will also see how we plan to turn this base into revenue. But let's be honest because the market underneath that install base is changing and is changing dramatically. I will walk you through three main changes. First, the value proposition is moving from pure energy export credits to market-based ROI, as we know that net metering and feeding tariffs are fading out market after market. Second, the home energy systems themselves are becoming more and more complex, more EVs, batteries, smart loads, backup requirements, it means that a single inverter or a single purpose mindset is simply not longer enough. And third, the go-to market itself is changing and you see it here in the U.S. from homeowner led cash purchased business model towards third-party ownership and installer driven fleets. While some of you can see these transformation as challenges to the market? We actually see these as opportunities, pure opportunities, because Nexus was built exactly for that. And when I just, what I showed you, it's not just theory, because when we see it with our three main core customer groups, we hear it loud and clear. For our homeowners, this is personal, because without any guaranteed export rate to fall back on, the value now comes for them from using, storing, and timing their energy very well. So they're looking for energy predictability. They want to know how much they pay in the end of the month, how much they earn. They want to maximize the return on each kilowatt hours. And of course, like we all do, they want it automatically with AI in their smartphones. Then our installers, they keep telling us one thing. We want more jobs per day. Less complexity, less O&M, easier upselling. The TPOs, they think fleet-wide. They don't carry a home by home alone. They want to maximize their returns across every system and they want it for years, for decades. They want fewer track rolls, full portfolio visibility and eventually extra revenue from virtual power plants. Nexus is built to satisfy all three audiences and when a single product serves all audiences that's exactly what turns into real share game and better economics per site. Meeting those shifts isn't just good for our customers it's a major opportunity for SolarEdge as well. Everything I just walked you through is going to turn into real measurable upside and it's coming from two main directions. The first one, our existing install base, as I said and as Shuki mentioned, SolarEdge powers more than 4 million homes globally. Only 10% of them has storage. It means that 90% are PV only systems with battery potential and we estimate that as dozens of gigawatt hours of battery potential and we don't have to win a single new customer they are already our customers and even when we look at net new customer when we sell PV plus storage system from day one we actually double out more than double our revenue per site it means that with the same sales effort it's worth more than twice that much to us so I think we're already now let's meet Nexus. So here it is SolarEdge Nexus. We just launched it but it's already turning heads. Forbes named it as shaping the home of the future of home energy and it won the 2026 green builder award for sustainable product of the year. But the number we are coming back to is one extra seven thousand dollars of savings per home throughout the system lifetime compared with our leading competitor. Now that's the number that wins the homeowner and a product that wins a homeowner by value is exactly what protects our pricing and of course our margins. And if you take only three words from my part today, make it be these. Powerful, flexible, and durable. Powerful means a real leap in energy and savings. That's what wins the homeowner. Flexible means installer can move faster and earn more on every visit. That's what swings the channel. And durable means a system keeps performing and keeps earning for years from now. And that's what swings the TPOs. Together they make this the strongest platform we've ever built. Now let's start with Powerful. So Powerful breaks down into three concrete validated numbers and I want to walk you through each one. First, harvest more. This is the story of our power optimizers. Ten and a half percent more energy per year validated by VDE renewables. Store more. Together with the value of the power optimizers, our homeowners can save up to $7,000 in lifetime savings versus leading competitor. And additionally, they also save more because more than additional $1,000 saving per year is done with our Sarah AI energy management companion. Now let's go a level deeper starting with how we harvest more energy. Let's talk topologies for a second because this is exactly where SolarEdge shines and separates itself from anyone else on the roof. Starting with string inverters, the old guard, simple, but unforgiving because one weak panel drags down the whole system. It means that one shaded panel and the whole string pays for it. Now micro inverters, they try to solve that going panel by panel, but eventually they still cap every panel output at the micro inverter's own power rating. So even on a perfect sunny day, you're still leaving power on the table or on the roof in that case. And SolarEdge, eventually, with our power optimizers technology, each panel produces its maximum power independently. So shading, mismatch, or just one weak panel never drags down the system. That's why SolarEdge roof simply produces more value than anyone else and in this case ten and a half percent more energy. Store more. So on the storage side we have a real advantage as well and I could just repeat that number again and again but you're here and I want to take you through the bits and bytes of our Nexus technology and this is exactly why we brought it. So first we will begin with new product requirements that are coming from the field. Because of the market trends that I mentioned earlier, the export payments that are drying up, and the homes, the energy system that are getting more and more complex, storage is now the king. And once you add storage to the system, it's not just working a few hours a day, it's working around the clock, charging and discharging as conditions change. So I want to ask you the question we've been asking ourselves when we designed Nexus. If the job now changed that much, shouldn't the way we measure performance and the way that we design products change as well? So to understand the meaning of this change, let me take you somewhere completely different for a second, the car industry. For most of the 20th century, it was very simple. When a new car launched, when a new car launched, all the commercials and the window stickers, all mentioned one thing. Peak performance. How fast the car gets to 60 miles. Top speed highway driving. But the world as we know has changed. Urbanization pulled people back into cities so suddenly nobody cares how fast the car gets to 60 because in real-world performance a car goes from zero to seven to five to 30 back again. And the solar industry is actually going through a similar transition now when solar is entering the picture, because for years every data sheet mentioned either peak efficiency or CEC, more focus on high power rather than the low power. It's a simple number, it's easy to sell, we did it as well, I did it all the time, but today it's simply not the picture anymore. This is exactly how it looked like before storage. You can see here in the curve, 90% of the time it really expands at a high power, only 10% of the time in low power. So that's exactly why our industry was very comfortable in measuring and marketing according to peak efficiency or CEC. But as I said, when we enter storage to the picture, you see here that 70% of the time we're actually in low power now, not just then. And 30% of the time alone we're in high power. Nexus, what we did with Nexus, is to optimize it for how systems actually operates today, not for the data sheet of yesterday. So what exactly did we do here? This is our unique power control design. I invite you, of course, all to see that afterwards. It optimized for silicon carbide-seek semiconductors, a cutting-edge architecture that cuts switching losses and heat compared to traditional silicon-based inverters. It's a big part of how we hold high efficiency across that full power range we just talked about and that efficiency is what turns into real dollars for our customers and a real competitive advantage for us. Thank you. I'm sorry I got a bit... I'm excited about the Nexus launch and that you see it here. We paired it also with an important innovation. From the creators of the power optimizer and with the same principle, we designed the battery optimizer. You can see it right here. What it does, it lets each battery block has its own brain. So one weak block never drags down the stack. So just think of it like at midnight when the home only consume around 200 watts because only the fridge and the Wi-Fi are working. The brain of each block knows exactly not to make all of them work. Only one of them can work in 200 watts and this is why we drag down the waste of the energy. Okay so there is no wasted energy and no weakest block that is dragging down the whole stack. It's the same philosophy that transformed solar harvesting now applied to storage. And after we reviewed the importance of high efficiency and low power that is in charge again on 70% of the time, let's put some real numbers behind it. This chart compares round-trip efficiency for Nexus against leading competitors, focusing on low power range from zero to two kilowatts. You can see Nexus has a meaningful advantage here. That exact gap is what enabling our homeowners to earn more. That advantage sits deep into our architecture. It's hard to copy and as Malz will show you later, it's a part of why nexus carries stronger margins than anyone else the third piece of powerful is Sarah our ai energy companion and frankly one of my favorite things we've ever built Sarah automatically optimizes how the home uses storage and export energy so the homeowners doesn't have to leave the finger and as you recall it was their number one requirement so Sarah brings value that worth over $1,000 every year but for us Sarah is also strategic because it's adding a software layer on top of the hardware so it deepens our relationships with every homeowner and it's the foundation of future services like virtual power plans that's powerful but a great system still has to be simple to sell and easy to install, which brings us to the second pillar. Flexible. Nexus is scalable by design, so no rewiring is required, and the system is field expandable at any time. That has two business implications. First, every service visit can become a potential upsell. Installer can add capacity without a full system rework. Second, the system grows with the home. More EV chargers, more loads, all on the same platform. Flexibility also translates into simpler operations. Nexus ships as single SKU covering 3.8 to 13 kilowatts. That's easier forecasting and easier inventory management and it's installed under 30 minutes, which means installers can fit more jobs per day every single day like they ask. Thus complexity pays off in better installer margins and of course stronger loyalty to our brand and to our platform. That was flexible. Now let's move to durable. Complicity only matters if the system is standing from years from now, which brings us to this third pillar, durable. And here I'll admit, I get to drag a little because we build Nexus for the extreme. It operates from negative 22 to 131 Fahrenheit degrees. It survives underwater up to a foot for up to 72 hours, sand and dust sealed out. But why does this matter so much for you as an investor? because the residential system has to perform for 15, 20, 25 years in somebody's backyard, not just in a lab. Every year it keeps performing. It brings higher lifetime value to our partners. That durability is backed by manufacturing with the highest Western quality and safety standards. And Nexus is manufactured here in the USA. So for our installers and TPOs, that means resilience, confidence in long-term reliability, which matters enormously the compliance for large-scale fleet deployments. So to bring these three pillars together, powerful, DC-native optimized architecture with intelligent management. Flexible, it's a future-ready scalable solution adapted to any home. and durable, built from durable materials, built in safety and cyber security protections. And we bring this all to market in the USA and in Europe with two different approaches. In the US, we're winning the shift towards third-party ownership, and Nexus is built just for TPO economics, as we mentioned, with faster deployment, better lifetime value, and strong adoption. We're proud to share, as Shuki mentioned before, that we've signed a few billion dollars deals with a few leading TPOs in the U.S. In Europe, we're scaling through thousands of neighborhood installers, oftentimes mom-and-pop shops, and we offer them a full suite of B2B2C and B2C solutions designed to help them exactly to win and to market in their own region, in their own neighborhood. We do that with the help of our DC bus. This is, by the way, an actual bus. I drove it myself a few weeks ago. It travels right now all over Europe doing roadshows with the Nexus technology inside, exactly to help them to get to everywhere in Europe. Two different markets, two custom playbooks, but one platform unified underneath. I would like to close on why Nexus is the win for every stakeholder in the value chain, all at once. For the homeowners, more usable energy, backup and savings, easy management and sleek design. For our installers, increased capacity, more jobs per day, less complexity, O&M and, of course, easier upselling. For our TPOs, improved economics with maximum system ROI, reduced track rolls, portfolio visibility and control, and VPP monetization potential. And last but not least for us, for SolarEdge, we have an expanded revenue opportunity here per site. It's a modular platform with more attached opportunities, data-driven install base and improved margin potential for us. That alignment where every participant in the value chain in the ecosystem wins together is exactly why we believe SolarEdge Nexus is the right platform to power our next phase of residential growth. Thank you. And now let's hear the voice of one of our enterprise customers, Alta Yen from Prologis.

Michelle Spina, Other

Alta, thank you so much for being with us today. We're honored to have you. Thank you for inviting me to be here. Yes, the customer perspective is so important. First, I really want the group here to know about Prologis. Prologis is described as the world's most valuable logistics platform. Can you tell us a little bit about the company? Sure.

Alta Yen, Other

Palatis is a global leader in logistics real estate. We have about 1.3 billion square feet across 5,900 buildings in 20 different countries. We serve 6,500 customers, and we have approximately $240 billion under management. Our facilities tend to be located in commerce centers, And so we have about $3.2 trillion of goods that flow through or touch Prologis Warehouse. And just to put that into context, that's about 3% of the world's GDP. What we're increasingly seeing from our customers is demand and asks around power. Power has just been a theme. Access to power, power affordability, power reliability. And so the role of Prologis Energy Solutions is to deliver power to our customers. And so that can be either through solar or battery, EV charging, or on-site power.

Michelle Spina, Other

Great. I mean, so glad that you really dove into the span of Prologis. I mean, huge real estate footprint. But you mentioned power. You know, it's a resource that is increasingly constrained. And so why is renewable energy such a priority for Prologis and your customers?

Alta Yen, Other

So renewable energy allows us to use what would otherwise be underutilized rooftops to deliver energy infrastructure to our assets. This allows us an opportunity to deliver incremental value to our buildings, to our customers, and to, in some places, even for communities, to deliver incremental value there as well.

Michelle Spina, Other

Okay. And really the process that Prologis goes through is kind of a well-oiled machine. You're really building the same roof hundreds of times. What earns a vendor the volume across your fleet? What disqualifies one?

Alta Yen, Other

So at our scale, we're really looking for three attributes from our vendors. We're looking for safety, we're looking for scalability, and we're really looking for partnership. Safety is quintessential to the Pelagas philosophy, as you can imagine. Building live solar systems on operating warehouses is really, safety is really critical for us. And we really value SolarEdge's technology and safety features, especially around the module-level architecture that you all have, the ArcFlash detection, the module-level monitoring. it really allows us to protect our buildings, our customers, first responders, as well as other service providers. On a scalability side, we have developed a repeatable platform at Prologis, but every project still has its unique attributes, whether it's the EPC contractor or the utility territory or any kind of regulatory constraints. And so from our vendors, what we want to see is a high level of service at every project, but then also across our entire portfolio. We want to see consistency of service, whether that's a project that's in construction or a project that's in operations. And then finally, we're looking at partnerships. We don't want vendors who just look at every interaction as a single transaction. We believe that the real value from our solar assets is over the 20-plus year of useful life. and so we really want vendors who have that like-minded mindset and will work with us.

Michelle Spina, Other

Great. And SolarEdge and Prologis have been partners for years now. You mentioned safety was a big priority. How else has SolarEdge supported Prologis kind of in that relationship piece across that large distributed portfolio we have with you?

Alta Yen, Other

So Prologis and SolarEdge have had a relationship for at least six years. I've only been at Colleges for two, but I will say that I really appreciate the collaboration between our teams, whether it's on the engineering side or the construction side or all the way through operations. SolarEdge continues to be a great partner, one that we can bring ideas with and have just very candid conversations. What I will say is that it's very helpful to have a partner like SolarEdge in discussing all of these different issues because during a six-year relationship as one can imagine there will be unexpected challenges and difficulties and really you find the value of a partner when you're trying to solve these types of live problems and I for one worked personally with SolarEdge in a situation where we had we're trying to develop a procurement strategy that had some scheduling challenges and then also some internal business goals that we needed to meet and i very much appreciated solar edge just bringing a cross-functional team who included engineering and legal and commercial and operations to work with us to really solve that and then i would say finally what's been really helpful with solar edge is just your continued innovation and your proactive nature in terms of bringing forth ideas around how prologis can continue to improve and optimize our energy solutions business great yeah prologis passed a huge milestone in 2025 one gigawatt of solar plus storage on rooftops across the entire fleet so what's the path to scaling further kind of what's next so it was important for us to have that one gigawatt milestone it allowed us to demonstrate that we can build distributed energy at scale but we didn't stop there, so we're currently at 1.4 gigawatts right now, and we will continue to push and execute on our current pipeline. We're continuing to develop in markets where we have some tailwinds behind us, and then we're trying to continue to open up new markets. I would say additionally, we're also looking at other technologies, so pairing solar and battery together, and working across new geographies. But we're really optimistic about what we see ahead of us.

Michelle Spina, Other

Yeah, us too, together. Thank you. I just wanted to get a little note of a real customer experience. Thank you for sharing all of that with us today and being here with us and your continued partnership as well. All right. I will invite all of you for a quick 15-minute break and we'll come back here and reconvene and talk about AI factories. Thank you.

Shuki Nir, CEO

I hope you enjoyed the short break and some time to rest. The first half of the presentations, we focused on our core markets. Thank you to Liron, thank you to Alta for sharing their perspective with you. Now we're going to move to the second and exciting multi-billion dollar opportunity that we have, unlocking the AI factories opportunity, I'd like to warn you, this next session is a bit more technical. But what we figured is that with all the noise that people are talking about it, it will help you and other investors, we believe, decipher between signal and noise. So Mayer will take you through the details and the technology behind our solution and why we believe that's the right way to transition to 800 volt DC. So without further ado, one of our co-founders, May Radist.

Meir Adest, Other

Hi there, everybody. I hope you enjoyed the coffee break. As Shouki said, it might get a little technical, so I hope you're well caffeinated, and I'll try not to make this too painful. For the next 25 minutes, I want to talk about what I believe is the most exciting opportunity in front of this company, powering AI factories with our solid-state transformer and the DC native architecture we built around it. We're going to talk about why the industry needs it, why it's hard, why we're the ones to do it, and where we're standing today. And a note on why this is personal to me. 20 years ago five of us founded SolarEdge on a contrarian bet that the future of power lies with intelligent DC electronics. I've been living DC ever since then. It's physics, it's dangers, its economics. So when I look at AI factories, I'm not seeing a new market. I'm seeing the same bet at a hundred times the stakes. Let's start with the one thing you need to understand about AI infrastructure today. Demand is enormous and it's growing. But the bounding factor is no longer chips, it's energy. Because every extra watt you could get from the grid is an extra watt that goes to GPUs. More watts mean more tokens and more tokens mean more revenue. So don't think of this as a power problem, but as a production problem. In an AI factory, every watt is a token, and every token is revenue. Shuki, in his presentation, showed you the dollar figures. Keep those in mind. We'll be coming back to them later. Here's my argument in three lines. It's the structure of this whole section. First, DC native architecture is the key to the AI power bottleneck. NVIDIA, the Open Compute Project, OCP, and the hyperscalers all agree on that. Second, getting there means solving two hard physics problems, medium voltage isolation, where the solid-state transformer replaces the old transformer, and DC safety, because DC behaves nothing like AC. And third, SolarEdge is positioned to lead this transition, with 20 years of experience of DC at scale and a multi-year head start in medium-voltage electronics development. And I won't just claim it. Towards the end of the show, I'll show you the... Towards the end of the presentation, I'll show you the system running. So, quickly, the numbers. Global data center demand... Global data center capacity is projected to double by 2030, with 100 gigawatts being added to the grid, led mostly by the Americas, but growing everywhere. The scarcest resource in AI is no longer silicon. its electricity. And the industry knows it. 92% of data center operators cite utility capacity or transmission constraints as their most significant obstacles. And you could see it also in the chart. Starting by late 2024, the long-term ambition is far exceeding the grid connections. The demand is there. The grid connections aren't keeping up. What's clear is that however you get your megawatts, you cannot afford to waste a single one. Now, The second thing happening at the same time is that the power architecture inside these facilities is changing fundamentally. NVIDIA has realized that the current design doesn't scale and announced a new 800-volt DC ecosystem for power-dense, efficient, scalable AI factories. This is a once-in-a-generation shift. To see why this shift is fundamental and maybe even inevitable, we need to go back to when the grid was born. It's 1880. Thomas Edison has just perfected a practical incandescent bulb and suddenly everyone wants electric lights. But there's a problem. A bulb works at low voltage and low voltage power can travel. Try sending it down a half a mile of copper and most of it is burnt off as heat. Long distance demands high voltage. The home demands low voltage and it's 1880. There's no power electronics, not one semiconductor on earth. So an ugly fight started about how to build the grid. In the left corner, Thomas Edison pushing for DC, but because low-voltage DC can't travel, that means you need a power plant in every neighborhood. And in the right corner, Nikola Tesla pushing for AC, because only AC could change voltages with a transformer, two coils with an iron core and no moving parts. The war of the currents, as it was later known, got ugly, but eventually physics and AC won. Now, here's the takeaway. AC didn't win because AC is better. AC1, because the transformer was the only voltage-changing machine on the planet. Every grid in the world today is downstream of that one missing technology. Now, fast forward 140 years and look at what we actually plug into the grid today. Lights, LEDs, DC. Phones, computers, chips, everything that computes, DC. Industrial motors behind their variable speed drives, DC. EVs, solar, wind, every battery ever manufactured, all DC. And AI data centers, the most power-hungry loads humanity has ever built, pure DC all the way down to the transistor. We live in a DC world, served by a 140-year-old AC delivery system. The old justification for AC, no power electronics, has evaporated decades ago. We now have the semiconductors that Edison was missing. But the grid never caught up, because nothing ever forced the issue. AI factories are the load that actually forces it. So let's count what the old system cost. 34.5 kilovolts comes in from the grid and runs a gauntlet. Transformer, low-voltage distribution, double-conversion UPS, low-voltage AC power supply, 12% of the power lost before it reaches a single GPU. You'll notice that inside the rack there's another 2% to 4% lost, But since we're focusing on what's outside and comparing apples to apples, we'll focus on this 12% number. On a gigawatt campus, 12% is 120 megawatts, an entire data center's worth of power burned off as heat before it touches a single GPU. Now, this is a sneak preview of what you'll see later. SolarEdge's integrated DC-native solution end-to-end. Same utility connection on the left, same GPUs on the right. but one difference. Instead of 12% lost, there's less than 1% lost. One conversion instead of five. So remember Shukri's arithmetic. At a 100-megawatt AI factory, every 1% of power is approximately $20 million per year in compute revenue. So closing this gap from 12% to 1% means efficiency is no longer a footnote on a spec sheet. It's a line on the income statement. Hold this picture. The rest of my talk is what it takes to make it real and why we believe almost no one else can do it. So how does the industry capture this prize? Think of it as a maturity ladder for data center power, differentiated by system efficiency. At the bottom is level zero, legacy AC architecture, basically every data center today. Traditional transformer, low-voltage distribution, double conversion UPS, PDU, etc., going to the chips. As you can see, 84% to 91% of the power actually reaches the chips. Above that is level one. Basically, you keep the white space with the same AC that you have today, and you just bolt on a side card, a side card which converts from AC to DC so that you could power the 800-volt next-generation chips. You're able to turn them on, but you're not really gaining any power. In between are the compromises. Level 2 and Level 3 get you to the mid-90s, still dragging around a long legacy hardware. By the way, if you want to deep dive into this, we have a white paper on this. And at the top is Level 4, the integrated DC-native solution with a high-efficiency solid-state transformer, an intelligent power distribution unit, a DCUPS, which is connected in parallel so you're not losing any power, converting 800 volts directly to the chip. 98% to 99% efficiency end-to-end. Most of the market is stuck at level 1, band-aids on AC. The market is climbing this ladder one rung at a time. SolarEdge is going straight to the top, straight to level 4. So why hasn't everyone done that? Because between level 1 and level 4 are two walls of physics. Bear with me for a minute of science. This is the heart of the story. The grid is layered. Transmission runs at hundreds of kilovolts, and as it enters an area, it steps down to medium voltage. 10 to 20 kilovolts for the smaller loads, such as edge data centers, and 34.5 kilovolts for the large loads, such as AI factories. And the higher the voltage, the harder it is on power electronics. And this enemy has a name, partial discharge. As you can see in the middle pane, every insulator could have tiny microscopic voids, and under high electric stress these voids could spark tiny lightning bolts inside the insulator eating away at them. One of these sparks is not a big deal but billions of billions of them over the years cause erosion and treeing and degradation and eventual failure of the equipment and the stress is voltage dependent, manageable at 10 kilovolt but brutal at 34.5. Building a transformer that stays quiet at this level while switching thousands of times per second is one of the hardest problems in power engineering. So here are the two barriers. Barrier one, what you've just seen, direct conversion from medium voltage to 800 volts DC with a single stage and full galvanic isolation at more than 99% efficiency. Everything you see here is what makes it hard, and you've already met the hardest enemy, partial discharge. And barrier two is 800 volts DC because DC never crosses zero. When there's an arc in AC, it extinguishes itself 100 times a second every time the AC signal goes through zero. The DC arcs keep burning. So protection and detection and control and safety have to be designed from the ground up for DC. Safety by design, not by retrofit. This is completely new to the data center industry. Two problems, both hard, both physics, not software. And here's why we believe SolarEdge is best suited to solve them. Three proof points. First, proven solid-state transformer medium voltage isolation. Remember those tiny lightning bolts I mentioned before? We tested our solid-state transformer at 150% the rated voltage and measured partial discharge of less than 1 picocoulomb. That's less than 1 trillionth of a coulomb. That's 10 times better than what the standard requires. That difference is the difference between a lab demo and a product that survives for years in the field. Second, 20 years of DC architecture and safety. With 150 million units in the field, it has been our core business since day one. And third, we have the scale to execute. With 65 gigawatts shipped and global R&D, manufacturing and operations, we're not a startup with a slide, we're an industrial company with a running machine. The two hardest problems in AI are medium voltage isolation and DC safety. We've spent 20 years of training exactly for these two problems. So let's open the hood and look at how SolarEdge is powering the next phase of the AI revolution. Let me bring back the picture we started with. Five conversions, each taking its cut, 12% of the energy gone before a first GPU sees a single watt. That's the machine the industry is running today. And here, the same first mile rebuilt, one stage, 34.5 all the way to 800 volts into the rack with less than 1% lost. Five conversions become one. 12% becomes less than one. What we deliver is one unified solution all the way from the utility connection to the compute rack with three ones. one conversion where the AC chain needed five, one protection scheme all the way from the SST output to every individual rack feed with no coordination gap because there's one vendor and one control loop. The source, the SST and the buffer share a bus but both of them are controlled by us so stability of this bus is engineered in not integrated afterwards and hoped for and there's an important degree of freedom here everything you see could be located either inside the data center in the white space or in the gray space or outdoors based on the customer preference so they have a degree of freedom and it's not dictated by us as a supplier the solution is built from three purpose-built engines the SST the power distribution unit and the DC UPS and you can see all three of them all three of them come together here in a second this This is one architecture that scales all the way from a single hall to a gigawatt campus without requiring redesign. Here's how. Here's the SST. The medium voltage comes in three phases, and each of them has up to a dozen cells in series to stand off the medium voltage. The output of these cells is connected in parallel to provide the power for the 800 volt bus. In addition to that, you have the DCUPS connected to the bus in parallel, so there's no series losses. And this DCUPS is storage agnostic. We supply the conversion and the interface, not the battery cells themselves. So our customers are never locked into one vendor or one battery technology. From the bus, there are separate channels going to each rack, each one providing its own protection and isolation. And notice one detail. The galvanic isolation, the red part down there, is built inside of each cell's DC to DC stage, so any fault energy from the medium voltage physically can't reach the bus. The isolation isn't a component we added, it's built into the architecture. One design for any size. This isn't a concept. A few weeks ago, I took off the jacket, put on a polo shirt, and went into our high-voltage lab to show you the system running. Today, between the utility lines outside and the DC racks inside an AI factory, power is converted up to five times. Each of these conversions adds losses, a significant percent of the energy wasted. Our solid-state transformer is built on semiconductors instead of iron and collapses all of those conversions into one stage. 34.5 kilovolt in, 800 volt DC out at over 99% efficiency. Everything you see here is working today. station one a single cell under power behind me is one cell the AC grid comes in from the left goes through the transformer and back and comes out from the right as the DC that the servers run on those three parts are the whole cell it converts the power and it isolates between the high AC voltage and the DC side the isolation is built into each cell we are going to raise the power now. As you can see, the efficiency is above 99% and it stays that way across different power ranges. Station 2. Six cells with one shared output. The same cells you've seen in station 1. AC three-phase medium voltage comes in from the top and all six convert one shared DC output. Station 3. Safety. Before equipment like this goes anywhere near a data center, insulation must be proven. Here we test partial discharge. The tiny electrical leaks would show up way before equipment actually fails. The ground comes off and the test starts. At a 34.5 kilovolt grid, the voltage to ground is 20 kilovolt. We're going to 30. That's 150% stress. And as you see, the leak stays under 1 picocoulomb. That's 10 times better than the standard requires. At this level, the insulation is not wearing out. That measurement is what determines if the technology could be deployed at all, and we are much better than the limits. What you've seen today is working hardware at medium voltage. We're building it up toward a 5 megawatt unit, and the efficiency isn't a projection. You've seen it on the meter. 20 years of power optimization aimed at an industry which values it most. That's real hardware converting real medium voltage today. And we're not doing it alone. Earlier today, we've released a white paper about protection and grounding for 800 volt DC for data centers. NVIDIA contributed to this white paper as an ecosystem partner and a technical reviewer. We didn't wait for the 800 volt DC safety standard to be handed to us. We're writing the playbook, and NVIDIA engineers reviewed it. And the partnerships go deeper than paper. Yesterday, we announced an extension of our collaboration with Infineon, solid-state circuit breakers for high-voltage DC distribution. Remember barrier two about DC safety? If you try extinguishing an arc with a mechanical circuit breaker, quenching the arc would take milliseconds. At the power densities of AI factories, that's near eternity. With solid-state circuit breakers, we could do it in microseconds. We're leading the design with the Infineon silicon carbide devices at the core. Protection is what unlocks 800 volt DC at scale. And together with Infineon, we're building this solution. So, where are we on this journey? Through 2024, we've been doing scientific validation and medium voltage transformer testing, basically the deep physics work. As you've just seen, we have a demo currently working in our lab, and expect to have a working product by the end of 2026, leading to pilot installations in 2027 and commercial deployment in 2028, together with availability of the 800-volt DC racks. We're not aligned to a paper roadmap. We're aligned to NVIDIA's rollout. So when the next generation racks arrive, we aim to be there powering them. Maoz, in a few minutes, will show you what this ramp means in dollars as he takes you through the model. And when I say customers, were engaged across the entire AI infrastructure ecosystem, starting with the hyperscalers who are building the AI infrastructure for their own platforms, through the neoclouds, the GPU providers who are writing the standards, and the facility operators and energy providers with whom we're having strategic partnerships and joint system offerings. Basically, everyone who touches the power path of an AI factory. Before I end, five takeaways. If you remember nothing else, please remember these. First, the world is moving towards 800 volts DC. This is a once-in-a-generation shift. Efficiency is tokens, and tokens are revenue. To quote back, Shuki is a 100-megawatt AI factory example. Every additional 1% of power is $20 million per year in compute revenue with an NPV of $100 million. dollars. Not only 10 or 20 kilovolt, but the full 34.5 kilovolt directly to 800 volt DC at over 99 percent efficiency demonstrated running. 20 years of distributed DC architecture expertise. With 150 million units in the field and 65 gigawatts shipped, DC is not new to us, it is us. And And finally, as you've just seen, we're not promising. We're executing. And now, what I'm sure you've all been waiting for, the numbers. Everything you've seen today, starting with Shuki's strategy, the nexus, the SST, it all comes together in one financial model. And the man who built it will take you through it. Please welcome our CFO, Shuk Maoz Zygrun.

Maoz Sigron, CFO

Thank you, Mayo. Good morning, everybody here in the room and also online. I joined SolarEdge a few months ago, and I'm very excited to be here today. SolarEdge is an organization aligned with one main goal, translating and leveraging technology and innovation into a sustainable, profitable growth in the coming years. Today, I would like to discuss with you our model for the coming years. I will discuss the assumptions, the revenue drivers, the margin expansion drivers, the operating leverage that we have in the model. Our discussion today will split to two sections. The first one is the turnaround, what we did so far in the last two years. And second is the profitable growth plan for the coming years. Let me start with one slide that, in my view, gives a good picture about the two sections that we're going to have today. First is the turnaround, where we are today, and second is the profitable growth, where we are heading. The turnaround, the main objective, was to first sustain the business and translate this business into a profitable business. In 2024 and 2025 were years of first, restoring discipline. Second, extending the margin. And last, generating cash. In 2026, we returned the business to profitability. As we shared in the last earning call, the second quarter, we had a positive operating profit for the first time for 2023. This is the foundation for the model that I'm going to share with you in the second section on the profitable growth for 2027 to 2029. Three numbers to remember from this slide. First is the $2.4 billion in 2029, reflecting 23% KGEL for the three years. Second is the SST revenue that we're expecting for 2029 with $600 million. And the last is the EBITDA margin that we're expecting in 2029 with 18%. Significant improvement from where we are today. Let's start with the turnaround. During 2023 and 2024, the solar industry went through a significant correction. Our response was to simplify the portfolio, reduce cost and complexity, and to refocus on profitable market and products. Three years with three lines. First is the revenue that increased significantly from the $900 million in 2024 to almost $1.3 billion in 2026. The gross margin improved significantly from minus 90% in 2024 to 28% in 2026. 24% excluded GIPA. This is important for our discussion later. The last one is the EBITDA. moving from a negative EBITDA in 2024 and 2025 to a positive EBITDA that we're expecting for the full year in 2026 with a 2%. I'm going to share with you today our model with adjusted EBITDA rather than EBIT. We believe that the adjusted EBITDA better reflect the underlying of our business, and this is the way we're going to report moving forward our results. Let's start with the gross margin. The gross margin increased significantly from 8% in Q1 2025 to 24% in 2026 for the full year. As I said before, this is excluding the EPA impact. This is 16% improvement in two years. And this is the foundation of our model, so I will discuss now the main drivers, and then I will talk about the margin improvement that we're expecting also in the model. The first effort is the refocus on profitable countries and businesses. Part of what we did in the last two years is exited non-profitable countries and businesses. And we are focused and we are allocating today most of our investment into these countries and businesses. The second one is the single SKU. Part of what we did as part of the operational excellence efforts is making sure that we are running and moving to a single SKU. Single SKU, simplified manufacturing, forecasting, operation, logistic, and support. It saves real dollars. The next one is the offshore. Part of what we did in the last two years is moving operation and support function into offshore countries. It's helped us to improve our margin over the last two years. The next one is the business skills. Part of our cost of goods includes fixed expenses. The fixed expenses can absorb more revenue. And part of what is happening from moving from the $900 million that we had in 2024 to the $1.3 billion in 2026 is the scale of the business. And last but not least is the U.S. manufacturer that we have in the U.S. 90% of our production of optimizer and inverter is happening now in the U.S. This is good for SolarEdge from the benefit of the 45X, but also important for the domestic content from our client for our client so this is good for margin but also important for our revenue let's move one line down to the cost discipline the operating expenses decreased from four hundred forty seven million dollar in 2024 to three hundred sixty seven million dollar in 2026 this is a non-gap numbers moving from 50% operating expenses to revenue in 2024 to 28% operating expenses to revenue in 2026. This is a significant improvement. This is $80 million reduction in our operating expenses while the revenue is actually growing over the last three years. The main efforts that we have here in the operating expenses are the following. First, as I said before about the gross margin, the same is right for the operating expenses. The focus on the core market and core businesses is translating also here into the operating expenses. We are allocating our investment to where we are making money. The next line is the offshore. As we did in the cost of goods, we are doing the same with the R&D, S&M, and G&A. We have a global footprint, and we are allocating resources, in some cases, to lower cost countries. And the last one is the headcount. Part of what we did in the last two years is aligning the ad count to the current level of the business. Moving here with the ad count to the current level of the business while preserving our ability to increase and to scale the business in the coming years. Of course, discipline is not a one-time effort. This is ongoing efforts that are also going to be part of what I am going to share with you in the model. We are going to include AI and automation as part of our efforts to improve our operating expenses in the coming years. Let's move to the balance sheet and the free cash flow. The net cash increased from $82 million at the end of 2024 to $265 million at the end of the second quarter. Important to say that the cash level at the end of the second quarter achieves more than $600 million cash, cash equivalent, and marketable securities. Another improvement here that is significant is the free cash flow. Improving from minus $422 million in 2024 to $24 million in H1-2026. We are expecting the year to be with a positive free cash flow. The main efforts around the balance sheet and the free cash flow include the following. First is the working capital discipline. This is a very important effort that we are running. It includes, first, the inventory. We are making sure that the inventory level is aligned to our needs. We are making sure that sales forecast and operation are aligned and translating to our footprint. Second is the payment terms with our customer and vendor, making sure that there is alignment between them. The next one is the cash conversion cycle. At the end of the day, this includes the DIO, DSO, and DPO. Every day that we are improving, improving our ability to support our future growth for the coming years. In September 2025, we return our convertible notes on the maturity date with our cash, without using any external financing. The balance sheet that we have today is a strong balance sheet that will support our future plan in the coming years. Which leads me to the next section that we are going to discuss today is the profitable growth plan for 2027 and 2029. Before we are getting into the numbers, I would like to share with you the assumptions for the model and the process that we run internally. In the last three months, we run the process globally with the U.S. team, the EU, and the international market team, with sales and operations, with the Resi, CNI, and the SST team to make sure that we all align with the consolidated process that I'm going to share with you today. So the first thing that I would say is that this model is a bottom-up model country by country with the market share assumptions that we have in the model. The second is that we are using our 20 years of data in SolarEdge, making sure that we know where we are today, what was in the last three years, and where we are ready and what is the potential of the business for the coming years. We are using third-party data from Wood McKenzie, Home Analytics, and Solar Power Europe, and others to make sure that our assumptions for the coming years are aligned with the market analysis. After we have the market share per country, We are translating this to a product mix, taking the assumption of the price, and making sure that the market share is also reflecting the footprint that we're expecting for the coming years. The SST model built separately, based on our discussion with potential customers and based on the third-party data that are important for our assumption that we have in the model. The IRA, as I said before, the 45X, is another important element in our P&L. So the assumption of the IRA is part of the product mix that we are expecting and based on what we're expecting to have here in the U.S. from the U.S. made point of view. The indirect Cox expenses are based on our revenue level for the coming years. Part of them are variable, but part of them are fixed, and we're expecting a moderate growth in the coming years, not on the same level as the revenue. One line down to the OPEX, the same. We're expecting to see leverage. important to say about the R&D as our plan include innovation we are going to invest and keep an investment in the R&D and be expecting that 50% of our operating expenses will go to the R&D in the coming years so let's move to the model so first I would like to talk about the first line the revenue moving from 1.3 billion dollar in 2026 almost there to 2.4 billion dollar in 2024 in 2029, sorry, 23% Kegel. The next line that is important is the gross margin, expanding from 24% without the IPA impact to 35%, 7% with the IPA, 11% without the IPA margin expansion in the coming years. Let's go to the OPEX, moving from 28% operating expenses to revenue today to 20% in 2029. 8% improvement. Last, but not least, is EBITDA, moving from 2% today to 18% in 2029. In all P&L KPIs, we are doing better and we are expecting to improve our position. With that, I would like to dive into our growth drivers that we have in the model. So, four main drivers that I would like to discuss and dive into. First is the SolarEdge Nexus. As shared before by Liron, we are expecting to see the Nexus in the coming years. The move from PV only to PV plus storage is exactly aligned with the way our design Nexus. As shared before, we are expecting to see more revenue per site with the Nexus, more than two times. Additional, we're expecting to see better ROI with the Nexus, with a 10.5% more energy with our optimizer per year, with the $7,000 that mentioned before, that we're expecting to see on the life of the system, and the $1,000 additional benefit per year with our Sera, with the AI application, we're expecting the ROI to be much better also compared to our competitors. During the second quarter, we did a $60 million with the Nexus, and we're expecting more. Our assumption here in the model that Nexus is going to be a significant contributor to our growth for the coming years. The next one is the CNI. We are winning on the CNI on product, but also on our regulatory positioning. We are FIO compliant, FCC authorized, domestic content authorized, and we are U.S. made. We doubled our U.S. rooftop market share in the last few years, moving from 29% in Q4 2024 to 57% in the second quarter. This is a significant improvement. And part of the assumption here that we will keep and see more market share in Europe and in the U.S. with the CNI. Moving one line down to the safe harbor. Part of what we shared before, that before the July deadline, we signed a $1.7 billion contract with Resi and CNI customers. Together with the optimizer that is associated with the inverter and other products that are including the $1.7, we're getting to the $3.3 billion as shared before by Shuki. And if we are taking a conservative assumption of 30% batteries, we're getting to more than $5 billion. This is a significant amount. We're expecting this to translate to revenue between 2027 to 2030. As shared before, we have liquidity damages in the contract, but still some customer may cancel, some customer will go out of business, and this is part of what we are reflecting in the model. So we're expecting Safe Harbor to definitely contribute for the coming years and helping us to gain market in the coming years. The last one is the AI factories. With a $600 million revenue in 2029, we are here taking assumptions based on where we are today with the early stage. As Mayer shared before, in lab demonstration with prospective customers, the system achieved 99%. This is a significant achievement. Important to say that together with the AI and the core business, we are getting to the $2.4 billion. $1.8 billion is the business that we know and that we are running for 20 years. And another $600 million is the AI that is also using our knowledge from the last 20 years in the company. Let's talk about the SST for a minute. So our assumption for the SST is that in 2027 we will have an insignificant amount of revenue. It will be mainly pilot installation. In 2028, we are starting to see the ramp up from the nexus from commercial agreement, and the scale is expected in 2029 with $600 million. We are in early stage with the SST, so we are cautious about our expectation. but with our technology, with the lab demonstration and with the discussion that we have with potential customers and with our understanding with the RFI that safety and efficiency is the most important KPIs for our clients we feel that we are in a good position and we're seeing here a multi-billion dollar opportunity for the coming years With that, I would like to move to the gross margin in our model So the gross margin is expected to increase from 24% in 2026 to 35% in 2029. The main contributor to the 11% improvement are the following. First is the Nexus. As shared before by Leone, we're expecting to see more margin with the Nexus. This is not only more revenue. We're expecting to see more margin compared to the legacy margin that we have with the legacy product today. The next one is the AI factories. Our assumption in the model that the margin of the SST will be higher than our normal margin. So this is another contributor to the margin for the coming years. Operational excellence, as I said before, this is not a one-time effort. We are running different teams globally, making sure that we are running with different efforts to make sure that we are keeping and improving our leverage in the model and we're expecting to see more benefit from the operational excellence efforts in the company in the coming years. The next one is the business skills. Also here, we're expecting to see some leverage in the model as we have fixed costs that are expected to contribute to the margin as well. The last but not least is the 45X. Important to say that today we are not yet benefiting in full from the 45X. As we are getting into the next years, 27 and 28, we will move to full benefit, 100% benefit in our P&L. As we are ramping up the production and as we are ramping up our new products, the benefit of the 45X is not yet in full and we're expecting more benefit in the coming years. With that, I would like to move to the OPEX discipline that is also important in the model. So the operating expenses expected to increase from $367 million in 2026 to $480 million in 2029, a reduction from 28% to 20% in 2029. The main efforts in the operating expenses are the following. First is the operating leverage that we have in the model. As I shared before, we are increasing our revenue significantly and we are expecting a moderate growth in the operating expenses. This is the first driver. Second is the offshore. We built in the last two years a global offshore that can also contribute in R&D, S&M, and G&A, and we're expecting to benefit from that in the coming years. The next one is the R&D. As I said before, we're going to invest more than 50% from our operating expenses to support our future growth with the new product and technology. And last but not least is the automation. We are giving access to AI tools globally to all our employees. We believe that with automation and with AI, we can improve this line as well and have more benefit in the coming years from the AI. With that, I would like to conclude. So the turnaround returns SolarEdge to a profitability as I shared before in the second quarter. The coming years and the model expected to increase their revenue significantly while expanding the margin. together with our people, our technology our market position our operational discipline and our unique products we are expecting to gain market in the coming years and that will translate into sustainable, profitable growth in the coming years with that I would like to conclude my session so thank you and we will move to the Q&A so I would like to invite Shuki, Liron and Mayer So thank you for spending the time with us.

Shuki Nir, CEO

I hope that all these sessions were educating for you, that you learned something new today, which is very, very important. We'll now open it for questions, and then we'll wrap it up.

Alta Yen, Other

Just wait for the mic before.

Philip Shen, Analyst — ROTH Capital Partners

Thank you for hosting us. This is Philip Shen with Rothkabel Partners. Wanted to check in with you on the $600 million AI target for revenue in 2029 and the 35% margin. So specifically, can you share with us a little bit more detail on the assumptions that go into the $600 million? I think my quick math suggests you have 25% market share in 29. Is that the assumption? And then also, if you can, what's the rough price per watt? are we in the 30 35 cents a watt maybe 40 cents I think was on the slide earlier but are you assuming you know something in that range and then also on margins would you in the past we've talked about this Shuki you guys were still deciding if the 45 X could be claimed on the SST and so just curious if you guys have made a decision on whether or not 45 X will be assumed for the SSD. Thanks.

Shuki Nir, CEO

And that's your first question.

Maoz Sigron, CFO

I do have others. So thank you Phil. So first about the SSD. The model that we have here with the $600 million is the model that built based on our assumption. We need to separate between the time that we shared during the discussion to our model for 2029. We are first taking some assumptions inside the model that are more conservative. We are not going to share exactly what are the details behind the scene, behind the model, but let's say that the numbers that are out there in terms of, you know, as I said, we are using our conversation with customer, we are using third-party data, we are using all the available data that we can have in order to make sure that the model is reasonable and that is something that we can support. But this is very much aligned with the data that we have internally and the data that we have externally. So this is for the first question. About the margin, a question of the 45X for the SST. So the model does not include any assumption on 45X. So this is, you know, to be conservative and without having any question about where we are with the 45X. You know, we will keep and follow carefully. If there is a potential, we'll see how we're utilizing it. but the assumption here is without the 45X.

Philip Shen, Analyst — ROTH Capital Partners

Okay, great. Any thoughts on pricing or you can't share?

Maoz Sigron, CFO

We are not getting into this level of data. We gave the direction where we are heading, but we are not getting into the data.

Philip Shen, Analyst — ROTH Capital Partners

Okay, thanks. And my second one here, and then I'll pass it on. As it relates to, again, still AI and data center, you know, Meyer, I think you talked about a once-in-a-generational shift to 800 volts. One of our inside the data center EPCs, major one, Rosenden, you know, they've talked about potentially shifting to 1,500 volts. And there's a lot more standardization around that, a lot of volume in there. Could you see a shift to 1,500 volts pretty quickly? And if so, are you guys ready for it? and then how does that improve potentially your efficiency even more? Can you go from 99.2 to something even higher or closer to 100%?

Meir Adest, Other

So the short answer is yes. There's no problem to shift to 1,500 volts. The slightly longer answer is actually that's where we started. I was mentioning that we have a multi-year head start in developing medium-voltage electronics when NVIDIA just announced this 800 volt shift about a year ago and that's because we were working in the lab working on a direct basically utility scale inverter that's directly connected to the medium voltage so back then we were 1500 volts and actually what we've been doing is scaling down to 800 volts because this is what the ecosystem needs right now and we're hearing similar sentiment. I'm going to guess that the first few years will be 800 volts just because the whole DC infrastructure is new to the data center industry and it'll take some time for them to feel confident at 800 volts before moving on to higher voltages. But yeah, it definitely makes a lot of sense, not only because of the maturity of the components, but also just because you could push more if you have a higher voltage, you could push more power without increasing the current, which means without increasing the losses and You know, thermal issues are easier if you don't have to increase the current. So, yeah, I'm guessing the industry is going to start gravitating towards that. I'm guessing it's also going to take time, but we're there.

Mike Malte, Analyst — Deutsche Bank

Hey, Mike Melty with Deutsche Bank. Thank you for taking the questions, and thank you for the presentation. So my first question pertains to the 30% attach rate that you mentioned, representing up to $5 billion in additional revenues. Can you talk about how feasible you view that 30% possible upside or downside to that number and what the revenue opportunity could look like from that?

Shuki Nir, CEO

So when we talk about the touch rates of storage and safe harbor transactions are in the U.S., in U.S. resi, I think that one of the charts that Liron shared with you showed the touch rate that today we have in the U.S. And the average attach rate is above 30%. In some of the states, it's hovering above 90% as well. So that's the reason today we're above 30% attach rate. When we are modeling things out, we are trying to be more careful about the assumptions that we are taking. It's very easy to go to 50% and 60% and 70% attach rate. We feel that 30% attach rate is probably achievable. It's something that we can achieve with these transactions. and then that's the number that we're sharing with you today.

Mahib, Analyst

Hey, Mahib from Mizzou, and thanks for hosting me. A few questions on the SST revenues and business segments. First, on the $600 million for 2019, when could we expect the bookings visibility? What's the lead time on that? And on that delivery, is that for energization in 2019 or 2030? Like, should we look at the TAM in 2030 for that? And could you also, like, talk about the gross margin and EBITDA margin you expect for SST versus the core business? To understand how much of the 35% is from SST versus the rest.

Shuki Nir, CEO

So I'll talk about the revenue, and Marz will refer to your margin question. So when you look at the industry, and we were talking about exponential growth, and you could see that in the industry adoption. You can see that in the revenue evolution that we're sharing, and so forth. And the way that we look at it is the industry, the data center owners, the AI factory builders and everybody, they understand the value already. They understand the value, they understand the need as well. The rack density is going to be over one megawatt in 2028. You know, NVIDIA are not going to stop and this is just a calculation. So they understand the need, they understand the necessity to move to 800 volt DC, they understand the immense value, financial value, that they have in moving to DC Native Solutions, SST, and because of that we feel, and this is why we focused on safety as well here, we feel that there is one roadblock, if you will, in their mindset because safety is serious. They don't want to put as lucrative of a technology as it might be if it risks the entire building or the entire equipment that they have over there. I think that once we get over the safety hurdle, if you will, which is a big one, I think that then we will start seeing that the pilots are getting into higher gear and AI data centers are going to start placing orders, preparing for that, and so forth. As for lead times and everything, we haven't gotten to this point yet. Whether it's going to be before the AI factory is ready or not, it will be before. They need to place the power before they actually bring the equipment in. So it's for AI factories that are going to come online in 29 and 30, if you think about it this way.

Maoz Sigron, CFO

So about the margin, we want to be mindful with the way how we are building the SST model for 2028 and 2029 and including the margin assumption. I can say that the margin of the SST is better than the core business that we are running. This is our assumption based on all our discussion, and this is what we have in the model.

Mahib, Analyst

And maybe one follow-up, Ashuki, on the safety aspect. Something which we also learn is that SSCBs are pretty important for DC. Are you assuming some SSCB revenues in that $600 million or that package for the customers, or is it just plain SSG?

Shuki Nir, CEO

So, maybe I showed you the container. When we talk about SST, everybody assumes it's a small product. It's a 20-foot container, right? It's a 5-megawatt, 20-foot container that starts with the MV side and going all the way to the low-voltage side, including the protection over there. now some of the customers would like us to bring the entire container some customers may choose that we bring only the middle part or some other so it's kind of almost a configuration that will be dependent upon customer selection but if the customers that would like us to bring the entire container it will have the protection as part of it thank you yeah hi got a mic too here Dylan Asano, Wolf Research.

Dylan Asano, Analyst — Wolfe Research

Sorry, I just want to come at the assumptions behind the $600 million a little differently. I mean, should we think about that assumption as purely looking at the market and making a market share assumption? Are there indications of interest for orders already that maybe are contingent on the pilot and how that goes? just any more kind of color around how you're comfortable you are with that.

Shuki Nir, CEO

Yeah, so had we signed the firm agreement similar to the safe harbor contracts, we would have told you about them already. We have been in discussions with most, if not all, of the players in the ecosystem, whether these are neoclouds, the hyperscalers, and others. And in these conversations, we start with a conversation, then we have some technical discussions, Some of the customers have actually seen live demonstration of the system, and this is where you start seeing people kind of tilting towards, yeah, there are 20 companies that are doing it, to, oh my god, there is something that is unique here. This is why we keep saying that players in the ecosystem are telling us that we are ahead of competition in terms of timeline and in terms of the performance of our product. So we don't have any concrete PO or any concrete agreement with any specific customer, but we have the confidence, based on all the feedback that we are hearing, that these assumptions make sense.

Dylan Asano, Analyst — Wolfe Research

Great, thank you. Just a quick follow-up on the base business solar assumption for 2029. Could you give us maybe the split between CNI and Resi that's embedded in that assumption?

Shuki Nir, CEO

So it will be similar to today. And the main drivers, if you think about it, on the growth, are the ones that we mentioned, right? It's about the storage attack rate that is going to grow. In Resi, we are going to actually benefit from retrofit going into residential markets. And we will continue winning in CNI, as we said. So we expect both businesses to continue growing.

Colin Rush, Analyst — Oppenheimer

Thanks so much, guys. It's Colin Rush from Oppenheimer. Can we start with the channel strategy on go-to-market with the SSTs. Obviously, the data centers work with a number of existing vendors. You're looking at displacing a number of components in those designs. Can you talk a little bit about redundancy in the initial thought process of implementing these SSTs and if you'll go direct to the data center operators or be working with existing vendors that are currently designing a number of the infrastructure components for these folks? You want to take the redundancy first?

Shuki Nir, CEO

So, as you mentioned, and the go-to market to the, fortunately for us, the number of end customers is not unlimited. It's actually quite, one can count them on probably four ends. And between the new clouds and the hyperscalers, and as you mentioned, Colin, and we talked about it in the past as well, potentially there are two ways to go there. One is to go and sell directly to the hyperscalers or to the neoclouds, or go and partner with one of the power electronics companies that are already installing different equipment along the powertrain. And you named a few of them, and we know that there are several of them. At this stage, we're talking to everybody. And the reason for that is twofold. even if we chose to go through the channel as you call them the end customer would like to talk to us so that will not save anything and definitely if we want to go and sell directly to the data center owners we are going to work directly with them so the conversations are with everybody with different players in the ecosystem it allows us to be better it allows us to gain more feedback because end customers are seeing their own needs Horizontal players are seeing the industry needs and where the industry is heading and are more exposed to developments in other places. But in all of these conversations, we feel confident that the technology and the value that we bring is going to be very valuable to the end customers. And this is what matters at the end of the day, right? Whether you go here, here, or here, the end customer needs to appreciate what we are bringing, understand that the safety is something that we're going to take care of it and that we have the scale and reliability to support our business. I think that we have all the things that are needed. We haven't yet decided how to go to the market.

Colin Rush, Analyst — Oppenheimer

Okay, perfect. And then two questions on the supply chain. Just the evolving battery, the cell supply as you go through some of the logistics around tariffs and whatnot. Just curious what you're seeing in terms of that but also on silicon carbide and given with the scale up on EVs as well as a number of other applications you know are you seeing any tightness or potential tightness around components with silicon carbide?

Shuki Nir, CEO

So starting with the batteries it's been a very very interesting year enough you know with supply chain surrounding batteries because of non-fioca regulation tariffs that are changing and and where the source is and we kept saying and this is a very dynamic set of decisions that we have to make between reliability and availability of components, supply chain resiliency, and obviously complying with all the requirements that we have in the market. Because of that our supply chain team has been working with several different sources. We're continuously evaluating which one is the best for us, optimizing all of these parameters, and at this stage from a battery sourcing perspective we feel that we're in a good place but we are continuously evaluating where we are and how we are going to how we can do even better on the sixth side it's you know the the data center we are talking about the AI factories build out here from from our perspective about power but this build out is actually requiring is putting a lot of strain on the entire semiconductor industry including the silicon carbide manufacturers were needed on both ends of the data centers and fortunately for us we've partnered with these companies for a long period of time and and these vendors these are strategic vendors for us we partnered with them on six and another products for many many years and our supply chain team is working hard to make sure that we have enough supply at reasonable prices to continue supporting our customers but as you know from other industries and other markets There is tightness in the market, that is really.

Chris Dendrinos, Analyst — RBC

Yeah, thanks, Chris Dendrinos with RBC. I just wanted to go back to the safety topic here, and you all published a paper this morning on grounding, which kind of leads into this, but as we just think holistically about all the safety aspects that go into higher voltage deployment inside a data center, I mean, can you maybe talk about what else needs to be written from maybe a code perspective that could be a bottleneck, I guess, to this industry, And then as we sit here, I mean, what markers should we be looking out for to tell us that, hey, all the boxes are being checked that will allow, you know, call it scale deployment starting in 2029?

Meir Adest, Other

I don't think there's any code missing about the DC safety. The NEC covers it. There are plenty of safety standards. It's more just a matter of market education and adapting the standards that already exist to this new environment. So that's part of the learning mechanism and having devices which are adapted to this specific application. I don't think the industry needs to really wait for a new standard to appear. I don't believe that will delay implementation of these 800-volt GPUs. I think it's just going to be a matter of execution.

Speaker 4

Hi, thank you so much. I have learned so much. I've got two questions. The first is so that when I think through the architecture that you're proposing for the batteries and the inverters, you're going to put the batteries on the 800-volt bus, Does that mean that you'll have all of the system long-duration backup on that 800-volt bus, or you'd still propose to have a second battery farm outside of the building that would be connecting to that battery bus via 800-volt? So that's the first question.

Meir Adest, Other

We have the flexibility to allow both of them, and from discussion with potential customers, We see that different customers think about this differently and have different approaches. There are really two challenges here. One is the short-term stability of the 800-volt bus, especially in training clusters where you have the whole data center going, you know, 0 to 100 in milliseconds. seconds. And here it's important to have storage on the DC bus without reflecting it out to the grid, which could also cause grid issues and kind of wreck the THD of the AC of the grid. Grid operators don't necessarily like it. So that's short-term storage. My assumption is that almost everybody would want that. And then the longer-term storage for grid outages, minutes, UPS, et cetera, some would take the approach of wanting that to be as well on the 800-volt bus, and then you just add more storage, possibly with different chemistry batteries, whatever. But that could just be added. And others prefer to have it on the AC side, like an AC UPS and a Genset or whatever. Different approaches, we have the flexibility to, you know.

Speaker 4

And the downside of putting that on the 800-volt bus is it has to be near the training cluster and probably inside that building because you wouldn't want to go with a gazillion amps to a building further away.

Meir Adest, Other

Yeah, depending on how big the fluctuations are and what the architecture of that specific data center, that might be the case, or maybe not. Again, it's very dependent on what the actual application is and how they designed it, how they're using it, what the specific neural net needs. It's very scenario dependent.

Speaker 4

Thank you. The second question is just on the essence. I mean, at 40 cents a watt, which is what you had on one of those slides, like, I mean, the 2029 number is a tiny fraction of what NVIDIA is supposed to do in 2029. I mean, it's like one and a half data centers or something like that.

Shuki Nir, CEO

So we looked into that. And during the break, some people told us that our estimates are a bit too optimistic. You're now saying that our estimates are a bit more than a bit on the conservative side. I think that we spent a lot of time trying to understand, there's three ramp ups here, right? It's the build of the data center, of the AI factories, it's how many of them will go 800 volt DC, and how quickly they're going to convert to DC native. When we put all of these assumptions together, this is how we are getting, this is what we're getting at, but NVIDIA are definitely more optimistic than us.

Dave Arcaro, Analyst — Morgan Stanley

Hi, thanks, Dave Arcaro with Morgan Stanley. Wanted to ask about the 99% efficiency level on the SST, how defensible is that? Can competitors catch up or can you push that even higher as you continue to develop the product? What kind of a moat do you see around the efficiency for SSTs?

Meir Adest, Other

I'll talk about us, not necessarily about our competitors, but to answer your question of whether or not we could push it up, yes, we actually have been pushing it up. So I think there's some more room to go there. And I think different topology, the efficiency at the end is a product of what topology you chose and exactly what switching devices you chose. And finding that sweet spot, which we not only believe, we've been hearing from customers that seems to be better than what our competitors are doing, was a tricky ordeal. And I think, you know, if you choose the wrong components, you can't really get there, but, you know, anything up to 100% isn't against the laws of physics, so.

Dave Arcaro, Analyst — Morgan Stanley

Got it. And other patents or, you know, just other intellectual property that you think would come into play here and help, you know, protect that?

Meir Adest, Other

Yeah, I didn't mention in the presentation that we have a number of patent applications which we filed around exactly both the solid state transformer and the conversion side and the safety side. So yeah, there's definitely defensible IP there. Got it, thank you.

Gus Richard, Analyst — Northland

Thanks for hosting this event. Gus Richard with Northland.

Meir Adest, Other

I'm just curious about the timing for prototype shipment. you know when do you think you'd be ready to have something to ship you know what's the cycle time of building one what's the scale of of a prototype is it five megawatts you know can you give a little color on you know how you're progressing towards getting one in the field so most customers for the prototype are interested in in a unit a five megawatt unit some of them are asking for also some earlier you know partial prototypes but but when we say prototypes in 2027 we're talking about the the containers with with the full SST working at 5 megawatts and anything above that is basically replications of of that container in terms of having being ready to ship a prototype and the cycle time to build a system.

Gus Richard, Analyst — Northland

Can you give us any color?

Shuki Nir, CEO

Yeah, no, absolutely. So for the prototypes, as the mayor said, different customers are defining different pilots. Some are asking for the 34.5 kV going all the way to 800 VDC, and there we will ship the 5 MW. There are some potential customers that are asking for reduced level, potentially because this is the grid connection that they have in their test site, potentially because others do not talk about 34.5 kilovolt product yet, but they are looking for 13.8 kilovolt, for example, and over there it's 2 to 4 megawatts. From our perspective, our product is customizable, and for us to ship the full product or a product that has a lower AC going into it is more or less customization at the end of the production line. To a question about scale and lead time to build the products, obviously the first products are going to be built. We are going to ramp up manufacturing, if you will. As we scale, we are going to partner with one of... SolarEdge for the last 20 years, our model was around supply chain, was to partner with the likes of Flex and Jabil. They've been our close partners for many, many years. It's a model that is working well for us. We bring a very good technology. Together with them, we're ramping up manufacturing, but then we're leveraging or benefiting from their expertise in mass production. Our plan around SST is going along the same lines. So initially, the first few units that we're going to ship, we're going to build them ourselves, similar to what we're doing on the solar side. But later on, we're going to ramp up manufacturing. As I mentioned earlier, our supply chain is going to be based in the U.S. to make sure that we have the resiliency that is needed and, you know, in case of domestic content considerations and other considerations that these customers may have.

Joe Nussbaum, Analyst — BNP Paribas

Hi, this is Joe Nussbaum with B&P Paribot. Thanks for hosting. I know we're anchoring to 2030 to not get ahead of our skis, but kind of what informs your view on the scale of the industry into the 2030s with power equipment backlogs and load interconnection requests only going out to, like, 2032 today? And then within that question, like, how much of demand do you see coming from retrofits versus new data center constructions? Thank you.

Shuki Nir, CEO

Okay, so I remember the second part of your question about retrofits, and then you'll please remind me the first one. So about retrofits, initially we don't expect, the model doesn't assume retrofit revenues. What is the lifetime of GPUs and RECs? It's up to, there are different discussions about whether it's two years, three years, four years, five years. So if you go four or five years back, the size of the data centers, the retrofit opportunity is there, but it's not that significant. So we didn't take that into account. And your first question was?

Joe Nussbaum, Analyst — BNP Paribas

Yeah, first question just kind of on how you see the shape of demand going into the 2030s, how you're managing capacity expansions against that.

Shuki Nir, CEO

Like this. Everything that you read, everything that you look into who is talking about an acceleration of this build out. The power requirements of the rack. Look, today, a rack requires less than 100 kilowatt. In 2028, it will be more than a megawatt. It's a 10X factor. And the power density of the racks is going to continue growing. That was the conversation about 1,500 volt versus 800 volt. So as we move forward, the power requirements are going to continue growing, and our estimate is that demand for AI factories is going to continue growing as well, which I believe is something that many people share with us. So when you look into all of that and the transition to 800-volt DC and DC-native solutions is going to almost be over by that time, and at the retrofit and others, so that's how we come to that conclusion.

Gus Richard, Analyst — Northland

Are there any more questions?

Philip Shen, Analyst — ROTH Capital Partners

Thank you. So this question is from a large shareholder. He's asking a short question on SST. Silicon carbide has a lower switching speed than GAN. In what way does this become a challenge when competing with GAN-based SST solutions?

Meir Adest, Other

Yeah, so I actually don't think that's an issue for the application. We're talking about it. It's true that that GAN could switch at higher frequencies now Megahertz is instead of let's say hundreds of hundreds of kilohertz is for for silicon carbide and that makes GAN attractive for Certain applications like for example, you'll see a lot of it in small chargers where you could you know increase the frequency makes the magnetics smaller Okay, that's that's very nice. It isn't necessarily the pain point in data center actually in data centers when we're talking about medium voltage connection I think a more important deciding factor is the withstand voltage of the device and typically silicon carbon devices go to higher voltages than the GAN devices that's why I think if you if you look at SST's SST designs that are based on GAN you'll see not as they've shown today you know dozens of unit but maybe hundreds of units and also lower lower efficiency because I'm not I'm not I don't know I haven't seen anything working at these megahertz frequencies for this application so silicon Gans are great I'm not sure I mean when we were

Philip Shen, Analyst — ROTH Capital Partners

looking into designing the SSD and and Gans were on the table we decided to go with the devices that we decided to go with because we thought they're a better fit for this application investor question what would your target margins have been in the base business excluding SST historically they've hoped for or you guys have hoped for maybe 30 to 32 percent with five percent IRA boost so then maybe a 25 to 27 percent does that sound like the right ballpark you know between the two part of the business but as I said SST is definitely important

Maoz Sigron, CFO

contributor but we will not get to the split between the two of them it's it's it's a one number that we are sharing so thank you for the questions it keeps us sharp and I'd like to wrap it up and basically where we started.

Shuki Nir, CEO

SolarEdge is a very strong foundation. We built it for two decades. Meg was there from the start. I joined about a year and a half ago. And we're going to use this strong foundation in order to not only grow our top line, but actually expand our margins and grow our bottom line as well. We're going to do it by accelerating the growth in the core business. We talked about it. The markets are evolving. Storage is becoming a bigger deal. Our confidence level is higher because of the safe harbor transactions that we've signed. And we believe that we can continue winning in C&I. We are going to unlock the AI factory opportunity. We talked about it whether it's exponential that starts here or starts there, whether the ASP is here or that. Overall, it's a big, big, big opportunity. It falls right down our alley. This is an expertise that we've developed in the last 20 years, and players in the ecosystem are telling us that we're ahead of the competition, both in terms of the performance of the product as well as the timeline. And we're going to do both of those while expanding our margins. So we're very, very excited about what's ahead of us. I hope you are as well, and we'll definitely keep you posted as we continue to make progress. Thank you very much.