All right, awesome. Let's get started. Hi, everyone. My name is Akhil Bhateja. I'm going to be your host for the webinar today. Welcome, and thank you for joining us. I know everyone's really busy, so we really appreciate you taking the time to spend with us today. To make sure we get to the content, I'm going to keep everybody on mute. Whenever you do have a question, just put it in the Q&A section of the chat, and I'll address it at the end. Throughout the webinar, if you'd like to schedule a meeting or a demonstration with the Bloom team, please put it in the chat, and we'll make sure we follow up on that. Let me start by introducing myself. My name is Akhil Bhateja. I'm the Director of Technology Strategy here at Bloom Energy.
I've been at Bloom for about eight and a half years now. It's been a really fun ride. Got to wear a number of different hats across my career here, from business development to sales to product management, and most recently in this technology strategy role. In this role, I get to meet with a lot of people in the data center industry, learn about their challenges and opportunities. I'm really excited to be here today to talk a little bit about the fuel cell technology and how it can help address some of these challenges and help you maximize your opportunities. Before we start, I'd just love to draw attention to the slide I have here, which is one of our large-scale installations in Korea. This is in the multiple tens of MW running and operating today. With that, let's jump in.
Since this is a Fuel Cell 101, I thought I'd start with some basic facts just so we can level set everybody and make sure we're all starting from the same point of knowledge. Working in fuel cells for the last eight and a half years, one thing I hear very often is, "Are you a battery?" or "Are you a backup power generation source?" I want to address that head-on. We are a form of energy generation technology. We convert the chemical energy in some fuel, typically natural gas, into electrical energy and reuse it as a primary power source. That means we're always running, producing power. We produce power at a very high quality, often quality similar to a UPS. These things run [uninterrupted] continuously. As we dive in a little more, we're going to understand why that is and how that works.
The other thing I often hear is that fuel cells run on hydrogen. I want to make it clear, we run on pipeline natural gas. Yes, we are certainly capable of running on hydrogen, but a vast, vast majority of our installations today run on natural gas. We're an ultra-reliable solution. We can configure a solution to go even beyond the 99.999% availability. I'll talk a little bit about how the fuel cell works, but we're a non-combustion technology, and that means we have zero air pollution. That means no NOx, no SOx, no particulate matter. We're also really efficient, and that means we produce less CO2 per unit of energy than any other technology. Extremely scalable technology. On the right, I've got some pictures of an installation.
You can imagine them like Lego blocks where you just have to place more and more and more to reach the scale that you want. This gives us a lot of flexibility in how we install and allows us to often grow with the customer's needs. I'm happy to say we're ultra-price competitive with other solutions. This was not always the case, but over the last four- five years, we've certainly converged. Later on in the presentation, I do have some sample economics for us to walk through and understand that. With that, I put in some of the other myths that I often hear about fuel cells. Definitely the hydrogen, definitely the batteries, the backup.
The one I think that I find most interesting is the last one there is, "If they're so good, why aren't people using them?" That is a really fair question. To that, I respond, "People are." Bloom alone has over 1.5 GW deployed across 1,200 locations all over the world. Honestly, every time I hit refresh on this slide, that number gets bigger and bigger. I would like to draw attention to the left of this slide where we announced just a few months ago a marquee deal with AEP and American Electric Power for up to a gigawatt of fuel cells. On the bottom left, I have got our data center credentials where we have over 400 MW installed in data centers today. Equinix alone, one of our biggest customers, where we have exceeded over 100 MW.
On the right, I've got a smattering of some other logos from different industries, and this is just a small subset of our total install base. As I think about it, though, the reasons why people buy is really important, and it really comes down to four factors. One, time to power. We're often able to install power really quickly to meet urgent business needs. Businesses don't want to have to wait for the utility to come build out, wait for other on-site generation technologies to get permitted and installed, and we're able to get you up and running extremely quickly. Two, we're highly reliable. We're going to talk a lot about that as we go through this today, but we can really maintain a really high uptime and give you statistically significant superiority in resiliency. That becomes really important to hospitals, to data centers, to critical manufacturing.
Third, we're cost competitive. Often, we have a substantial impact on the OpEx spend of companies' energy spend. Finally, sustainability. Like I said, that no NOx, no SOx, no particulate matter, that really helps with companies' sustainability story. The lower CO2 footprint certainly helps too. All right. With that, let's get into the chemistry a little bit on how a fuel cell works. On the top, I've got an image there of a cell. That is the fundamental building block of the fuel cell. I'm going to age myself a little bit with this reference, but it's roughly the size of a floppy disk. It's made of an electrolyte, and we print on the anode and cathode layers. Natural gas comes in on the fuel side, the anode. It reforms into hydrogen.
That hydrogen reacts with oxygen from the air, which comes in from the cathode side and produces a little bit of CO2, H2O. Because you've got the oxygen ions moving, you have electrons moving in the opposite direction, and that's how we make electricity. On the bottom, it shows a little bit how we take that simple cell, that floppy disk-like device, and scale it up to the large hundreds of MW that we were talking about earlier. The fuel cells, we copy-paste them over and over again and make stacks, which we put into what we call the power module. A power module is roughly the size of a refrigerator, and that's capable of producing 65 kW. Five of those makes an Energy Server, and that's the Lego block that you just keep seeing repeating over and over and again. A couple of interesting things here.
There's the core chemistry part of it. There's no moving parts. This is an electrochemical reaction, non-combustion. So these are all reasons why we're highly reliable. It tends to be the moving parts that is what things that break. Because it's a static device, we're able to run for a really long, long period of time. How do they stack up? In my conversations with the data center industry, with you guys, I've heard that a number of things are important. Reliability is really important. I mean, all data centers have uptime requirements and have SLAs with their end customers. Cost is really important. It's a competitive industry. You have to make sure you're passing on the lowest cost possible to your customer. Time to power.
We're seeing this huge boom of data centers from the AI demand, and being able to deploy quickly is really, really important. Load following. This is one of my favorite topics, and one we've spent a lot of time recently, I've been thinking about. As workloads and data centers move more from CPUs to GPUs, we're seeing a dynamic variation in their power consumption. It is really important that your energy technology is able to keep up with that. Power density. Real estate's at a premium, especially where these data centers are going. Any space that's not used for a data center, not used for GPUs, is lost revenue opportunity. Sustainability. Every company has ambitious sustainability goals, and it's really important we are able to meet them. Through this presentation, I'm going to compare the three technologies you see here on the left across these parameters.
On the top, I've got a gas turbine with large scale, like the tens of megawatts, call it a 50 MW building block. Below that, I've got a reciprocating engine, also run on natural gas. At the bottom, it's the fuel cell technology. All of these are forms of on-site generation, which we're seeing data centers more and more adopting. A little bit about Bloom. We're on the 6th-generation of our technology. A lot of work has gone in making sure that they run longer and more reliably. We're very happy to say that the median time between our fuel cell replacements, the scheduled maintenance, is in excess of five years now. We have millions of hours of runtime across our fleet. Microgrids, sites that are designed to run when the grid's down, have a very high availability.
Getting close to 99.999% just from the natural technology itself. All right. Let's talk a little bit about some of the other technologies. As I go through these next few slides, there's a couple of key concepts that I want you to understand. What I'm trying to get at here is that, depending on the size of the building block of the technology and how available it is, you have to overbuild a certain % to meet a 99.9% availability. Why did we choose 99.9%, first of all, to level set? That's roughly the size, that's roughly the availability of the grid, and that's what we often hear from data centers as the target. Data centers obviously have UPS, backup, et cetera, et cetera. They want their generation technology to match what they already had, which was the grid.
With a turbine, which is a 50 MW block, roughly 95% availability, you need to have a full additional unit to meet that availability number of 99.9%. That means 150 MW. That is 50 MW worth of CapEx that you are investing in, hoping that you do not have to use. Similarly, for engines, engines now have smaller building block size. So when they go down, it is a much smaller loss of the total load. That helps. You can reduce the amount of overbuild you need because of that. However, they have a lower unit availability. They are down a little more often than the turbine, 93% versus 95%. With that, it is somewhere between 30%-50%, depending on the technology use of overbuild that you need. Now, let us contrast that with the fuel cell. Our building block is 325 kW, the smallest of the three we have seen so far.
Our availability is the highest, at 99.7%. With that, you just need a 9% overbuild, making it really competitive when you're deploying this technology. Let's try and look at what that means when it comes down to math. I'm looking at a hypothetical 100 MW illustrative data center here. Obviously, every data center has different install conditions. Soil conditions will vary, summary seismic zones, et cetera, et cetera. For illustrative purposes, on the left, I've got install cost per kilowatt of the generation equipment. Yes, even net of ITC, the fuel cells are a touch more expensive today. However, because of that overbuild percentage we were looking at earlier, your total CapEx outlay is significantly less. In fact, you can save up to $66 million just on a 100 MW data center just because of that overbuild.
Now, when it comes to OpEx, these things are running on natural gas. We are by far the most efficient technology with a lifetime average of 54%. That's not our startup life efficiency. That's lifetime average. With that, it's just linear math with the $5 a kilowatt. It's significant. It's close to $15 million every single year in fuel savings. What I did is I just took those numbers and put it in an illustrative five-year term for a 100 MW data center. You can see it leads to a total cost of ownership 15%-25% less. Over the years we've seen Bloom reduce our cost significantly to the point where now we are cost competitive with these other technologies. Load following. Like I was saying, this is something we've spent a lot of time working on in the last couple of years.
It's become more and more important as data centers have evolved. I've got a pretty busy graph down there, so let me take a step back. The green line is what the server, like a little voltage sensor on the actual rack, is tracking. You can see it's fluctuating up and down quite significantly. The orange line is the fuel cell system. As you can see, we're able to keep up and match those load fluctuations. This is real data running on actual NVIDIA chips using the Llama model, the open-source model from Meta. This is significant real-world data. We've been really lucky to get our hands on those chips and do this testing. We're very, very pleased to see how our system is able to keep up. Some of the other parameters that I think that we hear from you that's really, really important.
I touched upon real estate. Land is really valuable in some places. I mean, here in Silicon Valley, where I sit, Bloom has significant installations in Korea, Seoul. Let me tell you, it's really important. We're able to stack our solution to meet your need there. If you do have plenty of space available, land's cheap, you can just install us on a flat and kind of take advantage of that abundant real estate and would hit 30 MW an acre. More and more in these cities, we're seeing ourselves stacking where we're going up to even all the way up to high with four levels, up to 100 MW an acre. The pollution aspect is really important too. I mean, because of the more efficient, we consume less natural gas per unit of electricity. That means less CO2 per unit of electricity.
That's about a 35% mitigation. Because we're non-combustion, there's no NOx, no SOx, no particulate matter. Water. Our system is really interesting. We use a little water to reform the fuel, but then the next step of the reaction creates water. We are water neutral. Energy generation technology is really thirsty. This is becoming more and more important, especially in geographies which are water limited, like California, Arizona. Lastly, and I think my favorite point is the noise. These systems are really quiet. It comes down to the fact that there's no moving parts. It's unlike any other form of energy generation technology you can imagine. I often give customers tours of our system that powers our headquarters where I'm sitting here today. You can have a conversation right in front of the system.
Half a mile away, the freeway is much louder than the system itself when you're standing right in front of it. Finally, most importantly, time to power. We're hearing this over and over again from our data center customers and from you folks in the industry. The needs are now, and people need to deploy their data centers as quickly as possible. Bloom spent a lot investing in the infrastructure to meet this need. We have a manufacturing facility here in California where we do our cells and stacks, and we do final assembly in Delaware. Totally American-made product, by the way. Those factories have been designed with scale in mind. We're about a gigawatt a year today. As you can see in the graph on the right, able to double almost every year. This allows us to get the.
Unit on site in as short as 90 days. Typically, with projects, it's permitting. That's the longest lead time item. That's where I think we have another huge competitive advantage. Because we're non-combustion, that non-NOx, non-SOx, no particulate matter, we're a lot faster through the permitting process. In fact, here in California, where there's some jurisdictions which are some of the hardest permitting agencies in the world, where we're exempt from the air permit, that's hugely significant and can lead to a much faster deployment timeline for your project. Finally, I think my favorite aspect of the technology is how future-ready it is. You can take some of the waste heat from our exhaust and do a combined heat and power solution. We've got projects where we run district heating, where we run hot water, where we run steam for industrial processes.
More and more, we're running absorption chillers to provide cooling for the data centers. Carbon capture, because we're non-combustion, our exhaust stream is a pretty pure stream of CO2. That means it requires a lot less energy and a lot less CapEx to purify that stream to get it able to be sequestered or utilized in some way. DC power. We're a native DC source of energy. In fact, some of those boxes that you see are Northern boxes. As data centers explore the idea of DC data centers where they can reduce their CapEx, reduce their equipment, and improve their efficiency, we're right there ready, just having to take out the equipment to meet that need. Finally, we're fuel flexible. We can run on hydrogen today. We can run on blends. As and when the molecules become commercially available, Bloom is ready for it.
With that, that's the end of my presentation. Just wanted to quickly plug our mid-year data center report. This goes into a lot more detail than I did and has some actual data that we've surveyed from data center customers. It's a fabulous read. Highly recommend downloading it and visiting our website. There are some great resources there. With that, I'm going to start turning to the Q&A. Yeah, please keep the questions ready. First question, what's the expected lifetime of fuel cells? Can they be used in a 20-year PPA? Great question. Absolutely, yes. Bloom does contracts all the way up to 20 years. The system as a whole can last 20- 25 years. Inside the system, yes, the cells do degrade. I had the median life between replacement. That's what I was talking to.
Right now, I think we're about a 5.5 year median life before we do that replacement. It's a statistical distribution, so that's why I'm using the median. When we do that replacement, though, it's really interesting. That specific module we turn down, that 65 kW module, the rest of the system continues to operate. It's concurrently maintainable. The customer sees no drop in the load. Imagine 100 MW, you're only losing 65 kW. It takes about two hours to do that replacement, and we put the new one in, take the old one out, and then power it up. It's continuously maintainable. Also, what's really cool is that that old unit we take back to our factory and we recycle or reuse 98% of it by weight.
From David Orstaner, given that data center future will require 500 MW to all the way to 1.5 GW, are you seeing potential customers wanting that entirely to come from Bloom? If not, why do they shy away from 100% fuel cell powering? Absolutely, yes. We're seeing data centers that are completely powered by Bloom. The fact is, I think most data centers, in an ideal world, would just love to continue buying power from the grid. However, that's less and less available, and that's why they're moving more and more towards on-site generation technology. When it comes to on-site technology, yes, we're seeing customers fully comfortable with us and able to meet that need. The next question, can we explain how a supercapacitor is used in Bloom solution? Yes. A supercapacitor, for those who know, is a form of energy storage.
It's a device that stores very little energy, but it's very fast and responsive. In that graph that I had of showing how the NVIDIA chip fluctuates and how we're able to meet the need, what I'd say is like the nanosecond to microsecond sort of response, that's handled by supercapacitors. In fact, some capacitors already exist on the rack itself, some on our system. The second-level response, which is still really fast, is handled by the Bloom system. Next question, are we seeing demand for combined heat and power for cooling through absorption chillers? Cooling seems to be a great question. Yes, we absolutely are. The industry, I'd say the data center industry is adopting this as we speak. In fact, we did a project very recently in our own factory where we're running the spare heat through an absorption chiller.
That's been running great for a while now. Now we're starting to see data centers explore those designs and kind of test their waters a little bit with that. Next question. When the replacement is needed, can the latest-gen cell drop in and replace the older cells? It's a great question. Yes. I like to joke that our factory is at its most optimum when it's making the latest and greatest. We do not want to be wasting time and resources making the last-generation technology. We design them to be forward-compatible. It's like that iPhone plan where you always get the new iPhone. When Bloom drops in the newest cells, we put in the latest and greatest technology to that. What are water requirements, if any? A very tiny amount of water is required just during startup. Once we reach steady-state operation, we are water neutral.
In fact, we emit a little bit of water as water vapor on the outside. We are slightly water positive, but it is really minute. On startup, though, we do require a little bit of water. The cell degradation, I think I handled with the replacement. Can these fuel cells run on propane? It is one of those things where there is a theoretical answer and then there is a practical answer. Theoretically, actually, the solid oxide technology is really fuel flexible. With enough engineering, I am sure it can run on a variety of molecules, including some of the higher-order hydrocarbons like propane. However, commercially, we are only focused on natural gas solutions today. I have got a question on how do we compare to nuclear energy or the data centers in Georgia. We are certainly looking at Atlanta as a key market.
We've got a lot of projects being explored there. I mean, nuclear energy is very different. The way nuclear exists today is it's large-scale, very centralized, not an on-site generation technology. It still requires that transmission distribution technology. We do see technology evolve and some of the small modular reactors be planned and announcements be made for on-site generation. Candidly, I think that's still a few years away. It's going to be a while before permitting agents get comfortable with that. People are comfortable with having a nuclear plant in their backyard. How significant is the addition of fuel cells in the recent legislation changes? Happy to say that, yeah, fuel cells are still covered and that the ITC is still available to fuel cell technology. There's a question about specific Equinix data centers. I'm not sure if I can go into customer specifics like that.
Do we stagger the replacement of fuel cells to keep a customer running? Yes and no. I think that happens naturally. Like I said, it's a median life. It's kind of like the double A batteries in your remote. One will wear out faster than the other just because of electrochemical different conditions. That happens naturally. We do make sure that the customer, we're always meeting the minimum load requirements and what we have in the contract with the customer. We are very focused on that. We plan our replacements to make sure that we're always able to meet those contractual needs. How much does supercapacitor add to the system cost? Typically, it's pretty diminutive. It's a very small part of the overall system. Why do those nanoseconds matter? I mean. It's just the level at which the chips are fluctuating at, right?
I'm not a chip expert, so I don't really know how to go into that level of detail, but it's what we've seen in the chips. Can you discuss how natural gas is used in fuel cells? Yeah. If you think back to the image I had of the cell itself, the natural gas, which is primarily methane, lands on the cell itself. It gets reformed internally into hydrogen, and then the hydrogen pulls the oxygen ions across and causes an electricity move. Are we working SMR here? I assume you mean the methane reformation as opposed to small battery reactors. We do that on cell in-house. Are you being used in new gigawatt clusters? Yes, absolutely. I mean, every time I talk to the sales team, it seems like the deals are getting bigger and bigger, and it's a really exciting time.
Is water required from the carbon capture in fuel cells? Great question. Definitely not from the fuel cell component. To our exhaust, what comes out is basically a stream of CO2 and H2O with small traces of other stuff. The next step is to knock out the water. In fact, I suspect that we're probably water surplus, again, even in the knockout stage as opposed to water consuming. The percentage of new customers using Bloom versus backup power, 100% of customers. We're always a primary power source. We do not do backup power. Do we anticipate maintenance cost design? Absolutely, yes. Bloom has been on an incredible cost reduction story long before even I joined here eight and a half years ago. I have seen statistics that show us a double percentage cost reduction year-on-year.
That's not just maintenance cost. That's the cost of the device itself. That's the installation cost and the maintenance cost. We like to think of cost as total cost of ownership because that's what's important to our customers. We are extremely focused on bringing that down and ensuring that we deliver the most value possible to our customers. The ITC, I think it's till 2031. Oh yeah, how is the performance of fuel cells being monitored for performance? This is one of my favorite things about the Bloom infrastructure. Our devices are extremely heavily instrumented. We have tons and tons of data coming out of it. We have two data centers, one here where I sit in our headquarters and one in Mumbai, India, both perfectly redundant replacements of each other.
They're constantly—in fact, I think the last I checked them at, we're in just 8 billion data points a day. We're tracking voltages, currents, temperatures. Excuse me. That enables us to get really smart and really predictive on when we come in and do the replacement. This is part of how we're able to make sure that we're able to meet that uptime. The head of RMCC, the remote monitoring center that I'm talking about, actually came from GE, and he likes to joke that one single GE turbine has like 1/100 the amount of sensors that even our 65 kW module has. Another question of backup power. Again, I want to make it very clear. We're a primary power source. We're not used as a backup power. Are you running at full capacity? Typically, yes.
Typically, customers like to run us at full capacity to take advantage of the technology we have on site. That being said, very often, we do have sites where we do have to ramp, load follow, and fluctuate up and down. There are periods of time where we're not. For the most part, yes, we are running at full capacity. Laura asked a great question about what's driving longer than the 90 days. Is it just permitting? Typically, permitting is the longest tent in the pole. That's what we see as driving project timelines. In my enough years here, I'm not sure of a single project where Bloom was the limiting factor on the schedule. It's always external things and typically permitting. Maintenance cost typically lower for larger deployments? Yes. Of course. There's an element of scale for it.
That being said, because it's so heavily automated and it's so well run, it's pretty limited the amount that it scales down. Do you reform CH4 before this SOx cell? No, we do on-cell reformation. Next question, congratulations on 6th-generation. yeah, that's about the data sets. Yes, we do make some of the data available to end customers. Now, most customers don't want to see the temperature varying by the second. It's available either through a portal, which we give out web access to, which shows, I mean, some of the metrics that customers really care about, such as how much power is being produced, what efficiency, how much gas is being consumed, et cetera, et cetera. We can also integrate into different building management systems, et cetera, to make that data available real-time to customers. What are some of our larger deployments?
Are they able to replace generators or just supplement load for this thing power? Great question here. I mean, we've got multiple deployments that are as high as 50- 80 MW and beyond. It's an interesting question on being able to replace generators. In some conditions, yes, because we can configure ourselves as a microgrid where we're able to operate grid independently. Then often you can have the grid coming in as a backup. Some customers have gotten comfortable with this idea. In that place, you can displace buying diesel generators and UPS and save some CapEx there. The largest system installed in California, I'm not sure about California, but in the U.S., on the East Coast, I am aware of at least a 30 MW site that's the largest that comes to mind.
I'd have to look up with the team that knows what's the largest one in California. Is there a manufacturing capacity being run at full capacity? We're highly scalable. It's kind of interesting. I love doing customer tours of the factory. I think my favorite thing to do is to show off the technology and how it's come. Every time I go there, there's a new machine added, a new line added. The ball next always kind of moving around, and our operations team is really world-class at keeping it up and running. I'm sure if I was to ask my CEO, he'd tell me, he'd laugh and tell me that if we were to sell it, he'd find a way to build it.
I think we're running at full capacity with what we have now, but it's very easy for us to scale and add that machinery. We have tons of space there and are able to kind of expand as needed. How soon do you think we have a deployment using carbon capture in the field? Interesting question, right? I think the technology is really ready, right? The fuel cell has been running for a long time. I mean, we do not have to do much. We literally have to just replumb our exhaust to give it to someone who's doing the carbon capture. The carbon capture technology, like the purification of CO2, that's been known for a very long time. I think the use cases of what you do with the carbon molecule are still evolving.
Like if you're going to sequester it, which is probably what you're going to do on a larger scale, you need to have the welds ready. You need to have that infrastructure built. That's, I think, what the industry is waiting for, or utilization. That being said, there's project discussions going on all the time. Yeah, watch that space. Pay attention to our press releases. I expect it to be soon. How much of a project. Expenses versus CapEx? I think I had the numbers up there, like the fuel costs. I forget, but we could flip back and look at the ratio. Yeah, that pretty much covers it. How does the product address low transient site? I think I addressed this, but the supercapacitor handles the fast transients and our system ramps up and down on the second level.
Somebody's asking about the city permit and the honest answer, it depends. Like I said, there's HGs here in California where we're exempt from air permits. They tend to be much faster. If you take me to a new country and a new HG who's never heard of us, it might be a little longer because there's a little bit of the cycle and education. I think I have the maintenance cost question. The details of the non-fuel operating costs, the service costs. On a short term, like five years, like I showed, it's pretty much a wash with some of the other technologies, like the thing. Unlike them, though, we don't need a major overhaul with that massive shutdown and the engine rebuild and all of that stuff. Ours tends to be very modular, so it's like you can do the maintenance on site.
The way I like to think about it is like there's kind of three levels. Things like passive things like air filters and stuff like that, they need to be replaced every so often. Very easy, it takes just a few minutes to take out that air filter, put in a new one. There's a few moving parts too. A blower to build in the air. Like you saw, we need oxygen. That will break down every year and a half, two years. We're really good at predicting it with that maintenance, but that's again a pretty short one. Then the every 5.5 years, like I was talking about with the stack replacement. I've got a question on the CapEx for the overbuild. Yes. As you saw in that slide I showed, we're a touch more expensive per kilowatt on the CapEx.
When you take into account that overbuild to meet that grid level reliability, that's where you start to see significant savings. That's quickly changing. Honestly, if you're to get into a sales team, every time I talk to them, the vibe there becomes more and more compelling. Does Bloom work directly with hyperscalers? Energy? Yes and yes. Tom asked a really good question on where data centers initially deployed Bloom boxes to expedite operation and then transition to other sources. Or more often than not, the grid than other sources. I feel like the data center industry, I mean, you guys are the experts, have a great comfort with the grid. It's just how business has been done for the last few decades. Absolutely yes.
There are business models where we've worked on where we can move the system to another site as the grid becomes available. We can kind of work with the customer to redeploy them. Yes, that's definitely something we can look at and we've done in the past. Okay, can you explain the hydrogen generation system and the integration with fuel cells? The hydrogen generation system is a different thing. It's an electrolyzer. You can think of it as a fuel cell running backwards. A fuel cell takes in chemical energy, produces electrical energy. The electrolyzer does the exact opposite. Takes in electrical energy, uses it to split the bond between H2O , and there's no hydrogen. Our electrolyzer is the most efficient, but it's not really the topic of today. I'll just leave it at a high level like that.
There's some more information available on our website if you like. If you reach out, our sales team will be happy to talk to you more about the hydrogen generation system. How many megawatts an acre would regular and power tower installments? For flat, we're at 30 MW an acre. If we're going to do a two-story system, it's slightly less than double, 55 MW an acre. If you're to go four stories, it's slightly less than 4X, so it's 100 MW an acre. Again, asking our question on the capacity, yes, we're about a gigawatt a year annually today. Honestly, again, that's the other statistic. Every time I hit refresh, it changes. Where are we HQ'd? We're HQ'd in California, here in Silicon Valley, in San Jose. How are we working with the nuclear industry?
On the fuel cell thing, we're not a form of generation. Like I was saying earlier, nuclear tends to be large and centralized. We tend to be on-site and easily deployable. The biggest fuel cell competitors? Honestly, there's a host of companies developing the technology. There's different applications for fuel cells. So we're very focused on the stationary power generation. There's different types of fuel cells, and those are really good at mobility and kind of backup power. And you'll often see those, but they're going after different markets.
Hi, Akhil. Sorry for chiming in. We are a bit over the presentation today, and I know there are some great questions that are coming in, and we'd be happy to follow up with those after the webinar concludes. Thank you. No worries. Thank you, everyone. Appreciate everyone's time.