EMCORE Corporation (NASDAQ:EMKR)
Wall Street Analyst Forum
March 26 2008 12:30 pm ET
John Iannelli - CTO
Good afternoon, ladies and gentlemen in our ongoing attempt to adhere to the publish schedule, I'd like to introduce the next company in this afternoon's Alternative Energy and Clean Technology program.
EMCORE Corporation is a leading provider of compound semiconductor-based components and subsystems for the broadband, fiber optic, satellite and terrestrial solar power markets. EMCORE's Fiber Optic segment offers optical components, subsystems and systems that enable the transmission of video, voice and data over high-capacity fiber optic cables for high-speed data and telecommunications, cable television and fiber-to-the-premises networks.
EMCORE Solar Power segment provides solar power or solar products for satellite and terrestrial applications. For satellite applications, EMCORE offers high-efficiency compound semiconductor-based gallium arsenide solar cells, covered interconnect cells and fully integrated solar panels. For terrestrial applications, EMCORE offers concentrating photovoltaic (NYSE:CPV) systems for utility scale solar applications as well as offering its high-efficiency GaAs solar cells and CPV components for use in solar power concentrator systems.
For more specific information about the company and its products, their website is emcore.com.
Without any further introduction, I would like to introduce John Iannelli, Chief Technology Officer of the company.
Thank you. Good afternoon. I am John Iannelli, Chief Technology Officer with EMCORE and I will concentrate the talk this afternoon primarily on our solar business. The company again addresses both the fiber optics market as well as the solar, [solar being in both space] and terrestrial based. So, we will spend most of the time on the photovoltaic solar piece.
Okay. So, as we have mentioned, the company operates with four business units. Our two business units are focused on the fiber optic communication side of the business. The two other, the two remaining business units are focused on the solar or photovoltaic side of our business.
On the fiber optic side as briefly, we play in essentially two different markets. Our first business unit is what is known as EMCORE Broadband. This business supplies fiber optic hardware into primarily, cable TV operators or cable TV, HFC fiber optic networks and also into the fiber-to-the-home networks, namely, that’s Verizon has been rolling out now for several years.
Our second business unit in the fiber optic space is what is known as the EMCORE Fiber Optics or EFO. Traditionally, we supply products into the datacom market. More recently, we have now expanded this business to take a larger footprint in the telecom market. We did this through the acquisition of the telecom fiber optic business from Intel that we announced last December. And we just closed that acquisition late last month in February.
On the photovoltaic or solar side of our business, again, there are also two separate business units there. The first business unit is what is known as EPV or EMCORE Photovoltaic. This was the original business unit established, put together about 11 to 12 years ago and the markets that we predominately address here are very high-efficiency solar panels for satellite and spacecraft deployment. And I will talk a little bit more in detail shortly about these markets as well as this technology that we have developed for this area.
The fourth business unit is EMCORE Solar Power. This is a relatively new business unit that we put together about two to three years ago and the idea behind this was, could we take some of the well-proven and very high-performance solar technology we developed for our space business and could we leverage that over into a terrestrial solar application. And that’s precisely what we have done over the last few years within EMCORE Solar Power and again, I’ll talk more in detail shortly about the markets and this technology in that business.
Okay. So, talking into more detail now about EPV or EMCORE Photovoltaics, again, this was a business that we had business unit we started in the late 90s and just to put this market in perspective, in the mid-to-late 90s, anywhere from 80% even more than 80% of the solar panel that were being put on satellites and spacecrafts were based on silicon solar cells. At that time, silicon solar cells were and still are a very well-proven and very mature technology. Silicon solar cells were some of the first solar cells that were ever produced many, many years ago.
The downside of silicon in some sense is that it is very mature. If you look at the efficiency of silicon cells, the efficiencies are fairly low anywhere from 14% to 15%, coming up maybe into close to 20% with some of the better silicon out there that SunPower is talking about now. But still, it's getting very close to actually the theoretical limit of what silicon could ever produce, which is in the low-to-mid 20s.
At that time, we saw what we saw with a need here that if we could develop a much higher efficiency solar cell, there would obviously be a benefit. This is what we did. It turns out that the higher efficiency solar cells or what it called, triple-junction cells and I’ll speak on that in the next slide, they are more expensive than a silicon solar cell. However, because the efficiency was much higher, there were fewer batteries for energy storage that had to be put on spacecraft and the overall payload weight started to drop.
So, it's actually economically better for spacecraft and satellite providers to migrate over to the higher efficiency cells. If you look at the industry today, well over 80%, probably, well over 90% of all the solar panels that are put on spacecraft use gallium arsenide-based triple-junction and multi-junction solar cells. So, what used to be silicon having the lion share of this business, silicon now is a very, very small player in the satellite solar market. Again, three, five multi-junction cells are essentially taken this over.
You can see a picture here. This is a picture of our Photovoltaic business, which is based out of Albuquerque, New Mexico and on the Space and Satellite business. This is a business that we have a very large market share. We enjoy a large market share in this business along with essentially one other major competitor.
The other major competitor here is a division of Boeing, known as Spectra Lab and between EMCORE and Spectra Lab within Boeing, we roughly share this business. There are some other very, very small players who are now trying to come into the multi-junction solar cell business, but again us and Boeing really kind of have shared and continue to share this market.
Okay. So, I'll talk briefly about the technology and kind of describe triple-junction cells and how they compare and contrast against the silicon solar cells, since that’s the more mature technology out there right now, at least on the terrestrial front. What you see on the upper graph is a typical spectrum of sunlight hitting the surface of the earth. There are few things here that are important to note. One, the spectrum has a tail to it that extends out well into the infrared. You can see, extends out to 1,500, or even after 2,000 nanometers albeit there is a small amount of energy there, the tail still extending out well past 2,000 nanometers.
The other thing that’s interesting to note is that a large portion of the energy in the solar spectrum is essentially concentrated down in the blue and blue-green portions of the spectrum. If you look at a silicon solar cell, there are a couple of things to note here. And it's really a characteristic of not just a silicon cell, but what is known as a single-junction cell. So, it could be a cadmium telluride, thin film cell. It could be a CIGS single-junction cell or silicon cell. All of the solar cells have what are known as a cut-off wavelength. They will generate electricity based on light that infringes on the solar cell at a certain wavelength or below. Anything at a wavelength longer than its cut-off wavelength, the cell is blind to that light that comes on to it.
On the silicon solar cell, the cut-off wavelength is roughly a 1,000 nanometer. So, what that means is, first of all, any light in the solar spectrum that is longer than a 1,000 nanometers, the cell is blind to that light. There is no electrical power generated out of the cell.
The second issue to realize is that as you start receiving light from the solar spectrum, below 1,000 nanometers for a silicon cells, the silicon cell does produce electrical power from this. However, it's most efficient at its cutoff wavelength. As you start getting more and more or higher, higher energy photon hitting the cell, that extra energy that's below the cut-off wavelength is essentially wasted in the cell as heat. And again, this is just a characteristic of any single-junction cell, no matter what the material system is.
So, one approach to maximize or raise the efficiency of solar cell is to use what are called multi-junction cells and the way to think of it is slicing the solar spectrum up into several pieces, sort of state-of-the-art right now, our triple-junction cells are cutting it into three different pieces. There are some universities looking at four-junction cells, I believe some more seasonally done on five. But three is kind of the state-of-the-art at least in commercialization right now.
And you can see from the three colors here blue, green and red. We illustrate, three different solar cells that we would grow and we would staff the three different solar cells vertically on top of each other and each solar cell preferentially looked at as an individual portion of the solar spectrum.
So a couple of things standout, one, our bottom cells based on a germanium solar cell junction that has a cut-off wavelength that extends all the way out to 1,500 nanometers. So, the portion of the spectrum from 1,000 to 1,500 that standard silicon cell would be blind to, that portion of the spectrum is now received. On top of that, we preferentially receive with a gallium arsenide based cell, sort of the long wavelength of red they are very near to red portion on the spectrum shown here in green.
And then lastly, the portion of the spectrum where there is really quite bit of energy which is shown here in blue, e receive the Curr Top Cell which is based on an alloy of gallium indium phosphide.
The efficiencies on a flat-plate and by that I mean, just a panel which is completely covered with solar material and you will why I made that delineation, in a moment. The efficiency of a triple-junction cell like this is state-of-the-art right now is about 29% in change. Those results we have announced, similar results have also been announced by our competitor, Spectrolab. And again, that compares with other technologies that are anywhere from 6% to 7% up to may be 14% or 15% be it a thin film or a polycrystalline silicon.
Now, the higher efficiency does come at a price, as I mentioned earlier. If you look at a three, five days triple-junction cell on a per unit area basis itself is certainly more expensive than a polycrystalline silicon than a amorphous silicon, than a cad-telluride type cell. Again, on a space satellite application, there are other benefits that translate into making this economically viable. In the terrestrial application, economics do not work for a cell like this, if it's deployed in a full-flat plate type configuration.
So, how can some one utilize this very, very high efficiency cell on a terrestrial application? And idea here is very simple. What we do is we take our cell, again, which is a very high efficiency in terms of converting sun or optical power from the solar spectrum to electrical power. And they are very, very small cell. On our system we use a cell that's about the size of a postage stamp, it's one square centimeter. And on top of this cell, we put a very large collection optical element. We use I assume as a frenal land which is a refractive optical element. There are other folks out there that are using reflective elements, mirrors and so on. So, what we do is we capture and collect sunlight over a very large area. We focus this sunlight down on to our cell. This way, although the cell is higher cost on a per unit area basis, we can amortize the cell cost over a large area of light being collected.
That's the first advantage. The other advantage for this which is not obvious at first to most folks, is that for some reasons I won't go into right now. The efficiencies of our cells actually go up and they go up quite dramatically when they are put under concentrated light illumination. I mentioned before and sort of a flat-plate type application space, we see efficiency is around 29%, under what we call 500 ex-concentration meaning, we are collecting 500 times of the area of sunlight relatively to the area of our cell. We have reported efficiencies over 40% as has Boeing at Spectrolab, they've reported 40.7.
In production, we see cell efficiencies in our CPV systems running at around 39% or 37% in production. And again, 39 was announced in 2007, we introduced that and we announced over 40 late last year.
Just sort of a graphic that we do like to show, to produce a given amount of electrical power, one square centimeter of one of our triple-junction cell in our CPV system, which is shown here. One square centimeter produces the equivalent amount of power as seven five-inch polycrystalline silicon cells. So you can see there is a tremendous amount of savings in terms of the amount of semiconductor material that's required. Again, this is important obviously for a cost reason, also important for a second reason and that is, if we add capacity to build our solar CPV systems in terms of adding megawatts of power. We don't need to scale up on the wafer semiconductor fab side of our business, nearly, as much as someone that was doing a flat-plate type approach with. Since one square centimeter in our system produces about 17 watts of power, you can see that we don't need to scale up our wafer capacity very much at all in order to get several megawatts.
So to compare what are essentially three approaches that are out there right now for terrestrial solar. The first technology, and again the most mature and by far still has the lion share of the terrestrial market is polycrystalline silicon, conversion efficiencies at the modular level. Again, anywhere we see 12% to 14%. Again, some of the vendors are advertising larger numbers. There is a nice talk earlier talking about some of the thermal problems that happened with polycrystalline silicon. So, if you see these numbers, plus or minus a few percent. But that's sort of a ballpark on what poly is running right now. Again, it's a predominant technology. It's a very matured technology from a manufacturing point of view. Still very versatile in terms of several applications in the last, obviously the roof-top market is a big application for that you will see that our system is not geared toward.
On the down side, the [wells], talked about shortage supply constrain in polycrystalline silicon because of that it has held the prices up fairly high we see prices in the high $3 per watt range, they are from $3.50 to $3.80 per watt of polycrystalline modules.
Second set of technologies are thin film, again, there are many approaches at the thin film level, there is cad-telluride, there is amorphous silicon, there is CIG and from inefficiency point of view, thin film tends to be at the lowest end of the scale, somewhere between 5% and 7%. However, the manufacturing costs are very, very low for most of the thin film processes. So, the actual dollar per watt number on thin film tends to be on the lowest, as well. Right now, we see numbers in the low $2 per watt for thin film, you'll see first folks like for solar with their cad-telluride being very aggressive in pushing down some of the ASPs that they are forecasting.
Again, because of that low efficiency none of these numbers here talk about land use. Land use really seems that the real estate use in the sense if it's a roof-top application you have a fixed amount of area that you can cover, if you are going out and doing sort of a grid type in utility power application, you need to lease land or pay property tax on land, you do all things that need to be considered. And again, because the efficiency is so low, you take up the tremendous amount of real estate with most of the thin film approaches. Some of them again cad-telluride have seen environment concerns cadmium as a material is not environmentally friendly, and there are issues there to be worried about.
On the CPV side at the modular level, we see numbers up at around 30%, and again this is the complete module taking new account losses through the optics, losses through the electrical interconnects, the currents are very, very high on some of our CPV systems, so electrical losses are factored into this as well. We were at about $3.60 per watt we are actually slightly below that right now for larger installations in terms of ASP. And we are forecasting that goes down below $2 per watt in three years. And I will discuss in a moment why we think that's achievable.
In terms of land use, it is the most efficient, which is not surprising because of the high efficiency. And again, in terms of scaling-up production, we have the benefit of being a 500 X or even larger concentration ratio; we don't need a lot of semiconductor material there has been a lot.
Going through the down side, but things to realize on CPV, that CPV works in something that we call good sun, and by good sun we mean very clear sky, very low humidity because there is an optical collection element, we need to be facing the sun that need to be a direct normal illumination on our system. We therefore track the son in both axis throughout the day. If it's environment for example, such as here in New York where there is a lot of cloud covering a lot of haze, the economics of CPV aren't as favorable compared to let's say the Mediterranean coast or the desert south-west of United States.
So, as I mentioned, we have a fairly aggressive roadmap in terms of driving the cost and therefore the ASPs of our system is down over the next several years. There are two ways that we feel we will achieve that. One is at the system level and I will show that in the next slide. But the other way is if you look at the efficiencies of the cell, obviously if the cell efficiency goes from 20% to 22% your system cost in a dollar per watt metric just dropped by 10%. If you look at a lot of the mature technologies like silicon, the efficiencies are sort of incrementally getting a little better each year, but they are definitely starting to flatten out compared to the improvements you saw maybe 10 to 20 years ago.
On the triple-junction cells, it is still a, I should say, relatively young technology. It has been around for about 12 years now, but if you look at the efficiency improvements, its still has quite a bit of runway left on it. From 2005, we were demonstrating 33%. Again, at the end of last year, we were at just over 40%.We have a new technology called IMM, which is an acronym for Inverted Metamorphic. In short, it's a way that we can grow the layers in reverse to get an efficiency boost. We are looking at getting to 43% by the end of this year and conservatively to 45% by 2010. So, you can see the efficiencies are going up anywhere from 3% to 4% per year over the last four years. Again, this is a direct cost reduction at the system level at dollars per watt.
Now on the commercial side, we offer products at essentially three levels. We offer bare solar cells, which are shown here on the lower left. We also offer something which is known as a solar receiver. Because in a CPV system, you are putting so much sunlight onto a solar cell, it can be very critical in terms of how you heat sink that cell and also how do you electrically interconnect to that cell, occurrence are very, very high obviously. So, although we offer discrete solar cells to customers, typically what customers are asking for now are these receiver subassemblies where all of the thermal heat sinking through ceramic and all of the electrical interconnections are done for them, they now have a subassembly that they can immediately integrate into their CPV system.
We talked about the efficiencies in production right now, and our roadmap of 45% in two years. We are offering and currently shipping these receiver modules and have orders with several customers. The backlog right now for the cells and receivers is at $86 million. And again, we have announced several purchase orders and contracts with a variety of customers in the CPV industry.
Now, besides offering at the cell level and at the receiver level, we also offer products at the full system level. So, this is a product that we would sell to eventually a power producer. You can see, this is a 25 kilowatt array that is deployed in New Mexico. Obviously, we are not going after the rooftop market. These are very, very large systems. They really make sense on larger scale installations. And again, our target market here are utility grid-type power producers. There have been several purchase orders, letters of intent and MOUs that we have announced over the last one or two quarters. There is a 60 megawatt agreement we have with the Pod Generating Group in Ontario. There are two separate installations in Korea. One is 5.7 megawatt, which we have received the purchase order. For a second, which we are still negotiating a follow on 14.3 megawatt, also for deployment in Korea.
We are currently shipping again two deployments in Spain, one a 300 kilowatt and the second, an 850 kilowatt. And lastly, we've signed an MOU with a company called SunPeak Solar which is based in Southeastern California for two separate installations, one at 200 megawatt and the one at 500 megawatt. And again you can see here, if you look at the geography on several of these deals, you see Spain as being one of the very early adaptors of CPV, because of the geographical advantage. Italy and Greece are also looking very seriously at CPV and obviously the Desert Southwest of the United States. You see a lot of interest in New Mexico and Nevada and California and also in parts of Colorado.
So again, some of the growth drivers here on the component side as I mentioned were one of really just two major suppliers that are in the market right now. We have got a very strong performance roadmap with what we feel is very competitive pricing. The other issues if you remember here is that this is still a relatively new technology for deployment in the terrestrial market. One of the issues that power producers always have is that they look to these plans to be a 20-year life plan, they make with their business models for project financing. Because of that, the reliability of the system really – it needs to be looked at very carefully. And our system and really most of the solar systems out there right now you have to look at the solar cell itself.
It's one of the key active elements in the system. The fact that we have had this technology in these solar cells in space which is actually a harsher environment than a lot of the field deployment here on earth for over 10 years now and obviously, if you are doing a space deployment where there is just no room for error in terms of having callbacks from the field, you need to have a bulletproof reliability record. So that has really enabled us to go out there with the product that we can stand behind in terms of its reliability.
Right now at the component level, we have an annual capacity just over 150 megawatts of cell capacity. Again, we don’t need to have that many reactors online nor to produce this much power. So, this is based on a 500 X CPV system. At the system level, we initially came out with what is known as our Gen-1 system that was running on Sun for eight months. We have now finished development of our Gen-2 system and that is what we are in production with this quarter. We do have, as I mentioned, [demand line adopt with] various strategic partners and supply agreements at the system level. And current annual capacity we are brining in place right now at the system level is just over 50 megawatts. And we can add to both system capacity and components in cell capacity as needed.
So, one other thing I would like to go back here just for a moment and talk about the cost reductions, we mentioned how improving a cell efficiency directly translates to a cost reduction at the system level. The other point that's important to see here is that if you look at our system, the majority of our cost of our system is tied up in things such as plastic lenses, acrylic lenses, aluminum frames, steel pedestals that this array is mounted on. The amount of cost of our system that is represented by our solar cell is well below 10%.
So, the cell efficiency improvements as I said give us the cost reduction, but to reduce the rest of the system cost, we can go out and we can source things such aluminum and steel from a variety of vendors. We can also redesign the system in a variety of ways to take weight out of the system, to take material out of the system. So, I don't want to say those are trivial exercises. There are still engineering that needs to be done, but in order to drive cost out of the system, there is really non-innovation that needs to happen like that needs to be at the cell level. It's really just a lot of value engineering or reengineering for lower cost is what we are looking at right now.
And again to put that in contrast almost roughly, systems, the triple-junction or silicon and thin film, typically see the percentage of their system cost represented by the semiconductor, somewhere up at around 70% to 75%.
Now, I will briefly just go through the Fiber Optic business here before wrapping up. As I mentioned, the two fiber businesses that we have, we play into really two of the major networks that are out there now. One is the HFC network on the cable TV and the video providers. HFC is an acronym for Hybrid Fiber Coax. And secondly, we play into the datacom and telecom side of the network through our digital fiber optics business.
On the broadband or video side, you can see some of the products that we have here. On the left, there are shown some of the cable TV transmitters that will typically sit in the central office or [heading at a] cable company and a variety of other video transmission products. Here on the lower right, this is a product known as a triplexer. This is actually the product that would sit on the side of your house, if you subscribe to the FiOS service at Verizon who have been rolling out for several years now.
In the broadband business, this is a business we have been in for quite a while now. We came into this business through an acquisition we made five years ago. Going back several years, there was a company known as Ortel that essentially started introducing fiber into the cable TV network back in the late 1980. Because of that history, we keep somewhere of about a 70% to 75% market share in this business. So, this business has been a nice business for us and has being growing year-over-year. And because of the entrance of Verizon with their FiOS network, it has really helped this business quite a bit. The landscape here on video fiber optic products for the last couple of years has been the cable TV providers are now facing the telcos who are now looking at video as a major revenue source for them. So, you have the telcos and the cable TV MSOs are both competing for essentially the end users business for video, because of that there has been tremendous amount of CapEx happening on both businesses.
As I mentioned, we supply about 70% to 75% of the fiber gear in the cable side of the business. We are also one of the two major suppliers for the triplexers into Verizon FiSO rollout. So, we like to -- deploying a bit of a cliché, we would like to sell bullets to both sides here. The two of them have really been battling it out and we are supplying the hardware to both sides of this battle which has been good for us.
Quickly on the datacom and telecom side, you can see some of the products here. Again, the datacom side, CISCO is really the major customer; it has been and still is on the datacom market. And this is a business we've been in for several years. We were looking to move into the telecom side of the business which is really a different business, a different set of customers. Here you have the likes of the Fujitsu's , the Alcatel-Lucent's, Sienna's and Nortel. Our entrance into this market in a large way as I mentioned happened at the end of last year. We acquired the telecom fiber optics business from Intel that we announced in the middle of December of last year. And particularly, we acquired was known as our tunable transponder business. This is really, one of probably two sort of high-end niche markets that are less than telecom, where there is still a reasonable product margins out there.
And I won't go through all of the strategic value and the benefits here on this slide. Two things I would like to mention though. We are a completely vertically integrated company. We have our own wafer fab facilities for all of our three, five optical devices, both in the fiber optics business as well as in the solar business. We manufactured all of our own optical dye. Because of this, we can now use all of those optical dye in this telecom business, we get more cost advantage because of that.
Secondly, we also had overlap in terms of the contract manufacturing base that Intel was using, as well as the ones that we were using. Because of that, we are now in a more advantageous position with our CM.
Okay, and just to close. I'll summarize the guidance that we gave on our last earnings call which was just in the first week of last month. Factoring in the additional revenue from the Intel acquisition, which closed in the last week of February, we are looking to close fiscal '08, our fiscal year end at the end of September. We are looking to close with the range that that we're getting from $265 million to $285 million. We also gave guidance for calendar '08 at $360 million. The majority of that growth is coming from the photovoltaic business. Out of that business, the majority is coming from the terrestrial side, as oppose to the space side of the business.
And with that, I'll open up for questions.
Right now our…
Can I ask you to repeat the question?
Sorry, yeah the question was. What percentage of our revenue comes from -- I will answer it this way, what percentage of our revenue comes from of our photovoltaic business? And out of that, how does it breakout between space and terrestrial deployments? If you look at the entire business, fiber optics makes up roughly two-thirds of the revenue. PV makes up roughly one-third. And again, those numbers have moved slightly from quarter-to-quarter. But those are sort of rough ballpark numbers.
If you look at the terrestrial business, terrestrial revenue really started to ship in calendar Q4. And then in the current quarter now, which will end next Monday, this is really the first quarter where we are shipping in, I would say, significant volumes. If we look at the at the photovoltaic business, you would break it out to be -- we don't report on a segment basis by also our business units, but you would see numbers probably in about I think 15% to 20% range in terms of how much of PV is terrestrial, how much is space. Looking forward, those numbers has changed quite a bit, again, most of the growth is coming on the terrestrial side, and not the space side. Although the space side and that business still has good year-over-year growth, just not at the rate that the terrestrial orders are coming in.
Okay the question was, I guess, the first clarification that we are targeting 45% conversion efficiency in 2010. And then after that, what are the long-term goals? So, yes, the plan is that 45% in the next two years, and then after that, there are other technologies that we are looking at. I mentioned the IMM, the Inverted Metamorphic, which we think still may get up to even past 45%. There are four junctions that, I mentioned several of the research groups out there are looking at right now. There are other material systems, be it a three junction or a four junction cell that we are also looking at, as well as other folks are looking at that. The theoretical limit for a triple-junction cell is well above 45%. So, we don't think we sort of pushing the envelope on it right now. You start being limited in terms of the alloy concentration, the particular type of alloy that you grow and the crystalline quality of these different layers. And these are things that are typically refined overtime and processing and crystal growth gets better and better. So, we were comfortable looking two years out, we really don't want to put a roadmap much to pass that, but I wouldn't think that 45% that we are sort of hitting the wall mount and need to have a major shipment technology.
Yeah, the question was O&M or operating and maintenance expenses on CPV versus flat panel. From everything that we have been able to uncover and this is based on talking to folks out there, such as some of the National Labs who have been investigating CPV for many, many years now. Systems that we have had up and running in our plant in Albuquerque, New Mexico for quite a while, as well as talking to some of the other sort of early adopters of CPV. It does not appear that there is any additional cost in terms of maintenance cost over the flat-plate installations. It is something we looked at early on, I think, it's a good question because if you think about it if you are concentrating an image from this size down to the size 500 times small, if there is dust or dirt or debris on this lens and that actually steer a lot of your light off your cell and you see your efficiencies dropped dramatically.
It turns out -- it is not what we see. We've had systems running in Albuquerque, which is up at about 5,500 feet above sea level rain, snow, which had them running through the season. And our systems now are actually putting out 8% over the 25 kilowatts that we expect for. So there will be regular periodic operating and maintenance that has to happen in terms of cleaning the optics similar to flat plate. Above that, we are not hearing that it would be any higher than flat plate though.
Okay. So, the question is -- and let me know if I got this right. How does the cost of one of our triple-junction cells as used in a CPV system compare to the equivalent amount of silicon that's needed to generate the same amount of electrical power. Again, I don’t know what the current prices are just for bare silicon panels, if you look at our CPV system versus an equivalent silicon system and compare the dollars per watt number, given that O&M cost would be same and any of the ongoing costs are equivalent, we are seeing our prices below silicon right now at a dollar per watt level. A ballpark number for one of our CPV cells is again a one square centimeter cell you see numbers in $10 to $15. And again if it's a volume order of cells to supply a 100 megawatts or 100 kilowatts, pricing is obviously quite different. But it's not $1, it's not $100 for a cell like that.
And again, I just don't know street prices for bare silicon cells right now to compare it. And again on a per area basis, just because of the sheer volume of silicon production because of the added complexity of our system and our cells are grown on germanium wafers, it's not nearly the scale of germanium substrates as there is compared to silicon. Our assumption is, we'll always be more expensive in silicon on a per area basis than at the system level at CPV we can achieve those.
Yeah. Thank you.