Very welcome to this, Sivers Semiconductors' Capital Market Day 2023. My name is Anders Storm, and I'm the Group CEO of Sivers. We're gonna have a full schedule for you today with a lot of presentations, and I want to welcome, especially the audience here in that room, but also everybody online. So very welcome, and hope you will have a fantastic day in here. So this is the agenda. We're gonna go all the way to 5 P.M. We're gonna have Q&A in the end of the session, and no Q&A. Maybe some questions during the way. We'll see how much time we have, but the agenda is quite packed. So my name is Anders Storm. I've been with Sivers now for nine years.
I'm the group CEO, and I've been working within the data and telecommunication world for over 25, 30 years, and been part of the sort of turnaround of Sivers and making the company what it is today. Also, we have Lotte, Andy, and Harish with us from Sivers today that will have presentations. And then we have some special guest, GlobalFoundries, with Anthony and Charlie from Ayar Labs, and Pawel from Thorium Space. Both, Ayar and Thorium are customers, and GlobalFoundries is a partner to Sivers, as well as to Ayar Labs. So to give you a little bit of background for you who hasn't heard about Sivers before, the company is, of course, a semiconductor company. We provide chipsets within two business units. One is the wireless business unit, and the other one is the photonics business unit.
We have about 60 customer engagement, ranging from everything from very large Fortune 100 customers, all the way down to very exciting startups in Silicon Valley that, for example, Ayar Labs represent here today. We are addressing a very large total addressable market, which you can sell chips into about $10 billion over the coming 10 years. So this is a very exciting market we've been addressing, mostly emerging new markets, all of it. We have our head office here in Stockholm, and we have also offices around the world. For example, our photonics business is located in Glasgow. The team is 123 employees today, and we have a lot of PhDs that develop these really high-technology things that we are giving to the market.
We are today listed on Nasdaq Mid Cap in Stockholm, with 22,000 shareholders, and we have reached a very important milestone recently. Over the rolling 12 months, we passed $20 million, so SEK 212 million in revenue, and we have had a rolling growth of about 100% the last 12 months as well, which we're very proud of, and we're in a very interesting journey right now. Looking at the investment highlights, I'm gonna take you through five important steps here. First of all, why is this sort of rapidly, sort of expanding markets we're working in? Why is our technology in some way superior, and how could you think about that? Because this is hard material to, to maybe understand.
The sales are now moving from sort of hardware sales into volume sales, and a lot of the sales have so far been sort of consulting or what we call NRE, non-recurring engineering. We have a proven go-to-market strategy here and a lot of very interesting blue-chip industry partners like Intel, GlobalFoundries, NXP, and Renesas, and why do we have that? We have an attractive, scalable business model being within sort of the semiconductor industry, which will sort of reach in product sales, ranges of 40%-60% in gross margin. This is the target for the company. What are we actually doing? The chipsets we are providing is taking the digital world into analog world.
We have chipsets that are what's called RF chipsets, that takes it into a analog RF signal, or we have lasers that bring sort of light into to life. It could be, you know, in a cable or as a LiDAR, or as a radar, or as a 5G, or a Satcom product. Our chipsets is all focused on that side of things, which is sort of analog devices. We have had a very exciting journey over these last seven years, and in 2016, we put in a completely new strategy. We were at that time listed at Spotlight and also called Aktietorget, and we had about $2 million in revenue and a market cap of about $20 million and 1,600 shareholders.
We sort of moved that now to about $20 million in sales. 10x market cap is also more, almost 10x, but $150 million. Shareholders is 22,000, and so forth. During this journey, we've done a couple of very important things. We've sort of created partnerships with customers that is fundamentally important for us, and, and there are companies like NXP and Intel and others that are doing technology around what we are doing that sort of integrates with us to be able to deliver some very interesting and, and challenging technology in, in Satcom, 5G, and photonics. We've also had a journey, of course, for the listing all the way from, from Spotlight to First North and, and now Main Market.
We have been growing quite a lot lately, and that is the fruits that we sort of have from all of the work we provided. It takes some time to get the semiconductor company into sort of volumes because of the long lead times from sort of getting a design all the way to products. But we are now seeing that, and that is what we sort of going to see and focus on the coming three years. If we look where we are today, we have about 30% of the revenue in product and 70% in NRE. We're gonna move that to 80% in revenue in products and to 20% NRE. But of course, at the same time, NRE will be sort of maybe flat, but the hardware sales will increase and grow the company.
How are we gonna do that? First of all, we have the 60 customer engagements. A lot of those, and together with those, we doesn't actually need—we don't really need any more customers. They could give us multiple hundreds of millions of dollars from the existing base we have. Even so, we are actually adding customers, and on a basis, and we see a better sort of pipeline than ever. Just with the customers we have today, we've done a sort of a solid work to grow the company. A couple of those that's gonna come in in 2024, where we already seen orders now, is, for example, in Satcom with the $5 million order we got, now very recently, I think it was August.
And then we also have, of course, the optical I/O, which Ayar Labs will talk about today, which will start coming in. And then 2025, we see even more acceleration in the high-growth market. Also, seeing 5G coming in more. Right now, it's more niche markets for 5G. And then we can also see everything from the Fortune 100 customers and everything growing during these years as well. So looking at sort of the last 12 months and what we've done over the last seven years, we're very proud that we can see that we've been growing now 100%, and the last four quarters has been exceptional, with 113%, 130%, 68%, and 107% growth.
If we look a little bit longer back, we have actually a CAGR of 47% over these last seven years, including the outlook we've done for this year. So what is driving this, and what's behind it? So there are a couple of sort of things that really driven our stuff the last 12 months, and we've taken in about $32 million of orders. We delivered approximately 50% of these orders so far since over the last 12 months, so 50% of this is still not delivered in revenue recognition. And we have a lot of orders that we don't announce normally, but all of these is sort of connected to Satcom, 5G, and our sort of photonics business.
Especially Satcom has been very successful going forward, and I'm gonna talk a little bit more about that. You can see also the marked orders here with the boxes around them. That is product orders that are now coming in, especially Satcom, the last couple of months. In that sense, we've got $6.5 million in orders for that. Just a little bit more on today's guest before I go into the detail. So we have then Charlie, who actually was with us last year as well, and GlobalFoundries. Together, they are sort of representing extremely interesting area now within silicon photonics, where you can sort of make sure to deliver things of high-performance computers in the future, including how AI will be driven in the next generation.
They are delivering together with our chipsets, four terabit of data today, which I have demoed together with first other companies like Intel. The other guest is Pawel from Thorium Space, who is a very interesting company from Poland, who has deep connections into the European funding area and the European Space Agency, and so forth. They are focusing on both the ground terminal, but also on these small satellites that they are working really hard on. You will hear more from him today on exactly what they're doing. This is a customer that Sivers signed up in, I think, March this year on a $1.8 million development contract. The markets, I mentioned them already. We have the high-speed broadband with 5Gs and fiber.
We have silicon photonics, which is sort of driving a lot of the photonics in this. We have the satellite communication. We have consumer sensing, which is with our Fortune 100 customers. Augmented reality, also Fortune 100 customers we're working with, and that's mostly connected to our photonics business. Then we have the autonomous driving for LiDARs, and we could also do radars that we don't do today, for example. This is, of course, high-growth mega trends that will sort of get to an unimaginable sort of TAM in the future. The really interesting thing here is, of course, that this is sort of virgin markets where we're breaking into, so we have not really as many other semiconductor companies today, something to lose here in the market or any big stock out there or something like that.
We are sort of in a different place where we talk about this superior technology. What is it we actually do, and why are we sort of getting all of these 60 customers currently, and sort of even the biggest type of tech companies in the world? Why are they working with Sivers? There is... In our silicon photonics business, it is the marriage between silicon and photonics, and photonics, in this case, are indium phosphide lasers that we make. You can take those small little red dots there and put them straight onto silicon with the sort of patents we have today of marrying sort of the waveguides for the lasers together with the silicon, and that's a big step.
We work together with that, with all of these different companies, including IMEC and others, to make that happen, including companies who actually can pick and place these lasers onto the silicon. That is sort of the technology revolution that is changing how lasers can be built. This is completely new in that sense. The other piece within our 5G business is how we actually can make technology work with steerable antennas, and we make RF chipsets and antennas for that. That is used both for satellites today, because satellites are much closer to Earth, and they are moving ahead over us. They're not geostationary anymore.
So this technology is fundamental, and also in cases where there are track-to-train examples, for example, where we actually could get the trains to pass, and we can beam gigabit speeds to the roof of the train from small base stations along the trains, which we now have been able to do in several countries. And of course, embracing these and enabling these convergence and all the different customers we have here, and many investors always say, "Yeah, but it's hard for me to understand the technology. Do I need to understand all the, all the power or DBs or whatever?" No, I don't think you need to care about that. It's our customers that cares about that, and they are, you know, the biggest Fortune tech companies and everything. And our technologies, we're not a technology start-up.
We are actually very far when it comes to the technology now, and people are deploying it or about to deploy it everywhere. You might not have to think about that. It's our customer who have made the choice already that this technology that we have is working. Here is three examples on things that are semi-working, or demonstrating, or in the market, or just to get to hit into market. These are sort of new emerging markets. Ayar Labs, we'll tell you more about that today, but they are well-funded, as well as backed by Intel, NVIDIA, HP, Lockheed Martin. ALL.SPACE, who does satellite terminals, they are about to launch in Q1, and that's why we got the orders for $6.5 million, and they are backed by SES, Boeing, Inmarsat, the big satellite companies.
We have a smaller company, Airvine, that is backed by sort of the VC world in California, that does 5G things that they just launched last year. All of these things are now coming to fruition, and they are working, and they're sort of in the market. This is another thing we've been working on, which is sort of one of our big projects, and we're very proud of it. We've been working with a Fortune 100 company. They have invested $70 million, or about SEK 170 million now, over five years to get to a sensor. Unfortunately, we can't talk about the name or what the sensor does, but this is for consumer electronic devices, and the sort of the market is huge, and Andy will tell you more about these things.
They have recently got an order delivered to them for 30,000 chipsets, and we're getting closer and closer to the next phase of what they are going to do. How are we gonna move sales from 30% to 80% in 2026? We can see here, of course, that we are sort of-- it's fueled by product sales, and it needs to be product sales. All of these companies I'm talking about, they're going from, you know, development into products, and they're all at, at sort of a point where, where that's gonna happen more and more. Here you can see, you know, the quarterly, take-up and, and so forth.
If we look at just the hardware now year-over-year in 2022, 2023, we actually grown the sales by over 100% when it comes to the hardware sales. Also in actual money, it's grown. Even if we've grown the company 100%, we're also growing the hardware quite a lot. That's a very good thing about that. If we look at more in detail, how will that happen now near term with satellite communication? For example, one of the products we have is one of these beamformers for satellite communication, where this product has been finished, and we got the first volume orders now from ALL.SPACE.
These orders are now coming out in the market, and they give us the orders to build these terminals, and the terminals are still just in fairly low volumes because they are ramping up. This growth will be sort of 2x-4x, to 3x-5x, to 3x-4x type of growth going forward in that sense. You can take the numbers we have now and then see them grow year by year, and that is a very, very large market and a very high content of chipsets in those terminals that we are selling. These are sort of forecast by the customer. We also signed an extremely important contract with the same customer in December last year that we're delivering on.
It's a SEK 16.4 million, and there's two other chipsets, which is not sort of infringing sort of on the chip we're already doing with them. It's new chipset that would add value in the terminals and give us even more money for each terminal they sell. The plan is to have that ready by 2024, but we're delivering a lot on that right now, and that's also supporting the growth in the company. And how much that will grow in 2025 and forward, we don't know really, but the most value is actually in the product that is already now in, you know, in volume. Thorium Space, we've done the first development project. We're about to do the second phase now on that project as well, which will then lead into new volumes in that sense.
We have another customer, which is a Satcom company, where we're doing these things as well, or a satellite owner. There is a lot of things that are gonna happen over the coming three years in this that will support this heavy growth that we see in front of us. Another thing that is more niche is sort of the track-to-train applications that we have. They are now deployed and getting deployed more and more. They're deployed in the U.K., they're deployed in Spain, and we're now seeing also a deployment with Caltrain in California. These things are, of course, even more exciting because they're sort of the first carbon neutral communication system ever, and they will be that already in 2025, actually.
Promoting this, and hopefully, we're trying to promote this in Sweden as well, so we soon can get this to Sweden, and this will give you gigabit speeds with 5G onto the trains. Another thing that is really fueling everything in the future is generative AI, of course, and we were able to get an article into Wall Street Journal. You have the link there, you can read that, and that's together with Ayar and Intel and others in there. We've also written an insight article about how generative AI will drive and need to have these optical I/O solutions, which is actually changing the way computers speak inside from electricity to light. That's where our photonics business comes in, and it's going to be very vital to be part of that in the future.
The proven go-to-market strategy is based on that we have a lot of partners in the ecosystem, ranging from Intel, NXP, Renesas, GlobalFoundries, and all of those. We've done a very fundamental photo work together with all of these to be able to deliver our technology into market, because as a semiconductor company, you cannot stand alone with just a, a chip somewhere. You need to have all of these partners, and this has been a fundamental part of our strategy. That sort of makes it into the next phase of sort of the, the sort of business model. Why is it so attractive to, to have this business model? In our wireless business, we have the-- what's called the fabless model, where you don't have sort of your own fabs, a very light when it comes to CapEx and so forth.
We have GlobalFoundries and other partners. We are sort of designing, and they make things from our design. In our photonics business, we've had our own fab, and we're still going to keep our own fab, but we're also seeing a big, big flow of new products there. We've been looking at, and Andy's going to talk more about that today, but we also looked at something we call hybrid fabless there, where we have our fab, but we're also being able to outsource a lot of that to reduce the CapEx investments quite heavily. All of this flexibility gives us sort of a light manufacturing strategy in whole. Also, to understand the ecosystem within semiconductors, it's easy to put them up like this and see what other companies are doing.
Some of these are our competitors, some are not. In general, in semiconductors, the companies have something like 30 to 70 or 60%+ in gross margins. Sivers' target is to have the 40%-60% here, depending on volume, depending on products, and so forth, and depending on if it's NRE or if it's product sales. This is for the product sales target, we have this. Summing up a little bit, we have a very strong Nordic and EU active funding base. About 20% are institutional investors and long-term only. A lot of pension funds, as well as Swedish governmental pension funds like AP2, AP3, AP4. AMF is here today as well.
We have Swedbank and other funds that as part of Sivers and have been part of Sivers for quite some time. We're very happy for this strong support we're having in this space. We also recently got the Swiss company investing in the company before. Also, we're looking, of course, to widen this space for the future and, in these kind of events, marketing ourselves. Of course, one of the things is to get into the U.S. and get U.S. investors to look at the company that also understands the semiconductor industry quite well, and we have had a lot of meetings during the last year around that as well. What did we say here in 2021?
We said that we're going to have a strong organic group, keep on doing what we're doing, but add more. What did we do? Yeah, we kept on growing: 46% in 2022, 100% now over the last 12 months. We have been adding verticals in this Satcom, optical I/O, and so forth. We have increased our partnerships, but we also acquired MixComm, which was sort of evaluating that and getting to sort of a, a add- adding verticals per that. That has been, I think, very successful on the plan and the strategy we have. What's going to happen forward? We will look at that. We'll keep on building. As you can see, we now have even more partners.
We want to build, but we also now really need to move in to deliver on product sales as well as other verticals, and moving from 30% product sales to 80%. Also evaluating if there is sort of other interesting verticals that are sort of cash flow positive that we can add in via M&A or not. This is sort of the overall picture that we see in the future, that there's a lot of things that we're doing right now, and we want to keep doing that, of course. Summing this up, rapidly expanding markets. Yes, we are in six, six mega trends that are enormous, that we have a huge TAM, and we'll keep on building on our product sales into that.
Superior technology, and, I mean, we're very proud of the customers we have, and a few of them we present here today, but there's very big markets that we will be able to sell in, and they've chosen our technology already, and the technology is mature. It is not sort of something that might not happen. It is there. Sales moving to production volumes. Yes, we can see that we are moving more and more, and we're seeing customers coming to fruition and coming to market, so we're quite confident that we'll be able to move that. In 2026, the target is to get 80% of the sales from, from hardware. A proven go-to-market strategy, we have great partners. We had Intel and NXP with us in Mobile World Congress in Barcelona this year.
We have GlobalFoundries here today, which is sort of a large foundry, and all the other customer engagements... and an active scalable business model. As I said, sales growth have grown 100%. We passed $20 million in revenue. We have product and gross margins that we see that is easy to get to. Scalable assets in the foundries that we have, that we can actually grow the company without too much CapEx, and now also, trying then to end the year with a very positive 2023, where we have a plan to, in our outlook, to get to positive EBITDA. That's a really important part here for what we're doing. That is the investment highlights. I want to thank everybody now.
I'm coming back later for Q&A, and with that, I want to introduce our CFO, Lotte Saks.
Thank you, Anders. Sivers is a company with fantastic opportunities, and our job as management is to turning that potential into substance. As CFO, I'm really happy to provide the Q3 update, where we see strong current momentum in both top line and profitability level. To start, just a recap of the financial targets that were defined for 2023. Net sales growing 100%, a little bit variation over the quarters, given the nature of the business, and turning EBITDA positive during second half. How are we doing so far? In Q3, we had a revenue of SEK 58 million and a growth of 107%, and the Wireless business, a staggering growth of 210%. The Photonics business has a stable and strong growth of 38%.
If we look at year to date, we reached a growth of 96%, which is really close to the financial target of 100%. To Anders' point, looking at the past four quarters, we are at 100% growth. One of the strengths of Sivers' business model is the diversified sales into multiple segments. Photonics is growing in optical I/O and data centers, and, you know, also in a good position for continued growth in the consumer sensing. The strong growth that we see now in wireless is driven by the increased demand in the Satcom business. Another strength with Sivers' business model is the combination of development projects and product sales, where product sales provide the resilience and the ability to scale.
That's why it's so important, and we're so happy to see in Q3 that we confirm an increased sale of products, both in terms of increased share of sales, but also the growth rate, 109% growth in the product segment. We are growing fantastically in the development part as well, and the reason for that being primarily that we reallocate our internal development resources from last year, more working on developing our own products, and now we're focusing on customer-funded development. We're focusing a little bit our development resources. This is a really important key strategic signal in the quarter that we're able now to step up in the product sales.
Our third view of sort of revenue segment is looking at the geographical distribution, and also here, we have a good split between the regions of Europe and North America. Right now, the growth is coming from Europe, and that's relating to the Satcom growth in the wireless business. In Photonics, we're still growing in the U.S., so year to date, we had a 35% growth, or SEK 13 million. Zooming out a little bit, looking at the sales distribution, this is something that is really important, that we have these balances. We're not a one vertical, one region company. We have a portfolio that balance and offsets risk in our business. We have the two business units, 60/40 split, quite balanced.
We have the revenue type split, really important, and the signal that we now reach 31% share in, in third quarter. As Anders mentioned, this is a clear strategic target that we want to grow this to 80% or even more. We have the geographical split, Europe right now being stronger than North America, but still a good balance. Revenue and growth, fantastic, what about our ability to also turn that into increased profit? Looking at the quarter's year-to-date, Q3, as we said, 107% growth in revenue. EBITDA grew 57%, and adjusted EBITDA, 46%.
This is really important that we have a balanced investment in our future growth, so that also will let some of the increased revenue improve our profitability level. We are, at the same time, I would say, maturing the operation, investing in the operation, and building our capabilities and processes so that we are prepared for the coming scale-up in product sales. Really good pattern so far this year. Just to recap that, strong momentum, 107% growth, growing in all segments. Year-to-date growth close to the financial target, improving EBITDA, and increasing the share of product sales. If we put the year-to-date and the quarterly performance in perspective.
On the one hand side, looking at the business unit proportion of sales development, and then, on the other hand, the revenue type development. We have really strong growth in both business units from 2021. As you can see right now, Wireless is growing at a faster pace than Photonics, also increasing the share of total revenue. Coming from, in 2020, a situation where Photonics was much stronger and much larger part of the total, now it's a more balanced business in the company. Product sales, as we said, is growing. The share is still a little bit too little compared to where we want it to be, but again, we see an improvement now in the quarter.
To that point, I would say this is probably the most important sort of financial KPI that we can be tracking. This is the quarterly sales from 2021. The first thing you can see is it's clearly a step change for the past four quarters in the revenue level, coming up from around SEK 25 million to now SEK 50 million and above. As we want to point out here in Q3, picking up the share of product sales again to above 30% from a few quarters of a lower share, maybe even more important, the absolute value, SEK 18 million. If we look, it's the highest revenue from all quarters that we have been looking at from 2020.
Zooming in a little bit in the segment reporting, and the Photonics business. As you can see, strong growth in 2023. On EBITDA level, we haven't really been able to turn that into an increased profitability. I would say that's primarily driven by the nature of the business, that we have a factory, we have operations, and we are now preparing that for the increased scale-up for commercial volumes. We are really well positioned, I would say, for what's coming next here. If we look at wireless specifically, here, it's a more clear view with a really sharp revenue increase and as sharp improvement in the profitability level. As you can see, basically halfen the loss that was made in 2022 to the last 12 months now in 2023.
I would say in both business unit, there is a work ongoing of improving operations and making it more stable, including implementing ERP systems. In a small company like ours, that's-- I mean, it's a big thing, but it's not a huge risky project that it can be in other companies. This is quite small scale, but it's still something that really is good and important and prepare us for the future. Another thing being, I mean, a growth company, of course, it's the access of capital and how we work with our cash flow and funding.
Given the sort of market and access to capital, we have taken a view now to, as we said, focus more on customer-funded development rather than maybe in the past few years we have done a lot of development of our own products. So the capitalized R&D has come down, and also both in absolute values and in percentage of sales, and that's a clear priority that we're making right now. And also in the total CapEx, we're keeping on a sensible level, and I would say the CapEx that is non-R&D is also, to a large extent, driven by customer products. So it could be investments in IP that we need for the customer project, for example. And this area is something that, of course, it is absolutely central to us.
That's why also Anders mentioned we're now looking into not having to invest massively in new production sites for the Photonics expansion, but really looking at a hybrid solution, which is much less capital intense. Again, looking at our financial performance in perspective, growing revenue, improving EBITDA, and most importantly, probably, we're reducing the burn rate and really showing, I would say, a dramatic improvement now in our cash flow consumption in the last periods. Creating sort of a palette of our important financial KPIs and how these curves are looking, we are quite satisfied now with, as I said, the current momentum that we have. Really strong growth curve pointing the right direction, the revenue in absolute value also increasing.
The product share, in percentage, again, in Q4, that in, in Q3, it's kicking back up. In absolute value, which is not shown here, we're definitely satisfied with the improvement. EBITDA also turning in the right direction, and the cash flow curves, as well, are improving. To summarize that up, again, last twelve months, 100% growth. The company has a balanced and diversified sales distribution, really important. We are improving our capital efficiency. Capital efficient, profitable growth is really our, what we're focusing on. We're really planning our R&D and where we spend our resources. The target now, the next step is to become EBITDA positive, and obviously, after that, is to scale up the business and also become cash flow positive, and we are on a good path towards that.
Again, capital efficiency, a flexible business model, and we're looking really with positive feelings for what's ahead of us. Thank you.
Thank you, Lotte. We are a couple of minutes ahead of time. I hope we already have Andy McKee online, and we're going to go into our Photonics business that Andy is driving. So here we go, Andy. Over to you.
CTO, and also the managing director of the Photonics team over here in Scotland. Next slide, please, Anders. So today we're going to sort of talk about the Photonics business unit and try and give an overview of the markets that our products go into and you know why we have a competitive edge in those markets and why our customers choose to work with us. Maybe just quickly go back to the agenda slide, please, Anders, which is slide number two. So we'll go into a bit more detail about the markets that we participate in, not get too technical, of course. That includes Silicon Photonics, a very important area for us, Optical I/O that we've touched on already in the presentations for generative AI applications.
Wearable technology, LiDAR, two very important areas for us. Talk a little bit more about the foundry capacity expansion project that we've embarked on over the last six months, and then finish off with a summary. Okay, next slide, please, Anders. Just a sort of top-level view of Sivers Photonics. We are the most advanced supplier of custom III-V semiconductor lasers. We are based over here in the U.K. We are a relatively well-established business. We've been working for more than 20 years. We're very well known within the industry, both designing and manufacturing a broad range of III-V photonic devices. As Lotte just mentioned, you know, we are slightly different to the wireless team in the sense that we have our own fab here in Scotland.
We have a world-class 100 mm indium phosphide wafer fab with a potential capacity of 5,000 wafer starts per year. Today, we offer to our customers a complete end-to-end chip solution from design through NRE development, prototyping, qualification through to volume manufacturing. That's really where we sort of find a niche in the industry today. We're almost unique in terms of offering that full service across the supply chain. Today, we have 82 members of staff based here in Glasgow. As Anders mentioned, we are a key strategic supplier to many Fortune 100 companies and Silicon Valley startup customers. You know, to that end, I think around about 75% of our sales are based in the U.S. Historically, we have produced a very broad-...
range of devices within the fab, but now we're very much focused on indium phosphide-based lasers. And those lasers are going into all of the applications and markets that we'll talk about throughout the remainder of the presentation. Next slide, please, Anders. Okay, so just some highlights over the past 12 months on the photonic side. So we are sitting at year-to-date revenue of GBP 5.1 million after Q3. That's a 26%, you know, strong organic growth from last year. In terms of the orders pipeline, this year, we're gonna exceed financial year 2022 and all previous financial years, so that's a very sort of strong trend on the orders front. In terms of customer highlights, we have now started on a sort of second development project with Ayar Labs.
That's a $1 million development project working on next-generation laser arrays. We have a follow-on order with our U.S.-based sensing company of $1.3 million, where we're continuing to work on a partnership with them to develop sort of sensing technology. Of course, we have our historic largest customer, the Fortune 100 customer, where we're entering the sort of next phase of the program of development, and we've got year-to-date revenues of GBP 3 million with those guys. Over the last six months, we've also kicked off this foundry capacity expansion project that we've touched on already. You know, that really is evaluating potential partnerships and options to really sort of set up a sort of hybrid manufacturing capability expansion in anticipation of the high-volume production that we have visibility and forecast of already.
Next slide, please, Anders. Okay. Today, you know, Sivers Photonics, we are a sort of critical technology company, providing advanced laser sources for next-generation silicon photonic applications. We are developing partnerships with world-leading silicon fabs, including IMEC and GlobalFoundries, to provide customised laser solutions via established process design kits, that's PDKs. We are supplying advanced laser arrays for high-performance computing, next-generation artificial intelligence clusters and applications, and that's through our partnership with Ayar Labs. We have a five-year ongoing partnership with our U.S.-based Fortune 100 customer, which has generated to date, $17 million in NRE revenue. We see sort of rapidly expanding addressable markets across the optical comms space and optical sensing, with a combined TAM of more than $5 billion in the 2027 timeframe. We're developing partnerships to support the high-volume production to that end.
Okay, so quickly discuss the four, the four main markets that our products are going into, today. So at the top, we have obviously the, the AI and machine learning, sector. So we are supplying high-power DFB laser arrays into this market. Those arrays are being assembled into innovative, high-speed interconnect modules, Ayar Labs being a perfect example of that. So, so these modules are really sort of driving the sort of next generation of sort of cluster design that companies such as NVIDIA are deploying into the, the market just now. So that's a very exciting area to be working in. Of course, we also work in sort of wearable technology. That, again, you know, we're supplying high-power DFB lasers into this area. When we say wearable technology, we're talking about watches, glasses, headsets, hearables.
And this has very much got a sort of biometric consumer, you know, healthcare aspect to this. So you can imagine, obviously, the market size in this particular area, and how that may be applied by some of the big companies in this space. We've then got the more sort of traditional data center space, so this is supplying, again, high-power lasers into pluggable transceiver applications. Within the data centers today, you know, deployed by the hyperscale guys, then this is typically sitting at 200G line rates. That's progressing up to 400G, then 800G, and eventually 1.6 terabit line rate. So we are supplying devices that can enable those transceivers to move up to higher data rates. And then we have the sort of 3D sensing and LiDAR market.
You know, ultimately, that's driven by the autonomous vehicle market. That's by far the biggest end market for this application, for this technology. Again, we're supplying high-power devices into this market. We see some of our customers looking at sort of different, you know, more short-term markets to gain revenue, but it's a very exciting space for us. You know, ultimately, we have access to sort of very large TAMs across multiple end markets. You know, it may appear that we're quite thinly spread and lack focus, but that's certainly not the case. You know, we're delivering fundamentally the same lasers into different end markets. Really, the sort of key takeaway from that slide is, you know, the TAMs are huge for indium phosphide lasers, and all of these markets have the same laser requirements.
You know, they're looking for high-power lasers, high precision, high reliability, of course. And often, the lasers are delivered in, you know, an array format, and to that end, we've got a very strong competitive edge in these markets. So a few slides on Silicon Photonics, just to put that into context, you know, where a lot of our commercial activity is driven by. Next slide, Anders. So Silicon Photonics is basically leveraging the sort of existing, well-established, high-volume, low-cost manufacturing of the silicon CMOS industry to fabricate photonic ICs. Now you can actually achieve a lot of optical functionality on a silicon chip, you know, such as routing, muxing, demuxing, modulation, and also detecting. But critically, you cannot create light on a silicon chip, and that's where our laser technology is fundamentally required.
Without III-V lasers, silicon photonics would not exist today. So basically, our lasers illuminate and optically power these circuits. Next slide, please, Anders. So this slide is really just a snapshot of the companies that are publicly working in silicon photonics. You know, you can see, for example, we have Meta, Microsoft, Google, Amazon, Apple, Cisco, you know, some very large companies, of course, in this area. Also, the foundries, you know, TSMC are active in this space. We have GlobalFoundries in there, who we'll hear from later in the session. And we are very much one of the sort of key enabling tech companies in this whole silicon photonics ecosystem, you know, delivering our custom laser devices. Obviously, you know, we've got multiple fast-growing markets are supported by this technology.
You know, data comms, optical I/O, the wearables, and the LiDAR market, all very heavily dependent on the sort of evolution of the silicon photonics ecosystem. So this slide really visually represents how our lasers are used in silicon photonic hybrid integration. So we are the leader in this technology and have sort of granted patents protecting our IP in this area. Now, we have multiple commercial customers using this integration approach. It's by far the most established integration technology and the most flexible. So what we can see on this chart here, on the left-hand side, we have a gel pack of indium phosphide lasers. So we are fabricating these sort of little red chips. We're then shipping these chips out to our customers.
So the customers then manage the assembly of the laser chips onto the fully fabricated 8-inch or 12-inch silicon photonic wafers. That's done by typically sort of outsourced partners, guys like Fabrinet in the optics industry. Sanmina would be another example of that. So they're using sort of very high precision die-bond placement technology to put our chips very accurately down onto the silicon photonic circuits. The positioning is extremely critical, but that determines the optical coupling efficiency between our laser into the silicon photonic circuit, so that's really a sort of critical part of the process flow. On the right-hand side, you can see our laser chips, which have been flip-chip assembled onto the silicon photonic platform. So we have an ongoing development partnership with IMEC. We've been working with them for about five years.
You know, we've got further developments that will be published in the short term, and it's been a very sort of successful project with us and has generated sort of commercial activity, you know, based on the publications that we've made with them. Next slide. Again, we've spoken already about developing closer partnerships with the silicon photonic foundries. What we have here is a sort of non-exhaustive list of those foundries. Today, our lasers are already being assembled into all of the main silicon photonics platforms, but we need to develop an easier portal, you know, through the existing silicon photonic PDKs to make it easier for the customers to access our laser technology.
We're actively doing that today, with the assistance of, you know, PDK software companies such as Synopsys, a very large company in the sort of semiconductor industry. Okay, optical I/O, let's talk about that. So if we dive into this, this space, we can see here, again, we've got a very fast-growing market size. This is driven by the sort of ongoing displacement of copper by fiber, in the network. And that, that's due to a number of reasons, including sort of bandwidth requirements, signal integrity considerations, and also power consumption, efficiency. One of the most interesting numbers on this chart is the large percentage of the BOM, the bill of materials, for these optical modules.
So up to 50% of the BOM cost is the laser itself, so clearly, it's a very valuable component within these systems. Okay, so this chart here really talks a little bit more about the sort of generative AI landscape and where it's sort of headed. So these large language model complexity clusters are... They're really increasing at a computational rate, which is outpacing traditional optical pluggable I/O technology, even up at 400 Gb. So Meta have actually recently, you know, publicly stated that GPU, GPUs in these clusters can be sitting idle for up to 33% of the time, awaiting on input data. I mean, it's a staggering statistic.
So really, optical I/O technology, therefore, needs to show a sort of step improvement, perhaps an order of magnitude in bandwidth, to support the requirements between the GPU accelerators within these large language model clusters. And Sivers lasers are being deployed into these innovative high-speed optical modules operating at 4 Tb and above today. So this chart shows the sort of overall AI landscape. We have the OpenAI software tools on the left-hand side of the chart. A lot of publicity about some of those companies today, through to the laser suppliers on the right-hand side, where we sit. Now, clearly, NVIDIA are at the heart of the overall hardware supply chain here. And with our existing partnerships, we are very well placed to sell into that supply chain, so that's very exciting for us....
Now, on the right-hand side, we've only got three laser companies listed here. We've got ourselves, we've got Lumentum, one of the big incumbents in the industry, and also MACOM. So we believe there's very, very few companies that are actually capable of producing these very high-precision laser arrays for the four terabit optical I/O solution. So we're in a very, very strong competitive position, particularly within this marketplace. So we've just got the one slide on the Ayar Labs activity. I'm sure Charlie will cover that in a lot more detail later on in the session. But what we fundamentally supply to them is an InP DFB laser array. So each laser is individually addressable. Each laser emits at a different frequency.
Ayar Labs then assemble those into their SuperNova light sources. They mux the wavelengths together, then split them into each individual fiber. So there's 8 fibers coming out of that module. Each fiber's got all 8 frequencies as a carrier signal, and, you know, this is the first product based on the CW-WDM MSA that has been produced so far. So the output from this SuperNova module goes into their TeraPHY module, which then can produce this sort of bi-directional bit rates and links of up to 4 Tb per second. Okay, so I mentioned the CW-WDM MSA. So we were actually invited to be one of the founder and promoter members of this MSA. We managed to influence the standard that's now been defined and issued.
And actually, all of our array development activities today are based on this standard, so it's been a very sort of powerful process for us. Now, you can see on the left-hand side some of the other large companies such as, you know, the incumbents in the industry, such as Lumentum and Coherent. They were also involved. And then we've got some very large companies on the right-hand side as observer members, like NVIDIA, Broadcom, et cetera. So, you know, this is really a sort of a very sort of strong endorsement of our capability and our standing within the industry, that we were invited onto this consortium, and it's given us a lot of sort of very positive publicity over the past, sort of two years.
Okay, in terms of wearable technology and LiDAR, with wearables, again, you know, again, we see the sort of same strong growth predictions. We're defining wearables as watches, earphones, AR headsets, glasses, and so on. You know, ultimately, lasers are at the heart of these advanced sensing technologies. So again, we're in a sort of very strong position to drive growth in these areas. This is a slide that Anders has shown previously, but let's just go through this again because it's important we sort of reiterate the position that we are in with our largest customer. So, within the past six months, we've delivered in excess of 30,000 devices, you know, for qualification and system-level testing.
So it's a very exciting, you know, juncture in the project. This has been a very successful project over the last five years, you know, working on both custom lasers and detector technology for sensing applications. And, you know, it's, you know, at a position of sort of critical position in the program where we're awaiting feedback on the system-level test to define the next steps of the program as we move forward into next year. And just to reiterate, the RFP numbers that we've supplied, so that's request for pricing, you know, ranges from the sort of $30 million per annum right up to several hundred million per annum revenue, that we've submitted to the customer.
So it's a very exciting project and a very exciting opportunity for us going forward. Okay, we move on to LiDAR. So again, we can see a fast-growing market for LiDAR. Our customer base is exclusively focused on FMCW LiDAR, which can achieve the most demanding requirements and meet the hardest specs in the industry. For example, what they want to do is develop technology that can detect a brick on the road at 300 meters in front of the vehicle in darkness or very bright sunlight conditions. And this actually can only be achieved using very high-precision lasers with narrow linewidths and long coherence length. So we can supply those today, so again, we're in a very, very strong position to address this market going forward.
The LiDAR applications, of course, autonomous vehicles is the key application for the technology with the biggest market by some distance, but that's still some years away. Most vendors, you know, i.e., our customers, are still looking for other markets to generate short-term revenue, and those are listed above, e.g., sort of robotics, logistics, and 3D mapping. So again, a very exciting area for us. Just a couple of slides towards the end of the presentation here on the sort of foundry expansion project. This is something we kicked off around six months ago. So we've been engaged with a number of prospective partners, you know, to secure sort of high-volume wafer fab processing capacity. So just to put this in context, you know, we're currently processing just under 1,000 wafers per year through our fab.
We see volumes well in excess of that, from the markets that we've previously discussed today. We have a shortlist of three companies that we feel can support the ramp-up in volumes that we're predicting. Each one has particular strengths, of course, and, you know, we have two US-based companies. We have one Taiwanese company. We're evaluating the detailed technical capabilities and capacity of each of those, with a view to taking that down to a shortlist of two partners, where we will do a sort of technology qualification in the first half of 2024. So that will put us in a very good position to sort of realize the demand, support the demand of the volume ramps that we see across the markets, from sort of 2026 onwards.
This slide sort of schematically shows where that would really take us. So on the left-hand side, you know, we currently outsource our epi wafers. That is our scalable supply chain, moving out into the future. So what we do is then focus on steps two and three. So those are the key proprietary steps in the process flow that we do today, and we will scale up this fab in order to support with some very modest CapEx investment. So really, the hybrid outsourcing will be supporting stage four in the process. So the outsource partner will be supporting the wafer fab, test singulation, and potentially even shipping the products out to our customer.
In terms of capacity, you know, we're looking at putting in place a capacity which is 40 times the current run rate of wafer fab that we actually process today. Basically what we're doing here is, you know, setting up a sort of high-volume supply chain, but also maximizing the productivity of the existing site that we have here today, you know, focusing on the sort of key, proprietary steps. My final slide, just to summarize what we've spoken about today. You know, we see a sort of potential $5 billion TAM across the datacomms and the optical sensing markets. You know, we are a critical technology partner to some of the world's biggest tech businesses.
The Indium Phosphide laser arrays are key enablers for next-generation technologies and applications. We are the world leader today in custom laser chips for the silicon photonics ecosystem. Ayar Labs and Sivers are developing, you know, innovative optical I/O technology that will revolutionize high-performance computing and further release the potential of generative AI in the future. We have an ongoing partnership with our large Fortune 100 customer that's delivered five-year revenues of $17 million. We're developing closer silicon photonic foundry partnerships, which will accelerate scalable, you know, PIC solutions into high-volume manufacture. And the high-volume foundry capacity expansion project will significantly increase our production capabilities to meet the demands that we see going forward. And with that, I will close and hand back to the floor and to Harish.
Thank you, Andy, for that deep dive into our photonics business. And now it's time for Harish Krishnaswamy, who is running our wireless business. Harish is also a professor at Columbia University within electrical engineering, and I'm handing over to you, Harish.
Thank you, Anders, and it's very nice for me to be here to represent Sivers Wireless and tell you a little bit about how our business has been doing over the last year. If we could move to the next slide. Sivers Wireless today, especially through the acquisition of MixComm in February of 2022, has the broadest millimeter wave portfolio on the market, and I'll tell you a little bit more about that as we go down the slides. This broad portfolio that also has the highest performing ICs and module solutions on the market has enabled us to have a very strong pipeline of design wins across our three verticals, namely Satcom, 5G mmW ave, and 60 GHz. We're very excited to see that our lead Satcom customer is ramping to volume production as we speak.
We've received several orders that are in the millions of dollars in value and millions of chips in quantity, that we are delivering on as we speak, and we have several more customers behind that we expect to ramp to volume production over the next couple of years. Also critical is, of course, our partnerships with key players in the ecosystem, and these include GlobalFoundries as the leading specialty foundry for the types of chips that we manufacture, but also Intel, NXP, Renesas. These are really blue-chip ecosystem partners that enable us to go to market in these different verticals and serve these design wins. And all of this is being reflected in our revenue growth over the last four quarters. So over the last four quarters, trailing twelve months, we have booked revenue of...
We have shown revenue of SEK 122 million, which represents 272% growth over the previous four quarters, which is certainly impressive growth. But we also have order bookings of the order of SEK 270 million, which guarantees future growth for 2024 and beyond. We go to the next slide. So, you know, what do we do in Sivers Wireless? So we serve the need for high-speed broadband wireless everywhere, which is something that we see growing within society without any bounds. Every year, the wireless capacity that society needs grows exponentially, and millimeter wave uniquely has the ability to serve that capacity. There are three different verticals through which wireless can be delivered to society, and that is through the 5G ecosystem, 5G mmW ave FR2.
It can also be served through the unlicensed 60 GHz band, which is sort of part of the Wi-Fi ecosystem. We also see non-terrestrial satellite communications beginning to serve that societal need, and we're active in all of these verticals. Really, what enables us to be active in all of these verticals is our foundational underlying technology. All our solutions are built on high-performance silicon millimeter wave beamforming technology, which equally applies in each of these verticals. While we have different products for these different verticals, they all are based on the same underlying technology, the same highly differentiated IP, and the same skill set that we bring to the table. The other mega trend that really supports our business is the resurgence of interest that we are seeing within the semiconductor industry.
We've seen that across the world, whether it's in the U.S. or in Europe, or in India, or in, you know, in Japan, for instance, there is a recognition that semiconductor manufacturing needs to be supported at the federal level, that, you know, governments and countries need to invest in semiconductor manufacturing as well as chip design. We're seeing a lot of investment through the form of these CHIPS Acts being made into the semiconductor ecosystem. We are already benefiting from this investment, both in terms of a strengthening of the ecosystem more broadly, but also in terms of funding directly into Sivers Semiconductors. You know, this is enabling us to address a very exciting market. This is a view of the total addressable market for Sivers Wireless in the millimeter wave space.
We see that today's Satcom in 2023 is a $400 million market. 5G millimeter wave has been a little slower to materialize than what folks had anticipated and is currently roughly a $50 million market. Both are projected to grow, and we're already seeing evidence of this, not only through market reports, but actually through engagements with customers, at a compound annual growth rate of 53% for 5G and 15% for Satcom, reaching levels of $400 million for 5G mm Wave in 2028, and $800 million for Satcom. In 2028, basically five years from now, we will be addressing a $1.2 billion market, which is a very exciting opportunity, given our positioning within the competitive landscape.
This exciting technology is summarized on this slide here. We have the broadest millimeter wave portfolio on the market. We have beamforming ICs for the 5G mmW ave space. You see that on the top left. We have similar beamforming ICs, both bespoke as well as broad market chipsets for Satcom in the Ka-band. We have highly integrated RFICs, as we call them, for 5G as well as 60 GHz. We do more than just chips, we also work on modules in which this chip is integrated, and we find that this approach of building modules around our chips both allows us to increase the value creation that we're bringing to the table, but also allows us to accelerate the time to market for our customers because they're able to take our modules and immediately put a system solution around them.
And so on the system or module side, we offer antenna modules for 5G mmWave and 60 GHz as finished products, but we also offer them as reference designs for base stations, CPEs, as well as repeaters. And we also work on algorithms that boost the performance of the overall system that would be designed into modem solutions offered by our partner. And in this regard, this broad portfolio is really unmatched within the competitive landscape. So on this slide, we're looking at a comparison of Sivers Semiconductors, Sivers Wireless, to a variety of our competitors on the market. This includes very large semiconductor companies such as a Qualcomm or an Analog Devices, as well as smaller, privately held companies such as an Anokiwave or a Movandi. And you can see that other competitors are able to address a portion of the portfolio.
Some have beamforming ICs, while others prefer more integrated RFICs. Many of them do not do modules, many of them are not active in Satcom or in radar. Most are not active in repeaters. Given our ability to offer customer solutions across this entire, you know, horizon of offerings, we find that this enables us to get to design wins and offer full solutions to customers in a significantly advantaged way compared to the competition. But it's not just about the breadth of our portfolio, each individual product is also the top-performing product within that category. So on this slide, we're comparing the output power and the efficiency of our beamforming ICs to competitors on the market, and you can see these competitors being color-coded here based on the technology that they're implemented in. We have a couple that also use RF-SOI, similar to what we do.
We have a couple that use silicon germanium, and a couple that use bulk CMOS. Across the board, we have 10 times higher output power, as well as 3-5 times higher efficiency than our competition, which leads to significantly better products for our customers. And this is something that is not just a product of our own competitive analysis, but we have seen this validated in print by our customers as well. A little bit about how we serve the market and what the ecosystem around us is. So we are a fabless semiconductor design company, so we design chips, but these chips are manufactured at external semiconductor foundries. We are partnered with IHP as well as GlobalFoundries. GlobalFoundries being the leading specialty foundry for the types of processes that we use.
After manufacture of the chips at the foundry, they would typically be tested to ensure that they work, as well as assembled into a package. This test and assembly process is typically done at overseas assembly and test houses, or OSATs, as they are called. Here again, we're partnered with the largest OSATs in the ecosystem, whether it's Amkor, ASE or JCET. Our foundry partner, GlobalFoundries, actually often handles test and assembly in a turnkey fashion. Once tested and assembled, our chips are then sent to the customer, and the customer would handle the integration of our chips into the final end product, followed by selling off those products to the eventual customer, who would essentially consume the product....
So if you look at this slide, this really summarizes our various ecosystem partners who are all critical in various aspects of our go-to-market strategy. So you see, of course, several big ecosystem players here. As mentioned before, GlobalFoundries, Intel, NXP, Renesas. Many of these GlobalFoundries, of course, is our foundry partner, but the engagements with Intel, NXP, Renesas, as well as WiSig Networks as an emerging startup, MaxLinear as well, tend to be engagements with companies that can provide a modem solution, which when coupled with our RF solutions, can offer the full system. But we also have other very important partnerships. Richardson RFPD is the global distributor of our components across the world. Rohde & Schwarz is our test partner for in-house testing, et cetera.
And so all of these, partnerships, as well as products, have really led to the tremendous growth that I mentioned earlier. As I mentioned, looking at the last four quarters, we are seeing generally across the board, nearly 200% growth over the previous four quarters, and aggregated, as I mentioned, we have 272% growth for the trailing 12 months. But not only that, we have an exciting pipeline of wins that builds revenue for the future. So if we go to the next slide, this is a summary of some of our major wins and milestones since the last Capital Markets Day. Most exciting is perhaps the fact that we have received several two volume orders from our Satcom lead customer of value $1.4 million and $6 million and $5 million.
One in October of last year, and the most recent one in August of this year. And so this really is supporting the volume ramp of this customer, but it's just the beginning. We're expecting more POs for the rest of 2024 and beyond, and so this will be a significant driver for our business. Of course, this customer also funded the record order of $16.4 million for the development of new chipsets for their next generation terminal. But we also have customers behind them. Thorium Space is a customer that we're particularly excited about, and you'll hear their CEO, Pawel, present later in the Capital Markets Day about the exciting work that they're doing. But we're developing a custom chipset for them, which has been funded by them, earlier this year.
You'll hear a little bit more about that later today. I’d like to actually spend some time talking about what our unique offerings are in Satcom as well as 5G, and look at a couple of these customers as case studies. In Satcom, we actually have offered several unique features that allow our customers to then use those features to differentiate their products on the market. One feature is the formation of multiple beams. Of course, all our products center around beam forming, but in Satcom, one of the unique things that we are doing is building chips that can form many beams at the same time.
This is particularly important for what's called make before break, where you want your terminal to be able to connect to the new satellite before the connection to the old satellite is broken, because it's leaving the field of view. When you go beyond two beams to three and four beams, you can also have terminals that can form simultaneous links with multiple orbits, not just LEO, but also MEO and GEO as well. That's a very unique feature that one of our customers exploits. Another very important feature in Satcom is the ability to operate with very, very little noise, as well as very high output power and efficiency. This is particularly important because, of course, terminals are communicating with satellites that are extremely far away.
Here, it is universally recognized that our products have the best performance along these axes, and that is one of the reasons why customers come to us. And then, one of the new features that we're looking to offer in future products is the ability to share the antennas or the aperture between transmit and receive. This also significantly reduces the size and the cost of satellite terminals, and hence is very attractive to our customers. And so let's look at a couple of case studies. So, you know, our lead customer presented at our Capital Markets Day last year. They exploit the multi-beam feature of the chips that we build for them, and that's what really leads to differentiation for their products. As I mentioned, the beam forming ICs that we have built for them are ramping into production as we speak.
We already have, 1.5 million chips totally that have been ordered and that are being delivered by us over the next few months. But we also have a very large development program for new chipsets for their next generation terminals, which we expect to ramp to production in 2025. We're very excited about our next in-line satellite, customer, which is Thorium Space. And I don't want to steal too much of Pawel's thunder. He's going to present, later today. But we are developing custom chipsets for them as well. We announced the, development program for that in March of this year, and that work is proceeding exceedingly well.
One of the unique things about the work that we are doing with Thorium Space is we're building these beamforming chips not only for the ground terminal, but also for the satellite terminals that are in the space side on the satellites. So this marks the first foray of Sivers Semiconductors into building space-grade chips, and we think that that will be an exciting direction in the future. In the 5G space, the unique thing that we offer is that we have chipsets that can address all the different verticals. So if you look at this slide, this is showing the different kinds of devices that are built for 5G mm Wave.
Starting with user equipment, such as phones and tablets, moving to CPE devices, customer premises equipment devices, which are the gateways that bring the internet into your home, and then going up to small cells and repeaters, and then finally, base stations. As you go from left to right, the device becomes more and more complex with more antennas and hence higher unit cost. Of course, the volumes proportionately get smaller, where of course, smartphones and tablets are extremely high volume. CPE devices are still high volume because you have one for every home, and then, of course, as you get towards repeaters and base stations, the volumes become more moderate. And as a result, the chips that they demand are actually quite different.
On the left-hand side, the user equipment and the CPEs require devices that are extremely low cost, but highly integrated, and the performance can be moderate. While for small cells, repeaters, and base stations, the chip cost can be a bit higher, but what is critical is to have extremely high performance in terms of output, power, efficiency, et cetera. We offer solutions across the entire gamut of requirements and product lines. Our TRB Series of chips are highly integrated transceivers that can be used in CPEs and user equipment, while our Summit beamforming IC family is appropriate for the larger arrays that you see in small cells, repeaters, and base stations. You see that reflected in our customer base as well. I'm showing, again, two case studies here. We have our Tier 1 infrastructure vendor.
This is an engagement that has dated back to the MixCom days, March 2019, and the product development is, is still going strong. Of course, it's taken a while because these are extremely complex systems, as you can see on the right-hand side. Where we are today is that our first-generation prototype has been evaluated. The second-generation build is actually ongoing as we speak. We announced prototype orders received for that build in August of this year, and all of these are important steps towards the volume ramp that we see in the future. Another key customer is sort of an emerging startup company called WiSig Networks in India.
India is an exciting market for 5G mm Wave because, you know, the spectrum auctions happened recently, and there are lots of opportunities in India for millimeter wave to serve, you know, use cases which don't have a solution today. For instance, the distribution of internet to rural areas, and that's literally the application that WiSig Networks is targeting. We announced an engagement with them in December of last year to build a prototype CPE unit that uses our TRB modules and a WiSig modem that runs on an Intel Agilex platform. This so-called infra UE CPE is meant to be a backhauling unit that will be used to distribute sub-6 GHz signals to rural and underserved parts of India.
And really, why wireless is exciting for this, is that it is very cost prohibitive to sort of lay fiber all over the country in India because that infrastructure doesn't exist. Where we are today is that the project concluded last quarter. WiSig Networks is then using the prototypes that we delivered to them to do POC trials, which are planned in 2024, and in parallel with that, we will have an effort towards the industrialization and commercialization of these initial prototypes. And so that brings me to the end of the presentation. Hopefully, what I've shown you is that we not only have the broadest portfolio on the market, but also the highest performance chipsets and modules across 5G mm Wave, Satcom, and 60 GHz. This unique and highly differentiated set of offerings is keeping the stream of design wins going.
We see consistent wins that we are getting with lead customers because it's allowing them to offer differentiated products. We are seeing the ramp to volume from our lead Satcom customers, with other customers behind them to follow, and this ramp to volume, along with other large development contracts, has really shown tremendous growth in revenue, as well as order bookings, which ensure future growth looking to next year and beyond. Of course, what's important to be able to deliver on this are strategic partnerships with ecosystem players, which we have put in place. And then finally, I'll leave everybody with one thought, which is we—this underlying foundational technology goes beyond these three verticals. It has the potential to impact other areas as well, and examples include aut...
The automotive sector, radar applications, as well as new bands that are being considered for 5G plus and 6G, such as FR3, which is in the 7-20 GHz range, as well as the beyond 100 GHz or so-called terahertz. Of course, this is early stages for this type of research, but we're actively engaged in this as well. So that brings me to the end of the presentation. Let me turn the floor back to Anders, and it's been a pleasure to tell you about Sivers Wireless.
Thank you so much, Harish. We have now concluded the part of the internal presentations from Sivers, and we're gonna get some three very exciting presentations while we here take some coffee break, and we're gonna be back in about 15 minutes. It's gonna be some videos for you online, or you can go make your own coffee, and then we see you soon.
Today, broadband and being connected is equally important as water and electricity. Without reliable technology, there will be impact on your everyday life.
So Patterson can't leave-
At worst, with devastating consequences. Sivers Semiconductors is a vital link in the chain to secure a connected and safer world. With small, efficient products and strong commitment, we achieve great things.
With focus on semiconductor technology, our passion goes from innovation at every stage, all the way to our customers' success. It's a passion for customers, products, and technologies, and how they contribute to a more connected and safer world. That's us, passionate at the heart of innovation. Today, broadband and being connected is equally important as water and electricity. Without reliable technology, there will be impact on your everyday life. At worst, with devastating consequences. Sivers Semiconductors is a vital link in the chain to secure a connected and safer world. With small, efficient products and strong commitment, we achieve great things. With focus on semiconductor technology, our passion goes from innovation at every stage, all the way to our customers' success. It's a passion for customers, products, and technologies, and how they contribute to a more connected and safer world. That's us, passionate at the heart of innovation.
Today, broadband and being connected is equally important as water and electricity. Without reliable technology, there will be impact on your everyday life. At worst, with devastating consequences. Sivers Semiconductors is a vital link in the chain to secure a connected and safer world. With small, efficient products and strong commitment, we achieve great things. With focus on semiconductor technology, our passion goes from innovation at every stage, all the way to our customers' success. It's a passion for customers, products, and technologies, and how they contribute to a more connected and safer world. That's us, passionate at the heart of innovation. Today, broadband and being connected is equally important as water and electricity. Without reliable technology, there will be impact on your everyday life. At worst, with devastating consequences. Sivers Semiconductors is a vital link in the chain to secure a connected and safer world.
With small, efficient products and strong commitment, we achieve great things. With focus on semiconductor technology, our passion goes from innovation at every stage, all the way to our customers' success. It's a passion for customers, products, and technologies, and how they contribute to a more connected and safer world. That's us, passionate at the heart of innovation. Today, broadband and being connected is equally important as water and electricity. Without reliable technology, there will be impact on your everyday life. At worst, with devastating consequences. Sivers Semiconductors is a vital link in the chain to secure a connected and safer world. With small, efficient products and strong commitment, we achieve great things. With focus on semiconductor technology, our passion goes from innovation at every stage, all the way to our customers' success.
It's a passion for customers, products, and technologies, and how they contribute to a more connected and safer world. That's us, passionate at the heart of innovation. Today, broadband and being connected is equally important as water and electricity. Without reliable technology, there will be impact on your everyday life. At worst, with devastating consequences. Sivers Semiconductors is a vital link in the chain to secure a connected and safer world. With small, efficient products and strong commitment, we achieve great things. With focus on semiconductor technology, our passion goes from innovation at every stage, all the way to our customer success. It's a passion for customers, products, and technologies, and how they contribute to a more connected and safer world. That's us, passionate at the heart of innovation. Today, broadband and being connected is equally important as water and electricity.
Without reliable technology, there will be impact on your everyday life. At worst, with devastating consequences. Sivers Semiconductors is a vital link in the chain to secure a connected and safer world. With small, efficient products and strong commitment, we achieve great things. With focus on semiconductor technology, our passion goes from innovation at every stage, all the way to our customer success. It's a passion for customers, products, and technologies, and how they contribute to a more connected and safer world. That's us, passionate at the heart of innovation. Today, broadband and being connected is equally important as water and electricity. Without reliable technology, there will be impact on your everyday life. At worst, with devastating consequences. Sivers Semiconductors is a vital link in the chain to secure a connected and safer world. With small, efficient products and strong commitment, we achieve great things.
With focus on semiconductor technology, our passion goes from innovation at every stage, all the way to our customer success. It's a passion for customers, products, and technologies, and how they contribute to a more connected and safer world. That's us, passionate at the heart of innovation. Today, broadband and being connected is equally important as water and electricity. Without reliable technology, there will be impact on your everyday life. At worst, with devastating consequences. Sivers Semiconductors is a vital link in the chain to secure a connected and safer world. With small, efficient products and strong commitment, we achieve great things. With focus on semiconductor technology, our passion goes from innovation at every stage, all the way to our customer success. It's a passion for customers, products, and technologies, and how they contribute to a more connected and safer world. That's us, passionate at the heart of innovation.
Today, broadband and being connected is equally important as water and electricity. Without reliable technology, there will be impact on your everyday life.... At worst, with devastating consequences. Sivers Semiconductors is a vital link in the chain to secure a connected and safer world. With small, efficient products and strong commitment, we achieve great things. With focus on semiconductor technology, our passion goes from innovation at every stage, all the way to our customers' success. It's a passion for customers, products, and technologies, and how they contribute to a more connected and safer world. That's us. Passionate at the heart of innovation. Today, broadband and being connected is equally important as water and electricity. Without reliable technology, there will be impact on your everyday life.
At worst, with devastating consequences. Sivers Semiconductors is a vital link in the chain to secure a connected and safer world. With small, efficient products and strong commitment, we achieve great things. With focus on semiconductor technology, our passion goes from innovation at every stage, all the way to our customer success. It's a passion for customers, products, and technologies, and how they contribute to a more connected and safer world. That's us. Passionate at the heart of innovation.
Welcome back from this coffee break. My name is Anders Storm, and, I'm going to present to you now the external speakers of this Capital Markets Day, and I'm very happy to announce, Thorium Space CEO, Pawel Rymaszewski, who is with us here today. And we have about 20 minutes of that and a total of a hundred and forty minutes, I think, left. So very welcome, Pawel, and take it away.
Yeah, good afternoon. My name is Pawel Rymaszewski. I'm founder and CEO of Thorium Space. As you see on the slide, we are a space company, mainly, and that's what we focus on. Can I have next slide, please? Maybe easier to understand. Oh, yes. That's great. Thank you. So we've been founded in 2017, so now it's six years' time. And during those six years, we grow to more than 50 people. 45 engineers at the moment, and about 10 of them is PhD. Also, we are really focusing on the new technology in satellites or in satellite communication. Historically, I came from big operators. Some of the other guys in the team also, they came from the big operators, the main ones. And we are, at the moment, in three locations.
So we are in headquarters in Wrocław, which is in south of Poland, where is the main lab there. And I'm now in Warsaw office, which is like small lab and a business office. And there is a third one, which is on the seaside, the best one. That is the antenna and RF engineers mainly sitting. This, the third one is growing really, really strongly now. And if we can get another slide. We managed to come to solution to change the market a bit, through different R&D projects. So at the moment, we are shifting from R&D company, which purely R&D company, which was doing only R&D projects, to the company which is providing products.
Our all knowledge and IP knowledge, which we have, it came from five different projects, which been supported by mainly EU funding, also MOD funding. So in total, roughly, it took more than EUR 30 million. And we managed to come to the point where we become very good in the antennas, in arrays, starting from E-band, so it's like 70, 90 gig and going down to the K- band. Then we discover that we need something better, and this is the story with Sivers. Because we managed to find out that we need some special chips, we need somebody, we need some partner who will help us to give us this, technology or to, to make it as we want. And, and we start working, for more than a year now.
It started with 5G, then it moved to the Satcom, and this is where we are. As you see on the slide, there is something somewhere 5G, but we are maybe not giving away, but we are shifting away a bit from it. So we focus on two things. One is the space, as it's been said in some minutes ago by Harish. We mainly came for the chip, which can work in space, so space-grade chip, which will be very big adventure for both of us. I mean, for, for Sivers and for the Thorium. And then, of course, the chips for the terminals, which we are doing, for the ground segment. So if you can skip to another slide. So our terminals are active arrays, of course, double beam active arrays. We already managed to build the terminals.
We already have a product for Ka-band, which is not using Sivers's chipset or Sivers's Thorium chipset. It's using different one, and it's already been tested, and it's, you know, it's like in production version. The difference which we wanted, it was make it before break it on both up and down, so we need a special chip, and this is where we work together. We already know we will invest. This is important message probably for you, for all the investors in Sivers. We will invest quite a lot of money for the next year or more, or probably more than a year, in cooperation with Sivers because we need second generation, we need some agreement changes. But we got some roadmap which we want to follow and, work with, teams. Between the teams, it's from my perspective, it's fantastic.
There are no delays. We work pretty well. We managed to get what we wanted, even more than what we wanted at the beginning, and this is cutting edge of technology. And now we're just pushing harder and harder in space. That will be very interesting, and I will say that at the end, why. So if you can skip another slide. So one of the things is the terminal. The most important for me personally is because I spent all my life in big satellites, is the payloads based on active arrays. There was a problem or there is a problem still today. There is lack of devices which you can use for arrays, active arrays on the satellite side, because it's a millimeter frequency, of course. So this is where we've been pushing for that.
This is what we've been having the results already of the work together. We already know this payload will be absolutely game changer on the market, and it's becoming, 'cause what we are simulating and calculating, it's showing lots of changes in the way the system works. And it opens up the cooperation with you guys, with Sivers, it opens up additional opportunities and the doors, of course, to going in much bigger. So we managed to combine the cooperation with you and the cooperation with Teledyne and some other guys, and build a special system, which is we call it like a flat satellite. Which is not really true, but it looks like flat. So there are only arrays, direct radiating arrays, and this is where we're targeting as a main product in Thorium.
I think we are on the good way, which is, you know, digital beamforming or hybrid beamforming, using the sub arrays based on work with Sivers. So if you can skip another slide. Yeah, so part of it, so it's like a side effect of our cooperation and these chips for the terminals, it's the way we can provide a solution, critical solution for defense market. We are still Thorium Space, but in a few weeks, we will become Thorium Space and Defense. This is where we are changing now, slowly because of this. So we will be providing very small terminals or kind of modules, let's say, for the drones, for the medium and long distance drones.
And that's mainly because, you know, we are in a situation where it's extremely needed, and that's the only good solution where you don't lose the connectivity, and you don't risk the guys on the ground. That's purely defense business. We've been not trying to go there by ourself at the beginning, but that's the life. It become like very important part of us, and we already start working with big vendors. We shouldn't tell who is that, we won't, but let's say the big ones. And that's one of the legs which been separated from us. So we split, we really split the company into things. One, which is really satellite or space there outside the Earth, and the other one is the terminals on the Earth.
The third one, which we call it, is the chips, because there is a part of the team which is working with Sivers, mainly. Not only Sivers, but let's say 70% with Sivers on the RF chips, and we work also on digital side. So if you can go another slide, this is something which comes to critical part of the Thorium. What we started a few years ago, and the reason why I make this company happened, it was the way the new satellite should look like. So it should be software-defined, it should be smaller, it should be generic, which means whatever you want to do with it, or if it's commercial or double use or military or whatever, this is the same hardware. It's just the applications and the software you use differently. So we managed to come through many, many years.
Starting from 2019, we start working on the payload to come to the solution, which is already validated by ESA. So it means we have been given the teams from ESA for free, really, at some stage, to help us to develop new technology. And because of those work and those hard times where we've been, you know, checking if it's really working or if it's not working, how it will be, we managed to get involved in a national program and to get involved in ESA ARTES program. Which will be announced. We already know, but officially end of month, we will join big program as a company, and we will start doing or building the real payload. That's, for us, it's a huge step because that's the payload, which is for the 1-ton satellite, for GEO orbit, not even LEO.
So it's we're going with the full power now. And what we want to do during this time, it will take probably five years time. So during the five years time, so what we will have is we will try to get those chips done together, certified, and get space, you know, radiation ready and tested, because we want to focus on that in our payload. So it will be in use there. So that's very good times coming to us because that's, that's challenging for sure, because space is hard of... That's really hard, but that's something which push us a bit. And, what is going with this, it's the new market.
I mean, we're becoming one of few on the market in Europe who can provide the payloads, Satcom payloads, and one of two, probably in Europe, who can provide, well, maybe one, because we managed to work with somebody to provide very small satellite, which the big ones doesn't have. And that opens up new story in Thorium. So the next year will be a bit different. This is where we are. We don't want to go in other markets. For sure, we will move a little bit from RF to the digital RF, and this is what we are doing really with the Teledyne and some other NXP guys and the NanoX plore, also through ESA. And we want to focus on this. You can see some new features which will happen on the payload.
So we have beam hopping, we got beam steering, beam shaping, interference, localization, and all that stuff. This is all the technology we develop in-house. We all develop in-house, not the chip, not and you know the rest, it's ours. And that give the new way the satellites can work, and this gives the new tools for the very critical communication. And also, just to mention, we are added to the IRIS² European Constellation for the secure connectivity as one of few who can provide the technology. So this is comes with the chips for the space, mainly. That was the interest.
So this cooperation, which started more than a year ago, in Kielce, it started really in Poland, then become for us, this is the most important factor now, the, the most important milestone at Thorium, the company, really, because we shift completely to, you know, beam steering and the space, et cetera, with the nice partner who can provide us devices, which we couldn't get from the market at all because they didn't exist. And that would push us to the new life, really. And, yeah, and this is how it goes. So we will be working more with Sivers. What we see, we will be also trying to get more with Sivers Photonics, with the lasers, because we need them for digital. That's something which not many people knows that we do that.
And that give us very nice opportunity that it's still in Europe. That was what we've been facing because we didn't want to go outside the Europe. We want to focus on, you know, on the partners in Europe, mainly because of some governmental projects, and now it's possible. And because of that, we got what we wanted at the beginning. Just rough figures, it's more than EUR 30 million payload, which will be big step, and it will be all done in Poland. So, if you can skip another slide, I think this is the... Yes, this is what shows here, it's also kind of what will happen during this development. So we will give the payload, which can connect with any different type of network, really, and protocols and beamforming systems, et cetera.
This is like a, you know, hub which can or router can connect anything. And that gives really agile connectivity, and it gives the unlimited possibilities, really, really, what you can make from it. And again, the missing element, which we've had for many years, was the very small beamformer, which can do the basic subarray stuff on a satellite, which means the opposite frequencies, then the ground segment. And all the work which we are doing now, it was or it's possible because we will include that. Just to tell you rough figures, that's from interest, it's whenever we will use this technology, we can reduce the power on the satellite almost or more than 10 times, really, to get the same functionality. And that's a very huge impact on the cost of the satellite and the weight of it.
And this is what we will push, starting from the end of year, very, aggressively also on the market. So it will give for the Thorium, yes, for sure it gives, because we already are talking with the customers about that, but also for the, for the partners, it will give a bit of advantage, and we will need much more of work from on both sides. And of course, it gives the money.... So I want to thank you for the trusting in us at the beginning, that when we came with the Satcom chip, I know you didn't want to really look on it because it was just crazy a bit, the one which is on the space side. But, it opens up a new chapter here, and it now it's validated on as a system-level design.
It really makes sense. So for us, this makes sense completely, and we look forward in the future, really. So that's, that's the message from us. It's just the beginning of the big, longer cooperation. So yeah, this is just the business model, so this comes with the payload. We also will be providing satellite as a service. That's our idea. That's still in the negotiation, but probably will happen with the first one, so you don't need to be really experienced with the satellite. You can get all the support and the services based on that, but that's the future. We have some years to do that. There are already sponsors for that, so it's kind of good times, and of course, because of that, we will need much more from you. So yes, definitely, we will need much more from you.
Thank you again for the inviting me for this meeting, and thank you again for the really nice cooperation up to now.
Thank you so much, Pawel. Thank you for that. And, we're now coming into the next speaker. I'm checking here to see if there is people online. Yes, thank you for that. So we're starting a little bit early here, and it's Charlie from Ayar Labs, who was here last year as well. But now we're gonna talk a little bit more about the AI and actually how Ayar Labs is a vital part for the future there. Welcome, Charlie.
Thank you. Thank you, Anders. Can you hear me okay? Just a-
Yes.
Sound check here.
Yep.
Okay, very good. And you'll be presenting my slides, and I'll just tell you to advance as it proceeds. So, you know, first off, my name is Charlie Wuischpard. I'm the CEO of Ayar Labs, and, you know, I want to thank Anders and the Sivers team for inviting me back to the Capital Markets Day. I'm actually on holiday here in Mexico, so if you hear children in the background, hopefully you won't, you know, you'll have to forgive me. But I did want to participate this morning or this afternoon for you. You know, we've had a long-running relationship with Sivers, and so what I thought I'd do in sort of, you know, big picture is show some of the progress.
You'll be hearing from one of our partners, GlobalFoundries, after me, which has been also a key participant and partner in this journey for us. But, you know, first, just an overview of Ayar Labs. You know, the company was founded in 2015, based on, you know, research that went all the way back to 2010. And remember, we're focused on Silicon Photonics for data communication here. We're based in the San Francisco Bay Area. We have two locations. We're actually now over 130 employees, mostly still engineering and advanced degrees. More than half are PhDs in various science disciplines.
As you can see, we've spent quite a bit of time covering all aspects of our solution from a patent and IP perspective, and it raised substantial funding. I'd show a couple of, not all, but some of the notable, you know, financing sources, and I've separated in the top row, which are our financial investors primarily and have come in at various stages and, you know, have mostly all participated through multiple rounds. The bottoms may be a little more interesting, which is, you know, the strategic investors we have, and you can see it's quite a list in the industry, in the semiconductor industry, and it's, you know, in my experience, it's relatively rare. I will add that TSMC also invested just this summer.
We don't, you know, we don't have them on the page, but, quite a broad collaboration, and part of this is because we feel that for a technology like this to truly come to market successfully, it needs the help of all these partners. If you'd advance to the next slide. Again, a little splash page here of, awards and recognition we've received. We've actually got quite a bit more than this. But the nice thing is, it's, you know, we're, we're definitely a, a known quantity in the semiconductor industry, and closely watched by, by, you know, by both investors and, and participants in the industry. Next slide.
So, you know, the way I try to describe this, and I'll show a few slides and kind of relate back to AI a little bit, is that, you know, if you boil it down, the fundamental problem, the physics problem we're facing, is that the energy required to move increasing volumes of data, even over short distances, over copper, you know, starts to exceed the energy available to compute on the data. You're starting to see this today, and you can see how the industry is reacting by building, you know, much more powerful, data centers. Rack densities are going up, but you know that this cannot continue, you know, forever.
The opportunity really is that if you replace these short-reach copper links with optics, then it unleashes a whole host of benefits that can be had at the system level by creating different system architectures, new energy-efficient improvements, and so forth. This 1000x is actually, you know, a customer statement to us and sort of design points we're working on with our customers, not something we just, you know, invented. But that's the fundamental problem. If we go to the next slide, you know, that starts to translate into challenges we're seeing even today, and certainly, you know, with the explosion of interest and demand around generative AI just in the last year, it's really put a spotlight on technologies like ours.
And you can see that one of the problems here is that... And there's lots of slides like this, but one of the problems is that the communication - the percent of time spent in communication across some of these large generative AI clusters is starting to be, you know, quite substantial, to the point where it's becoming one of the major bottlenecks in the efficiency of these, of these new architectures. It's one of the reasons we have such interest in the industry by the, you know, by, by the, you know, the largest players. And so, you know, this is here today, and it's, you know, certainly as model sizes continue, it's not going to get any easier. We advance to the next slide.
You know, we've actually sized this market and, you know, this is a slide we've adapted from a presentation a few years ago that actually Cisco gave. But the idea and what we often talk about is that optical communication has been replacing, you know, copper for decades, really, at first long distance, and then at shorter and shorter distances as data rates rise. We're talking about short-distance interconnect. So you often see this on a motherboard in a system, or you see it going, you know, between a base station and, you know, a tower in a telecommunications environment. But we're talking about distances that range from centimeters to meters.
Now, our technology is tested out to 2 km, but we've really focused on this much shorter range, which today is predominantly served by copper and is in that transition phase. And, you know, in that sense, you can think of it as a brand new market opportunity, a brand new TAM for not just us, but others like Sivers and so forth. I mean, out in 2030, we've sized it to be quite a substantial market opportunity. And, you know, if you do your research, this has been validated by a number of industry watchers as well, at various relative sizes, I would say. But this gives you an idea of the focus. Next slide. And so we call it optical I/O, and I think that's really started to stick in the industry.
Now, there is another, category called co-packaged optics, or CPO. That's really been more associated directly with Ethernet implementations, and this is, actually a little bit different, but, but very similar in, in characteristic. And it's delivered, as you know, today, two pieces. One is a, a silicon photonic chiplet that we call TeraPHY, and I've noted that that is manufactured in the GlobalFoundries 45 nm photonic process. You'll, you'll hear from Anthony Yu, the general manager of that, group, here after me. And the other is, the SuperNova light source, and you can see that the chiplet is embedded as a, as an embedded device associated with, compute ASIC, or memory module, and the light source sits external. And I'll talk a little bit about that, in more detail.
So, by the way, my screen is half cut, so I don't know if that's, you know, I'm only seeing part of my slides, but maybe that's not the way for everybody. But, you know, I try to put this in context of, you know, the common devices you see in the data center today. So, you know what I'm showing, those 20 devices I'm showing on the right are 100 gig transceivers, something you would see in, you know, that's widely deployed in data centers today, and those are the pluggable devices that sit on either side of an optical cable. And I'm trying to draw. I draw a relationship between, you know, that optical technology that exists today, and it's moving to 400 gig, 800 gig, 1.6 tera.
So there is a roadmap for these pluggable devices, and I compare it to our first product, the TeraPHY chiplet, which today is doing 2 Tb in a very small 6 mm by 9 mm form factor. And you can see that, you know, the technologies obviously are quite a bit different, and the benefits and requirements are different. I mean, the devices used in those transceivers would not actually fit or be applied to the application that we're serving today. And you can see that we measure it in terms of, you know, aggregate data rate, energy efficiency, latency is very important, and also cost reduction. I mean, there has to be an economic benefit, both in terms of performance, you know, and price performance at the end of the day. We move to the next slide. So the way we've...
I've tried to just, it's quite a bit different. These slides are different, Andrew, than my slides, but, you know, I've tried to show here. There's usually a blue background with some words, but I'll try to describe what we're looking at here, which is, you know, the way the architecture works in general is that you've got a, you know, let's say that SoC would be provided from Intel, NVIDIA, AMD, could be a host of others. And what you're seeing are four of our silicon photonic chiplets manufactured at GlobalFoundries that are arrayed around there, with fiber attached. Now, I am also showing the external laser, and you can think of that as an external power supply, really providing the multi-wavelength light source into the solution.
And then the fiber that's going up to the upper left is really the transmit/receive fiber. So, you know, we're sending light in via the Sivers manufactured laser solution that we're using. And then we're transmitting and receiving through this electronic photonic chiplet, but like I said, is manufactured at GlobalFoundries. One reason we do this is because, you know, we could have chosen, and some tried, to put lasers embedded in the SoC. That's a very high-temperature environment, and, you know, lasers tend to suffer in terms of reliability and efficiency at high temperatures. And so I think if you were to do the research, you'll find that it's often the case now that this concept of an external light source is the preferred course.
So I have a few pictures kind of following here. The next, if I go to the next slide.
Hello, Charlie, can you hear me?
Yep. Yeah, yeah, I can.
So there, there is a cut and paste error here. That's why the slides are looking a bit strange, but we're gonna change that for, for Anthony afterwards, unfortunately, right now. But I think you know your slides, well, so hopefully that works.
Yeah, I know the slide again. I just... There's little words up in there that are missing because they were white.
Yeah. Okay, no worries.
No, it's not a problem. You know, I show a picture that Intel shared publicly just recently as an example, and also some pictures to try to put things in context. And then, you know, I think Anthony, who follows me, can really talk about, you know, sort of, you know, how important silicon photonics is for the future and the future of computing and the capabilities that GF's developed. But, you know, just gives you an idea of some of the different multi-chip packages that are possible. And very-- and this is, this is the whole promise of chiplet-based architectures going forward, is that you can mix and match various functionalities without changing the core, you know, compute silicon, if you will. And this is, this is the FPGA roadmap.
But you can see in purple there are the addition of optical chiplets, and some of them are, you know, coupled in the upper right with analog digital converters on the other side for various, you know, RF applications. There's other in the bottom right, that's, you know, coupled with AI, ML accelerators and, and HBM or high bandwidth memory, and those would be more, you know, sort of compute-focused applications. But you can see that, you know, part of the roadmap here is to add these, this, this capability of, of optical I/O, going forward. So just want to put that in context. Now, if you go to the next slide, the next few are... Oh, my goodness!
The next few are various pictures of our of our technology, and the the writing in the upper left, it's kind of blanked out by white there, really is showing the capability of what we're demonstrating, what we've demonstrated more recently. And you know, I had three bullets, and it was really that, you know, this demonstration that we showed, which is this, this form factor is a common form factor used in data centers, you know, called a PCIe card. So if you opened up your, generalized, you know, NVIDIA box or Intel box, you'd find a number of these compute accelerators of various types.
What we showed was really error-free transmission, which translates to latency, means we don't have to do, you know, we don't have to correct any errors in transit, which, you know, adds overhead and latency. But we're showing actually here in this demonstration is eight terabits of, you know, full duplex communication at less than 5 nanoseconds latency, consuming 5x lower power than other technologies. And keep in mind, the roadmap you often hear about in optical communication is the 100 gig transceivers, then the 400 gig, the 800 gig. And here we're showing, you know, 8 Tb, across really 2 chiplets. So, you know, game-changing level of communication, efficiency, and bandwidth, out of these very small, chiplet devices.
If you go to the next slide, I think I've got, you know, kind of an explosion here that shows the innards of it. And I've drawn some, and what you're looking at here first is the Intel FPGA. That's the one with the lid on it, and the expanded picture below. And you can see encircled in red, there are two of our photonic chiplets that are under that lid. On the right-hand side, you've got the two lasers powered by the Sivers laser arrays.
And so what you're really looking at is that the lasers are feeding the optical power around fibers on each, on the top and bottom of that card, comes around, the fibers are heading into the chiplets, and then the transmit, receiver, the I/O, is coming out to the connectors on the far left of that card. But, you know, what you're looking at here has never been accomplished before. So this is an industry first. It really shows what's possible in terms of capability for the, for the future. And it's been very exciting. In fact, if we go to the next page, I'll show just a few more things.
And I kind of touched on some of the high-level integrations, and what you see, you know, the reason I wanted to show the expanded picture, too, is that we're showing a 500-meter cable that's showing the connectivity, because people often, you know, for the longest time, people didn't know if this was possible, and they didn't know if it was possible at any sort of distance, and we certainly wanted to prove that. But in terms of just some key innovation areas, I mean, you know, for the chiplet, it's really the silicon photonics fabrication process and PDK that, you know, we've worked with GlobalFoundries for so long on, but that's actually fundamental to the technology we've developed.
You know, the new devices that we've developed, these are based on a microring resonator structure, which we could spend a lot of time. The fact that this is a monolithic chiplet, meaning we're combining all the electronic, digital, analog, mixed signal, and photonic devices in one piece of silicon, and the fact that we're employing advanced packaging and fiber-attached technologies. On the SuperNova side, you know. We chose to actually use proven laser technology. We've got it now able to be delivered in multiple form factors, depending on the application, and we believe that, you know, this is already high volume manufacturing cost optimized, and we're continuing on that journey. But there's quite a bit that goes into this solution. Now, if I go to the next slide, you know, we were. It's a little advertisement here.
We were, you know, blessed to be so chosen by the U.S. government, the White House. There was a demo day two weeks ago. This demonstration was showing the best of American, you know, made technology that's been partially funded by the government, and we were one of six demonstrations in the microelectronics regime. So, you know, nice accolades and good, good visibility, you know, for the continuing work that we do with the U.S. government and others in the industry. I'm almost done. I've two or three more slides, but I wanted to sort of give you a feel for where this- where we've come on this journey.
You know, it starts with back in 2018, I joined the company in 2018, coincident with the first, you know, substantial financing, with what I would say is the Series A financing. And so a lot of the early work was done on the actual devices, the microring resonators, the photonic process that we use with GlobalFoundries, just sort of refining the design. In 2020, we had our first samples of both chiplets and lasers, and then started to show full integrations and started sampling starting in 2022, and it's continued into this year with various levels of integration.
Where we've been spending a lot of time over the last 12 months is really refining the manufacturing process, qualifying the technology for high-volume manufacturing, working on all the things that are important for commercialization, you know, yields, costs, quality and reliability, you know, sort of, failure in time statistics. But certainly, this technology, you know, it's nice to build one, which no one's ever done before except, IR, but it more important is the capability of building, you know, these in the millions, and building them reliably, and working very closely with our supply chain and our customers to enable this, this technology. We believe this will be fully production-ready by, you know, end of year 2024.
However, we see the large volume ramp starting in the 2025-2027 timeframe, although as of this year, we've already shipped over 6,000 units. We just received another order for 1,000 last week, and we've tested over 100,000 units thus far as part of our statistical analysis and testing and qualification. So quite a bit of activity in that range. And then if I go to the last page or the next last page, I'm giving a sense of the roadmap. So what, you know, we've been working with to date is really in the 4 Tb chip, which think of, you know... We have three sort of scaling vectors, if you will. Macros can be associated with the number of fibers that are used, that are connected to each chiplet, and that are powered.
The number of lambdas or wavelengths, so the number of wavelengths per fiber, the laser that's powering those is provided by Sivers, as I've mentioned, and then the data rate, that's per wavelength of light that's carrying. And so you can see that we've got quite a bit of scaling that can be done across all three of those vectors. And our customer interest is kind of lies across the spectrum. One of the big volume areas I believe we'll see is in the 8-Tb domain, where you have 8-16, you know, fibers running 8-16 wavelengths and running at 32+ Gb per second. And that's one of the design points and one of the designs we currently have underway.
We are receiving already and working through test on the 22 and 24 generation chiplets, but I think the important thing is to see that in the world of semiconductors, you can't just build, you know, one solution and call it done. Things are moving so quickly. You have to have a roadmap of capability, and that's certainly what we're showing here. And I guess the last, you know, slide is really just a statement from, you know, one of our investor-important investors, NVIDIA, that it really recognizes that, you know, the next generation of speed-up, the advancement of Moore's Law as we know it, really requires new technologies. And you know, optical I/O is certainly one of those that's been selected as that high-promise technology for the future.
And so I'll stop there, and thank you for your time, and I think Anthony's after me, so you'll probably hear now the rest of the story, so.
Thank you so much, Charlie. I'm sorry about the slides. We exchanged them when we get it up on the internet afterwards, so we can see the nice slides you made, and I wish you a very good vacation now, and you can go back to the sun shore and hopefully some margaritas or something like that.
Yeah. Well, thank you. Actually, I'd like, if you don't mind, I'd like to stay and listen to Anthony's, and then I'll drop.
Yeah. Okay. Yeah, yeah.
I'll go off camera.
That's fine. They can wait for 20 minutes, I assume.
All right.
Okay, thank you so much for that. Now, we have Anthony Yu from GlobalFoundries, who's gonna have a 20-minute presentation. Thank you, Anthony, and well, welcome to Sivers Capital Markets Day.
Thank you very much. Can you hear me?
Very much so, yes.
Okay, so if I could go to the... Do you want me to present, or are you, are you presenting the slides from your end?
I am moving them ahead here.
Very good. Well, first of all, good afternoon, everyone. I'm Anthony Yu. Very pleased to be here as part of Sivers' Capital Market Day. I'm based in New York, and I manage the Silicon Photonics product management offering within GlobalFoundries. If you go to the next page, a couple of introductory slides. GlobalFoundries, it's a little bit cut off. I think I have the same problem as Charlie, but I think the relevant facts are there. GlobalFoundries is the third largest foundry in the world. In 2022, we did about $8 billion in revenue, shipped about 2.5 million wafers. We are a very focused foundry.
I'll explain in subsequent slides how we select our business and our customers, of which, Silicon Photonics and in particular, optical chiplets, are a main focus point for us. We've got about over 200 customers in 2022, with over 13,000 employees and a very IP-rich, innovation-centric company with more than 9,000 patents. Let's go to the next page, please. Our, our manufacturing facilities, our silicon wafers, are basically based in four locations, from the bottom to the top. Burlington, Vermont, which was a former IBM facility. IBM Microelectronics was acquired by GlobalFoundries in 2015. That's actually how I came to, to join GlobalFoundries.
We have a 200-mm factory in Burlington, Vermont, mainly focusing on mixed signal and RF technologies, RF SOI, silicon germanium, and wide bandgap gallium nitride are the main technologies in our Burlington facility. Do about 250,000 wafers per year of 200-mm out of our Burlington site. In Europe, we have our 300-mm factory, based in Dresden, Germany. There, where we have the capacity of close to 700,000 wafers per year. Our main technology in our Dresden facility is our extreme low-power technologies called FDX. That's a fully depleted SOI technology, as well as NVRAM and BiCMOS-lite technologies coming out of our Dresden very innovative facility in Germany.
We also have both 300 and 200 mm facilities in Singapore with capacity respectively of 730,000 wafers and about 400,000 wafers. We've got a variety of technologies there, including a little bit of silicon photonics, but I'll be end-of-lifing that as we center our photonics solely in New York. But Singapore has a variety of technologies that are listed there, including RF and silicon germanium, as well as some high voltage technologies. But the site that is of note for this presentation and of note for me is the New York facility. That's where I am right now, in our Malta, New York factory. That's our 300 mm factory. Brand-new facility or a fairly new facility, and certainly the flagship facility of GlobalFoundries.
There, we can do over 400,000 wafers per year. It is our leading-edge technology in the sense that we do 14 nm and 12 nm FinFET technologies, NVRAM, RFSOI, and we have a very exciting and flexible corridor for silicon photonics. Silicon photonics development, as well as high-volume manufacturing, will be in our Malta facility. We have plenty of capacity there to scale to the manufacturing volumes that you heard my good friend Charlie predict will happen as we look into this exciting market that will leverage the photons in the media of fiber. So go to the next page, please. As I said, we are a very focused technology.
If you go onto the web or if you listen to our CEO, Tom Caulfield, talk publicly about GlobalFoundries, our focus is on differentiated technologies. So as opposed to a foundry technology that we used to have, which was based on who could get to the smallest node fastest, moving to 7 nm, 5 nm, 3 nm, there's beauty and honor in that business model. We've chosen to work on a strategy based on highly differentiated silicon features, and it's really more guided by the mega trends that we see occurring in the world, in society, and I use those broad terms very, very specifically. Right, from left to right, you're seeing many things happening in the world today, including most recently with all the hubbub around things like OpenAI.
But you're seeing then a real veritable explosion of data. In 2026, they're projecting billions—I think the number here is 42 billion devices at the very edge of the network, generating about 180 ZB of data annually. So this is creating a real need for more capacity and more computing power within these data centers that are increasingly serving as a place where people monetize data. So business models are changing, data centers are changing. Back in the days of the pandemic, we were using data centers to work from home, go to school from home, entertain ourselves at home.
Now, we're in an area where data is being manipulated, and companies, especially in areas of things like OpenAI and ChatGPT, in terms of how to manipulate this data with very low latency and be able to create these very high-demanding large language models. Now, as a companion to that, all this work that's happening in the data center is creating enormous pressure on the amount of power that's generated within the data centers themselves. I spent a lot of time on the West Coast with, you know, the-
Name brands like Facebook and Amazon and Microsoft, and they all are very, very concerned, as everybody's concerned, with the amount of power that the data centers are taking up, and tied back to the enormous consumption of power linked to the explosion of data. There are doomsday scenarios, which are somewhat illustrative, that says that we may run out of electricity at our current rate, because by 2040, computing within these data centers may consume a significant percentage of electricity. Personally, I don't believe that 100% of the power is gonna be going to data centers, but a large percentage of it is going to the incredible compute demands that are placed upon these data centers by these new applications and these exponentially growing number of devices feeding into the data centers from the edge of the network.
That's creating a change within the semiconductor industry itself, of which GlobalFoundries is a part of, where tying back to the change in business model that we pivoted away from in 2018, the scaling law, known as Moore's Law, is changing relative to nobody can now continue both from an affordability standpoint or even from a scientific standpoint, to continue to proceed down the path of simply scaling to achieve more computing resource or power.
And that's driving new architectures, many of which, Charlie referenced in one of his couple of his fine slides, looking at things like heterogeneous integration and putting multiple chips of different functions with the same package, and allowing room for new technologies, such as photonics, to pair with computing chips built in traditional FinFET to provide these incredible solutions, driving the fourth mega trend that we're happening here, which is the growth of artificial intelligence, machine learning, and large language models. It just seems like I can't believe it was a year ago. A year ago today, I was showing my wife this interesting thing called GPT-3, but since then, Generative Pre-trained Transformer is a language model that's producing incredible innovations and incredible demand.
I think NVIDIA will announce their earnings today, but we're seeing tremendous demand and tremendous growth in terms of meeting the hardware requirements for this, this new, this new era that we're living in, based on AI and ML. So these four trends are kind of driving what we've done, and, and I'm the architect of silicon photonics within GlobalFoundries. We were fairly fortunate in the sense that we started to invest and build this technology in 2018, but since then, these four very powerful mega trends have kind of propelled us forward into, you know, the right time and the right place, and I believe that working with Sivers, we will provide a really, very powerful solution into these, these four trends, in the next few years. Go to the next page, please.
So as I hinted at, the main space for Silicon Photonics, and in particular, the work that Anders' company is doing, is relevant to how do we change the data centers? Now, data centers have been around for a while, and frankly, photonics has been around for a while. Photonics was really amply used to communicate between data centers across long distances. But now we're seeing the photonics being applied not just between data centers, and they'll always be between data centers, but not just between racks either, but now within the rack themselves. Data centers, the primary problem that they used to have is how to connect racks from rack to rack across these football field-sized data centers.
Now, we're seeing, again, part of the benefit of the AI, ML, large language models trend is trying to actually connect GPUs together to form a compute fabric. That's what Charlie was referencing when, when he said that co-packaged optics is mainly directed toward the applications for Ethernet, but we're also seeing now an explosion of a new market to connect GPUs together, to be able to create the computing fabric necessary for these large language models. So generative AI is driving a complete reconfiguration and disaggregation of the data center. How well can you reportion workloads across, computing facilities, across computing racks, across storage racks, and across memory racks, but also, how well can you connect GPUs in, you know, hundreds, if not thousands of GPUs, to be able to support these large language model demands?
From a standpoint of electrons versus photons, photons wins. My good friend, our CTO within GlobalFoundries, Ted Letavic, has a two-word slogan pinned up over his desk. We're nerds. I'll admit it. It says: "Physics wins" because you can't get any faster than the speed of light. So data from a standpoint of latency, data transport at the speed of light is way faster than copper, and more importantly, you have this effect within copper called Joule heating, which is the interaction of electrons with the copper as it travels, which creates heating, which creates less power efficiency. We have the ability, as demonstrated with Ayar Labs and Sivers, to be able to put multiple carriers on a fiber without interference.
You saw on Charlie's chart, he talked about 8 wavelengths per fiber, 16 wavelengths per fiber, 32 wavelengths per fiber on his roadmap chart. You simply cannot put 8 or 16 signals of copper on a single wire. That doesn't work. So multiple carriers in a fiber allows a very unique attribute to photonics, which is brought about by working with GlobalFoundries silicon, Ayar Labs design, and Sivers Lasers for having a bandwidth multiplier and allowing you to very reliably and very neatly relative to power efficiency, scale bandwidth while keeping power efficiency and keeping power very, very, very low, treating that mega trend that we have in preventing data centers from consuming all the electricity in the world. Let's go to the next page. So here is a marketing chart of the silicon photonics technology that I offer.
It's been branded. If you were able to see the top of the chart, you can see the trademark that we call this. This is called GF Fotonix, with an F, P-H-O-T-O-N-I-X. That's the brand name of it. We have actually qualified this technology about a year and a half ago, and we are very proud to be working with Ayar Labs to be featured at the White House, to be featured at various conferences, and to be, you know, one of the six best technologies highlighted by U.S. government, and recently as a ability to produce, you know, the most amount of data, bi-directional 8 Tb, using this technology with a Sivers laser coming, providing the light from the outside.
From top to bottom, again, you really can't see the names on the chart, at least I can't. But you can see the various unique attributes of the technology, which include the types of devices that you heard Charlie talk about. The workhorse of silicon photonics from a transistor standpoint, the analogy to a CMOS transistor, is called a modulator. So you can see the two different types of modulators that Charlie talked about, the Mach-Zehnder modulator and the ring-based modulators. These are all constructed with silicon, with introduction of new materials, to allow you to control photons on silicon, just as we have been able to control electrons on silicon for the past 50 years.
So this is an application of CMOS technology, of high-volume CMOS manufacturing in our factories within GlobalFoundries, but now instead of controlling electrons, you're controlling light as it passes across the chip with minimum loss. So you can see, we also have a variety of passive devices like waveguides, which allow us to control the photons, steer the photons, and control the polarization of photons. That's what was occurring at the 1 o'clock to 5 o'clock part of this chip diagram. And, you can see in the chip, the little green cylinders with the 2 arrows, that would be the fiber that would be bringing the light from the Sivers laser, providing the light into the chip and controlling the photons as it works itself across the chip itself.
On the left-hand side, in orange, are the electronic elements, and this is where Charlie talked about. What's unique about our technology is it's a system on a chip that includes the photonic elements, which is the photons coming in from the Sivers laser, with the CMOS, the high-performance CMOS, to actually control and capture the photons, convert them back and forth from photons to electrons, electrons to photons. So this is our workhorse technology. This is our manufacturing technology that will be used to scale to the different products and volumes that you've heard so much about in the Capital Markets Day. Next page, please. So what is this thing called Co-Packaged Optics? What does it look like from a manufacturing standpoint, and what are the innovations that we are investing in, just as Anders is investing in it at Sivers?
Well, as I mentioned on the left-hand side of the chart here, we've invested in a unique... This is the only product offering in the world among our different foundry technologies that have the monolithic integration, which is the same chip integration of CMOS FETs, high-performance RF CMOS FETs, with photonics. The advantage of putting the electronics on the same chip as the optical components is you get better performance. You get a higher signal-to-noise ratio, which you need as you scale this thing up. So monolithic integration, by virtue of reducing the capacitance there, allows you to get a higher performance, less bit error rate. Remember, Charlie talked about the 8 terabits bi-directional. You don't want any errors there.
You don't want to have any dropped bits or any mistakes that have to be corrected, and monolithic integration provides the highest level of bit error rate and performance across any of our competitive offerings. In the middle there is the critical thing. To take Anders's laser and deliver the photons seamlessly and cleanly, without what we call insertion loss, into the chip, requires very steady and very highly precise alignment of the fibers with the light and directed into the waveguides within the chip. And then on the right-hand side, the innovation that Charlie talked about, where we partnered very close with Ayar Labs, is to develop very powerful transmit and receive applications of using microring resonators to be able to handle the 8 different wavelengths per fiber and moving to 16 and 32 next.
These are smaller devices. They're less power hungry, and they are very, very fast and very, very active. So, as we move toward optical scale-out, which I said was the key differentiator between photons and electrons, we'll be relying on these types of devices, like microring resonators, which are about five microns in diameter. But, so they're much smaller than Mach-Zehnder, which allows you to do a more efficient chip. Next page, please... So, this is a picture from a publication called Applied Physics Letters, which is a nice depiction of what I'm seeing in the foundry world as this evolves. So this is a depiction of, it's for co-packaged optics, but it basically depicts the trends that we're seeing and which are moving quite, quite rapidly.
So if you look on the upper left-hand side, that's Gen 1, which is still prevalent in the industry today. What you see there, that dark brown color chip sitting within the rack, is feeding electrons out on those orange links all the way out to what's called the faceplate of the unit, which basically converts the orange photo-electron signals into photons, and then it would be passed out through the fiber coming out of the Gen 1 front part of the faceplate. That's what's called pluggable optics, and that's pretty much primarily in the data centers today. What we've moved to in the last few years is called onboard optics, where you can see the difference there as within the rack itself.
You've got the ASIC in the middle there still, but now moving closer, moving closer to the ASIC now are the transceivers. Those are those silver-colored rectangles. So you move that closer because as the data rate increases, coming out of the ASIC, the rate of compute, you tend to lose a lot of signal as you move across those long traces shown there in generation one. So to move to a higher bandwidth, you have to move the transceivers closer to the ASIC to avoid signal loss and actually allow you to keep the power lower. That's called Gen 2.
Where we are today is we are now seeing the first input of what's called Gen 3, and that's the work that we're doing very closely with Anders and with Charlie's company, with Ayar Labs, and that's called 2.5D co-packaged optics. So there you can see the difference. If you squint hard, you'll see that actually on the ASIC package itself, you now have the transceivers, the optical chiplets, similar to what you saw on Charlie's chart, sitting on the same package as the ASIC. So you've got them almost adjacent to the ASIC itself, which allows you to have, you know, very, very high bandwidth, very, very low power, so very good power efficiency.
Relying on many of the features that I showed you in previous slides within the foundry itself, to be able to support the features within the optical chiplets for co-packaged optics. Moving to the right is where we're probably headed to in the next two years, which is 3D co-packaged optics, where everything's gonna be on one chip. The optical chiplets and the ASIC will all be integrated together. But right now, I think the dominant space where we'll see the volumes come out, as Charlie indicated, I agree with them, probably in the 2026 timeframe, will be in the form factor of the 2.5D co-packaged optics. Next page, please. So in summary, it's been my pleasure to be here today and kind of describe the world as I see it within the foundry.
Now, foundries, we only survive and make money if we can see volume. We are a capital-intensive business, and we are interested in high volume to take advantage of the, you know, the large number of wafer capacity we have in these incredibly expensive billion-dollar factories. But I think that this is an inevitability based on the trends that we described on the on our, you know, first couple of charts. We think that if you look at the important metrics of data transport, of data transport within the data centers for things like generative AI, the key metric is gonna be: how much data can you actually stream out of these ASICs? That would be measured by gigabits per second, per chip size, so you have to normalize that by millimeters, divided by your energy efficiency, which is number of picojoules per bit.
That's a very, very wonky way to describe it, but the best way to describe it is you want to have maximum bandwidth for minimum energy, and you can't get that balance without moving towards photonic solutions. There's no other way to do that, based on physics. So to solve the problem that we have at the end user, which is the data center, the only way to solve this is with photonics. So you have to be able to break the electrical reach limitation. You saw that happening with my depiction from the Applied Physics Letters diagrams; you've got to be able to get around the fact that you can't get to those data rates with long reaches without losing lots of signal.
So, if you do this with photonics, it allows you to be able to basically re-engineer and re-architect your data center architectures. I think this thing called artificial intelligence and large language models is not a fad. It, it's here. It's gonna transform the way that we live, to transform the way that we work, and transform, you know, everything in ways that we actually can't describe today. But I think it's gonna likely require both photonics and co-packaged optics, and I would actually modify this chart to say optical chiplets, to be able to address what I described before, which is the way to link hundreds, if not thousands, of GPUs together to handle these large language model computing problems.
So there's a whole set of innovation knobs that I tried to describe in a couple of charts in terms of how we at GlobalFoundries are going to bring in new optical materials to work within the confines of our existing CMOS factory, which allows us to scale this with high manufacturing yield and high volume... but bringing about enhanced performance. So that's actually my trick that I have to do, is basically work within the CMOS factory, take advantage of the learning on the scale there, but bring in new innovative materials. You've heard about all kinds of cool SiP, I'm sure, earlier from Anders himself, the fine work they're doing at Sivers. You've heard Charlie talk about the innovation within the chip design itself and what they're doing with their product. We have done the same thing within GlobalFoundries.
We've innovated as well, and so my closing statement here is I'm really proud to work with people like Ayar Labs, and especially proud to work with Sivers, to develop an ecosystem that's gonna meet the challenge of this, as this thing rises to volume over the course of the next couple of years. Thank you very much for the time today, and it's been a pleasure.
Thank you so much, Anthony, and I just want to say to you that everybody's seen the slides well here, even if you couldn't see them correctly.
Oh, okay.
So, and we will make sure-
Good
That the slides looks really well. So thank you so much for that, and we're now going to go over to the Q&A session. So thank you, and me and Lotte is coming up here, and we're gonna take some questions in the audience, as well as we're having questions coming in from the online viewers. I've heard we have as much as 180 people online here, so that's really encouraging, I would say. So let's start with the first question. What is the biggest challenges for the coming years? And I would say that what we've been talking about today is to go from the 30%-80%.
So in itself, it's not a challenge to do that, but the challenge is, of course, to make sure that our sort of fabless things that works and we getting to fruition, and that, of course, our customers also becomes as successful as they need to for us to be successful. So that, I think, is sort of where we are in that piece. Another question, can your technology somehow be used in micro lens screens? And I would say that you could probably do things in that direction, but that is nothing we are doing today. Let me see. Is the TAM value a yearly market cap or a total value of the market?
It is a total value of the addressable market for these kind of chips or these kind of lasers and so forth. So it actually describes how much everybody can sell into a market like that in the future, and sometimes it's used for many years. It can be used for one specific year, basically. In the slide, it says wearable. Can you please tell me more about that? And that was in Andy's slides, where we see some future where consumer electronics needs to use different lasers for, for example, you can use them for touch or near touch or something like that, to be able to sort of steer what your - where your headphones are doing, for example. Standards should be the important, how you work with these.
So yes, so if we look at the wireless business, there are standards for 5G, for example. There is... And there are also standards for Wi-Fi and so forth. So all the chipsets and things are done according to those specific standards. If we look at what, what Andy said today, this very long abbreviation, there is a standard for how the lasers will actually sort of send the light and, and with which frequencies and that sort of, sort of thing. So yes, that it is very important, and we are part of standardizing this, standardizing these things, especially in the laser side of things. In the 5G things, there are many others that do that, and we, we more follow on that end. When do you think you need an answer for the Fortune 100 to get into full production?
I would say that answer is changing a little bit with the very good news now with the hybrid manufacturing stuff we're looking at. So I don't have a direct answer right now, how much it will take and how long time it will take, but at least it will be much shorter for us to get to full production with this hybrid part than if we do it otherwise. Then, when will the new photonic product line be ready for production? I think we need to come back to that as well. But in a sense, we have the production lines and our customer or our sort of new outsourced partners have production lines already that we can sort of use.
It's going to be more of work when, when it comes to sort of qualifying our chips on those production lines, and that is, of course, a task we now look at and specifically with those customers and, and the ones we will choose will be quite important. You're welcome to write more questions, but do we have questions in the audience here? We have a mic over there. Yes, Johan, Danske Bank.
Yes. Hello, can you hear me?
Yes.
Yeah, yeah, great. I have a small timing question on the first Fortune 100 customer, in terms of the hybrid solution. Right now, you have shipped 30,000 chips to this customer for qualification and system tests. How long would it take for the potential hybrid manufacturing supplier to also perform the system and qualification test that I assume is required? And a follow-up on that is, in terms of product release window, how would you say that the 2024 product release window is that window closing or for a hybrid solution? Or I know we're more looking towards 2025 product release for that.
Yeah. So if we look at sort of our outsource possibilities in the hybrid environment, of course, there is more of a qualification issue. And depending on the qualification, exact qualification hours for that customer, I would say it's up to something like six months for a full qualification on a new process in a new fab. When it comes to the window for getting something out in 2024, I think that window is of course challenging, and I don't see that they have sort of either talked about anything in that direction currently. So I don't think we see a sort of a product window that will fit for 2024, actually.
All right. Thank you. And then I have a question about the Satcom customer that you have. You received $5 million order for that. How far in advance would GlobalFoundries need to book capacity for potential follow-up orders from this customer?
So it depends a little bit, but I would say that the minimum is something like 8-12 weeks when you get an order that you can deliver. But it's, of course, our friends at Global, who, depending on what they have, therefore, we're also looking at these kind of things that Anthony mentioned in general, not just for the photonics, but also since we are using our 5G and Satcom development also in their fabs, in two of the fabs he mentioned. It's important to make what's called a corridor, to sort of address capacity in those fabs. So depending on how you do that and if you do it correctly, but it doesn't need to be, you know, six months or 12 months.
It could be as little as down to eight weeks, but it also depends what is available. But normally, I would say three more months in that sense.
All right. And I have one more question, in terms of financing, for this hybrid solution. I assume you still need to do some kind of investments in your own facilities. How, if you look at other companies in the semiconductor industry, it's not unusual that they receive, like, prepayment from customers. In this situation would be the Fortune 100 customers. How do you view on that, and the potential of receiving that kind of aid from or prepayment from your customers?
Yeah. We haven't entered that discussion yet, with that specific customer. I would say that there are, you know, possibilities to do that, if they're willing to, but it's too early to say, I think. What was the first question? Sorry, it's like-
No, I think that, that was the question.
That's the... Okay. So yeah, we have to come back to that, I think, when we can see the sort of the next phase and what they're saying in that sense. Yeah.
All right.
Yep.
Thank you.
Okay, while we're changing over the mic, we have some other questions here. "Tell us how things are going and developing in India. Thank you." So I think Harish went through that quite well, in his presentation, but we have already delivered our piece on this. They are finalizing their technology, and they're gonna do proof of concepts now, I've heard late this year, but also starting early next year and so forth. We have Jesper from Redeye here as well for a question.
Yes, so thanks for taking my question. So regarding Satcom, you're obviously excited about the, all your customers and so on. And, like, looking into the next quarter and also years, it seems like All.Space could ramp up quite significantly, and yet you say that you're, you're, particularly excited about Thorium Space. So could you just, like, talk about, like, how you view the potential of, of, of both of these customers?
I think we are as excited by both of them, and there's sort of different excitements. I think short term, it's what Pavel was talking about for Thorium, is that the space side of things, and what they are doing is quite exciting per se. And also from the beginning now, it's a lot of sort of development, partnerships with European space agencies and everything that Pavel soon is to announce what they are doing. So that is a very exciting piece on that side. When it comes to All.Space, we are in sort of two large ongoing project, where the first project is now ready, and we see, you know, volume production coming in. And of course, one of our main goals is to get to a lot of volume production.
So that is, of course, an important piece here to see that ramping, and we're just sort of scratching the surface on that ramp, and the possibilities are very interesting in the piece with All.Space from that perspective. And then, of course, in the longer term, we'll also see that from Thorium.
All right, and then, and then also, one from my side about the potential ramp-up of the first Fortune 100 customer. I mean, could that potentially start in, like, low volumes on your existing fab, as it kind of like ramps up in just small volumes, and then, like, alongside that, you will make the system test for the new hybrid solution? Could you just elaborate?
Yeah, I think that the ramp probably could start like that. So I mean, we provided now 30,000 chipsets. As Andy mentioned, we have a possible capacity of maybe 5,000 wafers a year in that fab. We're using maybe 1,000 wafers a year currently, so we have excess capacity there. But of course, seeing the total numbers they're looking at, and depending on the sort of two different volumes we've seen in the RFP, it's going to be sort of interesting to see what kind of device and how that ramps and what the needs are. But I think that the big, the absolutely biggest volumes are, of course, further away and more connected to smaller devices in bigger volumes.
All right. And then just one last about CapEx need. I mean, say that if you were to, like, buy a new fab, say that the cost would be something like SEK 1 billion, and this would dramatically reduce that CapEx need, but, like, ballpark numbers about the investments for even a hybrid solution.
So it's a vastly different cost, and I think it's even that we can fund that sort of with the sort of cash we have now and the cash we're going to generate in the future. So the preliminary numbers we have now, which is not sort of finalized in a way, shows that it's sort of a very minimal thing to actually get what we're doing in the step one, step two and three, which is still going to be internal, in the sense to... Rather the external step, which is the sort of $100 million investment to get to, you know, 30,000 wafers a year. Fortunately enough, then, we have three different sort of outsource opportunities here who have done that investment already, so we would not have to do it.
All right. And then, then just about, I mean, cash or the eventual need of cash for any investments. It seems like if this hybrid solution would fall into place, do you see any other, like, big investments that you would need to do, except for eventual M&A and so on, like, before you reach a positive cash flow?
No, not what we see right now. You know, we're making the plans for 2024 as we speak, and I don't see any sort of specific need. We sort of trying to reduce and only do CapEx when absolutely needed. The plan is, of course, to make sure that we can sort of do what we want to do and need to do with the current cash we have at hand. But Lotte, if you have want to say something?
No. I mean, just echoing that we are planning right now as we speak, but we have no initiative sort of in front of us, in terms of, you know, equipment or, or CapEx need other than what you talked about that we're planning for, so.
Okay, thanks.
Okay, thank you, Jesper. So we have some more questions coming in here, and I need to translate from Swedish to English immediately. For a while ago, you talked about building fabs in the U.S., and that's sort of one of the hottest thing that could happen, and why are you not doing that? So, yeah, one of the big reasons, of course, is that we don't see that if we don't have to, we don't need to do a CapEx investment in this. So it's a huge project, and it's a huge risk to do that, of course.
We might do it long term in the future, but right now, we don't have to take that risk, and in the current market, that type of risk is absolutely something that people are not sort of willing to to invest in, I would say. Another Fortune 100 question: Are they all or mostly developing photonics application within different areas compared to other Fortune 100s? Thank you. I would say that there are multiple things that the Fortune 100 customers do, but they, they are, of course, using a lot of photonics application in different ways.
When, you know, if we look at, you know, the top 10 companies in the world when we talk about, they all do data centers, they do technology, and they need to use, as we've seen here today from Ayar Labs and GlobalFoundries, they need to use lasers in different ways. So that's been a part where we've been really successful. Many of the Fortune 100s that we are seeing today are not so much into, you know, 5G or Satcom. So that's why we're so much in the photonics business, I would say. Do you believe in that you will go-- get both the sender and an emitter, or the emitter and the receiver for the Fortune 100 customer? It's hard to say.
I mean, they're we're still in the RFP process and everything, but there is definitely a chance for that since they have invested quite heavily in both chipsets. You are talking about 60 customers engagements today. Do you think that's enough for your future earnings? Is there any bottlenecks to bring in more customers? So no, not really, and I think that's one of the sort of efficiency in our model, that if we have these solutions with the fabless and the hybrid fabless solution, and there's a lot of chipsets that can be sort of reused in different verticals, it doesn't create a bottleneck. And we have definitely seeing a very exciting pipeline on new customers coming in, so we can definitely add to that stage.
How is the status for Fortune 100 customer two and three? Yeah, those are still sort of research type of projects. It are long times, as you've seen with this customer, it's taking five years. I think those customers have been in the pipe and discussion for around two years now. So we are sort of waiting, in a waiting mode if they're gonna go further or not. So we don't really know yet what's gonna happen there. Do you expect more orders from All.Space or the Fortune 100 this year? Good question. Maybe, maybe not. We'll, we'll actually see. But, I mean, it's a short time to the end of the year, so, we'll do that information when it comes. What's your expectation on Ayar Labs and how large orders from them?
I mean, we got sort of an NRE order here in, I think it was July, August somewhere. And that is, of course, to get to a phase where we get to volume. So from that perspective now, we're sort of going to deliver on that $1 million order, so we don't foresee orders, sort of short term in that sense. What are the expectations? Now it's same again. Yeah, there's more questions around the same here that we already answered around what do you need more money in the nearest 12 months? That's not the plan, as we said. When will we hear about the forecast for 2024? We have no plans as of now to give a forecast for 2024. We're looking internally how we'll do forecasting in the future.
That was the first time we did a forecast now for 2023, and we're gonna look at if we do midterm, long-term, or short-term forecast. It's a bit early to actually say. I can also see now that we're actually five minutes past, and there's many questions here, but I think we answered as many as we could. I would like to thank you all for joining us here today, and I would like to thank Lotte also for participating and everybody coming.
Thank you.
Thank you so much, and have a great day. Thank you.
Bye.