Hello. Good afternoon. This is the 28th Needham Growth Conference. My name is Charles Shi. I'm the Semicap Analyst here at Needham. Joining me on the stage is AXT. We have CFO Gary Fischer and Vice President Tim Bettles with us today. And the company will do a presentation, and then we can open up for audience to ask questions. Gary, Tim, stage is yours.
OK. Thank you all for coming today. OK. There you go. OK, thanks for coming. So, brief history. Company started in 1986, went public in 1998. Our products are specialized wafers for when silicon wafers won't work. A lot of process technology involved. We're based in Fremont. We manufacture in China, but there's been a lot of discussion today about maybe having also somewhere else. And we have a pretty sophisticated supply chain because we're vertically integrated. So, the raw materials that we need to make the wafers, we have partial ownership in companies that are providing that to us.
So, what's happening with the stock? If you've looked at all, when we were here last year, the stock was in single digits. And this year it's in double digits. So, I'd like to say that we caused all of this. But what I would say is that we've worked very hard to be prepared.
We have the best product in the market for indium phosphide, and between last year at this time and today, the world has become aware of a growing need for indium phosphide for data centers, and my colleague Tim Bettles can describe the market application when he speaks in a moment, so the demand's being driven by large-scale architectures.
Moving forward, indium phosphide's critical, and if you want to invest in the whole AI data center infrastructure, indium phosphides are one of the first items in the value chain, and we're the only pure-play indium phosphide company available on the NASDAQ or in any other stock market in the world, so we're uniquely positioned to help the market do what it needs to do, and it's a unique position in terms of a stock and an investment opportunity.
Okay, so I'll thank you, Gary. I'll go through a little bit about our products. I'm not actually going to start at the top. I'm going to start further down the bottom. We manufacture three compound semiconductor materials, as we call them, germanium, gallium arsenide, and indium phosphide. And then we back that up with a supply chain. We have, as Gary said, ownership in some raw material companies.
We own outright two main companies that supply gallium, obviously for our gallium arsenide business, and also purification and recycling in there, and then high-temperature crucibles made from pyrolytic boron nitride. Starting from bottom to top, because I know everybody's interested in the top line right now, germanium is a relatively small portion of our business. It accounts for less than 10% of our business. Predominantly, it is used for satellites. So it's a solar cell material for satellites.
It's a relatively large market, but it's something that we limit because it's extremely sensitive to raw material germanium price, and the raw material of germanium price fluctuates a lot, so that's something that we kind of hop in and out of. Gallium arsenide is the next material, and that's really where the company was founded on, to build gallium arsenide.
We use that for industrial lasers, VCSELs for data centers, LEDs for lighting and display, printer heads, cell phone power amplifiers, so there's some market opportunity there that we deal with, and we do have some market growth applications there that we are looking into and we're working on. However, indium phosphide is really why the stock has gone up over the past 12 months. Indium phosphide is the core material system for optical transceivers, pluggable optical transceivers, and co-packaged optics, and that's really what we're looking at here.
We're seeing a massive increase in demand for AI data centers. As the speeds go up, the connectivity starts moving more and more away from copper and electrical into optical connectivity. So right now, what we're seeing is we're seeing growth in what we're calling scale-out. So it's a laser-based product, usually an EML-based product or a DML-based product that's going into a pluggable transceiver that connects rack to rack. That's where we're seeing a lot of growth today. And we're seeing growth potentially in the order demand growing 2× this year, potentially another 2× next year just for scale-out.
Beyond that, we see scale-up. So this is the CPO opportunity, the silicon photonics. I know there's silicon in the title. And a lot of the modulation and the light manipulation is done with silicon and silicon compounds. But we still use light. So to generate that light, we still need indium phosphide. So that gain chip that we use there is all based on indium phosphide. So we're going to see an inflection point when CPO takes off. 2028, 2029, we should see some real growth beyond what we're seeing today. As Gary said, we have a sophisticated supply chain. I mentioned this briefly just a moment ago.
We have about 10 raw materials companies. So we make our own indium, our own quartz for the furnaces. We make our own furnaces. We make our own raw materials, gallium, arsenic. And we make the high-temperature materials that go into the furnaces as well. So these are all built on a campus in China. It mitigates our risks of supply chain issues. They're all there in China. It allows us to do a lot of recycling. It allows us to control the cost of the raw materials.
And as the markets increase, and as we see the markets increase with indium phosphide, the demand increases, it allows us to scale those raw materials companies up to match the demand for the substrates. So we're uniquely positioned in that marketplace. We're the only substrate supplier that has a vertical integration backwards. Some of our competitors in the substrate market have a vertical integration forwards where they make their own lasers and detectors, which of course means they're not only a supplier to our customers, they're also a competitor to our customers.
But we're uniquely positioned where we control the backwards supply chain. A lot of our customers, of course, really like this supply chain model because it gives them security of supply when we're seeing massive growth opportunities like we're seeing today. With that, I'll hand it back over to Gary.
Financials are available. So I'm not going to spend time on it right now. We did file to take our subsidiary in China public. The name of that company is Tongmei. Most of you have heard of us talk about this before. It's been in play for several years now. So it's been delayed, disappointingly so. We haven't given up. We're still in the queue at the Shanghai Stock Exchange to have Tongmei go public on the STAR Market of the Shanghai Stock Exchange. And we keep responding to requests for information. It's an interactive process. It's going fine, except it's taking much longer than we hoped it would take.
Ok, so what's the market opportunity? Well, indium phosphide demand is going up significantly. And customer demand, therefore, follows it, of course. We don't think it's building a bubble because we think that the consumption for indium phosphide is going to be significantly accelerating. One of the first things I was taught when I joined AXT in 2014 was from the vice president of engineering. He said, Gary, always remember this. If you're going to transfer data using light, you have to have indium phosphide at the end of both ends of the fiber optic cable. So that's how it fits in. If you want high speed, you use light.
If you use light, you need indium phosphide. So there's only one significant competitor. It's Sumitomo, and we share the market with those guys. But we have better quality than they do, but they're a bigger company, of course. So there's a lot of proprietary process technology in what we do. We're the only player to have raw materials.
Our markets and fundamentals look excellent, and we remain hopeful to take our company public in China, so Charles, this is kind of an overview for the group, and then maybe you can see if anybody has questions, and I also want to introduce Leslie Green, who is our investor relations executive, and she's been helping us for over 20 years, and so very, very knowledgeable about a pretty complicated business model.
Do you know that China has export control on certain key materials, both germanium and gallium arsenide? Is indium also on the list?
Indium phosphide is on the list. As of February the 4th, 2025, indium phosphide was added to the export control list. For every order that we have outside of China right now, we have to apply for an export permit. If you look at our financials last year, we saw a bit of a dead space where we received no permits in Q1 and Q2. We did receive permits in Q3. We've been steadily receiving permits since then. Yeah, that's a good question.
This is export controlled. We do work very closely with our customers and also with the Ministry of Commerce in China to drive permits for export. Of course, it is a Chinese government-controlled permitting process. We have very little control and very little visibility on how that moves forward. Do you want to comment more?
Yeah, we currently consider the permit process to be one of our major risk factors. We've documented it in our filings with the SEC. In every one-on-one meeting today, we've talked about this with investors like you. So we're hopeful that the economics will work out and be superior to the politics and some of these concerns. But it is a risk. And it can impact us. But we're confident that it's going to be okay. Although I have to say, it's frustrating and creates a lot of extra work for everybody, including the Ministry of Commerce in China.
Do you have the alternative source outside of China on the raw materials?
Yes, we're in good shape on the raw materials. But what we don't have yet is manufacturing of indium phosphide wafers by AXT in another site. That did come up in almost every meeting again today with your community. And it is something that we've thought about and are thinking about what to do.
Can you perhaps maybe give some color on what % of your indium phosphide revenue is outside of China?
Probably today, it's probably more than 70%, 75%. Now, the Chinese business is growing. The Chinese indium phosphide business has tended in the past to be more of the GPON market. So we're looking at the passive optical network market, telecom market, fiber to the home, that kind of stuff. China has historically been a little bit later on high-speed optical transmitters and detectors, even though the majority of the optical transceivers are built in China. So a lot of the optical transceivers are built in China using lasers and detectors from non-Chinese sources.
We are starting to see some growth in that market as they develop laser sources and detector sources within China. I think we made a comment on the conference call in Q3 that we think they're a few years behind the rest of the world. But we were starting to see some growth in China. We're starting to see some acceleration on there as well, so I think we'll see some growth on the China side of the business, and I think we'll see some rapid growth within China, specifically on silicon photonics, in the next 12 months or so.
Just to be clear, you don't provide any epitaxial substrate?
No, that's right. We're just bare substrates.
Yeah, so the next step after our wafer is to grow an invisible chemical alloy on top of the wafer called the epitaxial layer, and that's a value-added step. We don't do that, but we sell to the epi houses as part of the step all the way to the ICs.
What is your market share versus most of the global indium phosphide market?
I'm going to give you some rough ideas back in 2024 because, of course, there were some disruptions in the first half of 2025, but we're typically looking at about merchant market. We're looking at about 40% between AXT and 40% for Sumitomo, so 80% between the two of us-ish, 80% to 90%, and there's another competitor in Japan called Japan Energy. They're about 10% right now, and then we've got some other smaller competitors that make up the few other percent that's remaining, so that's kind of where we lay in 2024.
We're currently expanding our capacity, so we are doubling our capacity through 2026. By the end of this year, we should be double our capacity from the end of 2025. We know also that Sumitomo is doubling capacity. But they've announced that it's going to take them somewhere between two and a half to three years to do that. We're in a really fortunate position that we have a building ready that was originally purposed for gallium arsenide crystal growth that we exited from a few years ago and relocated. So that building has been empty.
It allows us to repurpose that for indium phosphide at a relatively low cost and at a relatively short time frame. We also know that Japan Energy has announced several times that they're making increases in capacity. Again, they're about 10% right now. I would imagine to see them grow a little bit. But this market is growing. And as I said earlier, we're expecting to see a doubling in demand in 2026 and then another doubling in demand in 2027. And remember, that is only for scale out. So we're just planning now on what the next level of expansion is going to be beyond there.
So as we look into the future and we look to next year, I think we're going to gain more market share over Sumitomo as we build capacity quicker. And then really, the market share is going to really depend on how that market grows and which one of us can increase capacity quickest. I think that's going to be AXT.
Some of you may be aware that there's been some political moves between China and Japan as well, restricting the transfer of indium to Japan. So this thing gets more and more complicated in terms of the global strategies in place. But we can compete effectively as a more fast on your feet company, I think, than our Japanese competitor. And we also have much lower defect density than our competitor. And as you make the chip bigger, the defect issue becomes greater and greater. So we should be in the number one position to gain market share on that issue.
Can you speak to companies like Umicore who are extracting indium and other [audio distortion] ? To what extent do you partner with those companies? Or say a company like Umicore. Or to what extent do you see them as a viable alternative?
Yeah, so of course, we use indium. And I'm just going to focus on indium because we obviously use gallium and germanium and the rest of it too. We typically buy from China sources. And then we either own ownership or we have ownership in a mining company or a recycling company. And then we do a lot of the purifying of the process itself. So we do that ourselves.
Of course, we do deal with Umicore, especially on the germanium side. We don't so much on the indium side. But I'll make a comment about indium. And that is this market, the data center market, the optical transceiver market uses less than 10% of the world's indium. The majority of indium is used there. So it's indium tin oxide. It's a.
There, meaning the screen.
It is the transparent.
For those that aren't in the room.
Yeah. So it is a transparent contact for flat panel screens. So that's where the majority of indium is being used. It's being used at a lot lower purification level than us. So we do buy indium from multiple sources within China, outside of China. And then we purify it to the needs that we want. But at this moment in time, the indium supply is pretty straightforward. There's even a company in the U.S. that manufactures indium as well. So it's pretty open. China controls about 60% of the world's indium mining and production. So 40% of it is done elsewhere.
Your revenue last year, can you tell us what percentage of your revenue generated in China and what's the non-China revenue? And what's the difference this year?
Our revenue last year in China. I'm speaking from memory. It usually runs between 40% and 50% goes to China, so we have a strong position there, and I think that will shift over time as we grow in other areas, but I don't think this year is going to look a lot different, especially with the permit restrictions, which may prevent some shipments taking place that we want to do to the rest of the world.
What percentage of the revenue is from the U.S.?
The percentage of revenue in the U.S. is usually high single digit, pretty small.
And Japan's?
Japan's probably double digit for sure. But I don't remember the exact amount. Of course, this is all listed in our filing, so you can see it.
Do you have any insight into just in terms of the export licenses that were issued before that we saw a delay of? If there's any [audio distortion]
Yeah, let me ask the question over the speaker and have Tim answer the question. But the gentleman's asking, do we have any color about the nature of the permits, which ones are granted and not granted, and the timing? And we had expected more to be granted in Q4, even down to the last day of the year. And then they weren't granted.
Yeah, obviously, this permitting process is relatively new. So we're still trying to understand the landscape. When we looked at Q4 and we looked at the Q4 numbers, we really only had one data point to look at, which was Q3. What we can see from Q3 with trend analysis is that on average, permit time was taken about three months. But what we've seen in Q4 is that that average is probably the same.
But there's variability. We don't fully understand that variability. Is it regional? Is it customer focused? Or is it volume focused? We're still trying to understand what causes delays through this permitting process. But as I said before, the permitting process is not transparent. It's extremely opaque. We work extremely closely with the Ministry of Commerce. And we provide them everything that we can and everything that they ask for.
They often give us insight into when they think permits are going to come through. But even that doesn't always happen. So typically, what we do, and looking at Q4, as Gary said, there were some permits that were coming in that we expected to come in in Q4. We built, and we build ahead of time ready to ship. So had those permits come in even on the last day of December, we could have still shipped and recognized revenue.
But that process is not transparent. It is a bit of a black box. It doesn't always follow the same structure from order to order. And all of those permits we're expecting were repeat orders from customers that have already had permits before. So it's really not clear on what happens here.
Tim, comment a bit. You said at one of our meetings with the one-on-ones that it's not just the Ministry of Commerce as an agency, but there are other agencies that are part of this review process.
Right. So it goes through a multi-agency process. So it starts off at the Regional Bureau of Commerce. It then moves on to the Central Ministry of Commerce. They get involved with, I think, the Ministry of Trade. And then it goes to the Ministry of Defense. And then it goes back to the Ministry of Commerce.
So even though the Ministry of Commerce may think things are running through well and we talk to them, it may have gone off to another agency, which, of course, they don't have any insight to either. So I don't even think when you're looking at this permit process within the Ministry of Commerce in China, I don't think they have a whole lot of transparency here either.
The other thing I think we can say is that Leslie Green reported to me that she had an hour-long call with a very large investment fund based in Taiwan just recently in this calendar month already. They're very familiar with the permit process. They have a number of companies that they've invested in Taiwan that have reported to them that they're also struggling to get permits. So this is not something that we don't think is just focused on indium phosphide or just unique to AXT. It's two countries with lots of bureaucracy.
And our job as business people and entrepreneurs is to navigate through it as efficiently as we can. It's not something we can control. But it's our job to do the very best that can be done. We believe with our effort and their moving forward, we have to show our respect for these agencies that it's going to work out. In the meantime, for people like me, Tim, Leslie Green, it's a pain in the neck. But it's part of our job.
A couple of questions. The first one is in data center connectivity. Can you name some well-known companies that are part of your downstream customers or end customers [audio distortion] , essentially?
Yes, we can. We can do that.
Yeah. So we directly sell into the epi vendors. So we've got LandMark, VPEC, WIN Semiconductors, IQE, all of those kind of guys. And then we also sell directly into the laser manufacturers. So those are the guys that build the laser diodes and then sell them either into InnoLight or something like that. So the list is relatively lengthy. But it includes Lumentum, Coherent, Broadcom. Intel is in there, MACOM, Applied Optoelectronics, et cetera, et cetera, et cetera. Those kind of guys.
Then they'll sell either into their own system, their own pluggable transceiver, or they'll sell into somebody like InnoLight who's making pluggable transceivers that then goes into the data center hardware companies, the NVIDIAs of the world, and all of those kind of guys who are making the racks. So that's really how it flows through. So we're right at the beginning of that value chain.
The second question is, what was the historical pricing threat, and where capacity is doubling, how do you anticipate the pricing threat?
So we don't really see the pricing remaining relatively stable throughout most of the world and most of the technologies and most of the applications. However, there are some curveballs there. There are some applications and regions which don't fit the normal curve. Specifically, when we're looking at the GPON market in China, it's an extremely low ASP. When we're looking at consumer products, the ASP tends to be lower. And when we look at geographical regions, sometimes some of those geographical regions are on the lower side.
We're reevaluating that right now because, of course, capacity is in demand. So we're trying to reestablish a benchmark price across the board. We're not automatically going in and turning the price up. One of our competitors is turning the price up significantly with like 40% price increases. We're avoiding doing that across the board. We're trying to set more of a standard baseline so that it's a fair ASP across applications and across regions.
Since 40%-50% business comes from China, if you want to extract cash from China, what would the net look like?
If you want to extract cash?
From China.
What would the net look like?
Yeah.
What do you mean by net?
You have to pay certain taxes, tax repatriation.
Oh. Well, there are ways to bring money out of China. It cannot be capricious. You have to have an arm's length transaction, a reason. One example would be to pay a dividend to somebody outside of China. And you would have to go through the SAFE processes called SAFE, which is an acronym for something about a federal agency that handles foreign exchange. And so dividend is the first thing that comes to mind. You could also buy something from China. I mean, if a company, Tongmei, wants to buy something made in the United States or in Europe, then it's a similar process, bureaucratic.
But what you can't do is, let's just say, let's suppose we get the IPO accomplished in Tongmei. And they get, let's just say, millions of dollars. We can't just say, send us all that money. There has to be a business reason to justify bringing the money. But I would point out that since most of our operation still is in China, the bigger question is, can we get money into China to make sure Tongmei has enough capital to add capacity this year, right now? And the answer is yes, we can't.
But it's a good question. It is something that comes up again as part of these things that, as business people, we've got to deal with. And we're very experienced with it. We have a great team. And I'm confident we can always manage through those kinds of issues.
May I ask one more?
Yeah, please. Last question.
I think I want to try to ask the elephant in the room, data center connectivity. If you can guess the percentage of revenue that comes from data center connectivity, is it reasonable to assume it is like single-digit % of your total revenue?
Yeah, especially if we're looking at AI data center connectivity right now. If you look at the TAM for indium phosphide, I estimate that data center is somewhere in the region of about 30%-35%, with the majority of that still being traditional data centers. So if we're looking at data center TAM for indium phosphide substrates, it's probably on the low side. It's probably in those single digits, pushing 10%. But it's growing. And it's growing really rapidly.
Yeah, that's the genius of it. It's a great question because it's not big right now, which means there's plenty of room to grow. And that's what we think the market's going to do, is grow.
Yeah.
All right. Thanks for the sake of time. Please close with the management afterwards. We have to close out this session. But thank you, Gary.
Thanks, Charles.
Thank you, Tim, for your time, and I hope everybody enjoyed the rest of the conference. Thank you.
Thanks.
Thank you.