AXT Inc (AXTI) 2017 Q3 法說會逐字稿

Good afternoon, everyone, and welcome to AXT's Third Quarter 2017 Financial Conference Call.

Leading the call today is Dr. Morris Young, Chief Executive Officer, and Gary Fischer, Chief Financial Officer.

My name is Glenda, and I'll be your coordinator today. (Operator Instructions)

As a reminder, this conference is being recorded.

I would now like to turn the call over to Leslie Green, Investor Relations for AXT.

Thank you, Glenda, and good afternoon everyone. Before we begin, I would like to remind you that during the course of this conference call, including comments made in response to your questions, we will provide projections or make other forward-looking statements regarding, among other things, the future financial performance of the company and our ability to control costs and improve efficiency, increase orders in succeeding quarters, increase our competitive position in the market, our ability to meet market demands for our products as well as other market conditions and trends, including expected growth in the markets we serve.

We wish to caution you that such statements deal with future events, are based on management's current expectation and are subject to risks and uncertainties that could cause actual events or results to differ materially. These uncertainties and risks include, but are not limited to, overall conditions in the markets in which the company competes, global financial conditions and uncertainties, market acceptance and demand for the company's product, the impact of delays by our customers on the timing of sales of products.

In addition to the factors that may be discussed in this call, we refer you to the company's periodic report filed with the Securities and Exchange Commission and available online by link from our website for additional information on risk factors that could cause actual results to differ materially from our current expectations.

This conference call will be available on our website at axt.com through October 26, 2018.

Also, before we begin, I want to note that shortly following the close of the market today, we issued a press release reporting financial results for the third quarter of 2017. This information is available on the Investor Relations portion of our website at axt.com.

I would now like to turn the call over to Gary Fischer for a review of our third quarter results.

Thank you, Leslie, and good afternoon everyone. I'll begin by saying total revenue for the third quarter of 2017 was $28.2 million compared with $23.6 million for the second quarter of 2017.

Of our total revenue, substrate sales increased to $22.4 million compared with $19.1 million in the prior quarter. Revenue from our raw material joint ventures was $5.8 million in Q3, compared with $4.5 million in Q2.

In the third quarter of 2017, revenue from North America was 8%, Asia Pacific 68% and Europe 24%. Two customers generated more than 10% of revenue in Q3, and the top 5 customers generated approximately 39% of total revenue, reflecting again our diversification of both products and customers.

Gross margin in the third quarter was 39.5% compared with 30.8% in the prior quarter. Of course we are very pleased to see the gross margin improve. While no single factor accounted for the majority improvement, the following factors contributed.

Number one was favorable product mix. Indium phosphide in Q3 was at a record high for revenue, and germanium was also better than expected.

Number two, higher volumes across the AXT substrate university meant that overhead was spread over more units, and that helps gross margin.

Number three is a convergence of significant manufacturing efficiencies, including improvement in overall ingot yields and crystal growth, consistently high product quality and good raw material costs.

Number four is, to a lesser extent, we also had the benefit of a few other factors related to the three partially owned companies that we consolidated as raw materials.

So gross margin improved to 39.5% from 30.8%.

Total operating expenses were $5.9 million in the third quarter compared with $5 million in second quarter. Total operating expenses in Q3 included one-time expenses of approximately $500,000, consisting of $250,000 in special R&D expenses for our indium phosphide development program as well as setup fees related to our new facility.

We expect total operating expenses in Q4 to remain at this level, including approximately $500,000 in one-time items related to personnel and the end-of-the-year employee bonuses in Q4.

In Q1 2018, we expect to return to more normalized and sustainable operating expense level of $5.2 million to $5.3 million.

Total stock compensation expense for the quarter was $339,000. Operating profit for the third quarter of 2017 was $5.2 million compared with $2.3 million in the previous quarter. Interest and other income for the third quarter was a net charge of $820,000.

This net number consists of four categories. Number one, net interest earned of $122,000; number two, equity accounting for our unconsolidated joint venture companies, a loss of $266,000; number three, other items that add up to about $150,000; number four, a foreign exchange loss of $339,000. Exchange loss is higher than usual, and it's largely due to the fact that we transferred funds to our Beijing company in order to purchase the new site. And during the process of that transfer, the Chinese renminbi appreciated in value. As a result, we incurred a foreign exchange loss in the conversion.

The tax provision for the third quarter is 181k compared with 321k in Q2.

For Q3 of 2017, we had a net profit of $4.4 million or $0.11 per share. By comparison, we had a net profit of $1.9 million or $0.05 per share in the second quarter of 2017.

In Q3, the diluted share count was 40.1 million shares.

Cash and cash equivalents and investments closed at $78.3 million as of September 30. During the third quarter, net cash generated from operations was positive at $4.3 million. So that's a good number. This was offset by the purchase of the new facility. As a result, cash, cash equivalents and investments at September 30, were down from the previous quarter by $9.2 million.

Depreciation and amortization in the third quarter was $1.1 million, and CapEx was $15 million. And that includes the new facility.

Accounts receivables net of reserves were $20.9 million at September 30th, compared with $18.3 million at June 30.

Net inventory at September 30, was $40.8 million compared with $40.6 million in inventory at June 30. Ending inventory, which is pretty consistent consisted of approximately 52% in raw materials, 42% in WIP and 6% in finished goods.

So the financial review is completed, I'd like to turn the call over now to Dr. Morris Young for review of our business. Morris?

Thank you, Gary, and good afternoon everybody. Q3 was another solid quarter, contributing to what is expected to be a year of strong growth for AXT.

Once again we posted revenue growth in all of our product categories, including record revenue in indium phosphide. We continued to see a resurgence in new applications for compound semiconductor substrates, and we're pleased that our growth is coming from a diverse set of drivers with additional opportunities like 3D sensing still on the horizon.

Along with solid revenue growth, we also achieved positive gross margin improvement and outperformed our profitability expectations for the quarter.

The key to these results is that our products are well-positioned in seeing strategic growing markets. In addition, we're seeing the benefit of the operational efficiencies and the improvements that we began successfully implementing more than a year ago.

One of the most exciting area of growth we have seen in recent years has been in indium phosphide. This is a highly specialized material in which AXT has made considerable improvements, investments both in R&D for the [stringent] technical properties, as well as in customer support to assistant our customers in overcoming the technical hurdles in the development of indium phosphide-based products. This investment has allowed AXT to establish itself as a leader in the space in which technical barriers to entry are very high.

Further, the long-term increasing demand for higher bandwidth and faster speed for networking, and communications, lead us to believe that any phosphide will have a (inaudible) long lifecycle, in which we are just at the early stages.

One of the fastest growing applications currently is datacenter connectivity, driven by indium phosphide-based silicon photonics. This technology uses light of photon to move data at very high speed over the thin optical fibers, which is built directly into a semiconductor device.

Today, silicon photonics products are being deployed to connect switches within large datacenters in the future, as bandwidth to the servers increases, silicon photonics will be used to connect service. Leading global companies such as Intel, HP, Cisco, IBM, Broadcom, Infinera, Mellanox and many others are driving this adoption.

As the exponential growth in data traffic converges at the datacenter, hyperscale cloud provides providers such as Microsoft Azure, have publicly stated that they are deploying silicon photonics in order to drive scale for data rates of 100 Gb/second and beyond.

In the relatively short period of time, we have seen the datacenter connectivity applications becoming the largest contributor to our indium phosphide revenues. In addition to datacenter connectivity, silicon photonics is also being leveraged in a wide variety of applications in telecommunications, Internet of Things, metrology, consumer electronics and display, healthcare, sensing, autonomous cars, security and others. While demand from these applications today is small, the opportunity they collectively represent for the future business is encouraging.

In addition to silicon photonics, the passive optical network market is the other contributor to our indium phosphide revenues. As has been always well publicized, this market, particularly in China, has been weak in recent quarters, largely as a result of a inventory rebalancing.

We continue to take appropriate to a conservative view of it, but believe that its eventual recovery will provide additional growth opportunity for our indium phosphide business.

AXT supplies into a number of regions and customers worldwide and is well-positioned as the market improves.

And finally, a third area of focus for indium phosphide is telecommunications, where the current infrastructure upgrade cycle and preparation for 5G are providing opportunities in short haul, long haul and metro deployments. We believe that this application will be a market driver and contributor to our revenue growth for years to come.

In total, we are pleased with the continued growth of our indium phosphide business and with what we view as encouraging signals that our customers that demand will continue to be positive in 2018.

As a result, we're currently increasing our indium phosphide capacity at our Beijing facility in a meaningful way. We have always maintained that a certain amount of lumpiness in our revenue is to be expected, as we are selling into many emerging application that are working towards volume production. And also, we're in the early stages of it so that it could be seeing lumpiness.

However, we are seeing customers more proactively planning for future demand and working with us to ensure that their need can be met. For example, during this quarter, we will receive our first multimillion dollar long-term order for indium phosphide substrates.

While this order is not a guarantee of revenue, it does demonstrate that customers are viewing our product as a strategy part of their planning and a important part of their growth.

Now I'll turn to gallium arsenide. While the applications driving growth in gallium arsenide have changed meaningfully in the last several years, the potential business opportunity from this substrate is once again sizeable. This is a result of the unique properties of the material that makes it suitable for a host of interesting new applications. For example, [thinking] gallium arsenide is highly effective for use in solar panels and is also transparent enough that it can be used [under] the glass in automobile sunroofs.

Audi and BMW are using solar panels on certain new models to provide powers to the vehicle's climate control system fan without draining the battery even when a vehicle is turned off. In addition, Audi and Alta Devices recently announced that a partnership to integrate solar cells into panoramic glass roofs of Audi models to generate solar energy that increases the range of Audi electric vehicles. The first prototype is expected to be developed by the end of 2017.

Thinking gallium arsenide is also being used in solar (inaudible) to extend the flight time for certain drone applications as the material is lightweight and provide efficient solar conversion.

In addition, as we mentioned last quarter, gallium arsenide covers a spectrum of light that is unseen by the human eye, making it a ideal-suited material for applications that call for infrared light.

We're currently experiencing growing customer demand for a variety of other high-end applications for infrared light in which low EPD requirement limit a number of competitors that can meet the stringent specifications. These applications include virtual and augmented reality, retinal recognition and automotive sensing, among others.

3D sensing is the other application garnering much attention as a result of its use for facial recognition in Apple iPhone X, announced in Q3. While AXT is not participating in initial customer commercialization of 3D sensing technology, we view these as opportunity for high-end substrate manufacturers because the very strict technical requirements for 3D sensing provide a strong barrier to entry to new and or low-end players. Today, only 3 competitors, including AXT, are able to provide low-EPD substrate in sufficient volume for production [there].

We understand that our substrate meets the technical qualification for these applications and believe that we could see incremental opportunity in 2018, once we have demonstrated good progress in moving our gallium arsenide production line to the new site and can submit substrate for qualification.

We are on track to have the first 3D sensing substrate for customer qualification in Q1 for our new site.

Turning to germanium substrate, this segment of our business grew again sequentially in Q3, as the satellite industry continues its positive trend. The Satellite Industry Association recently reported in its annual report that small satellite used for observing conditions on earth are the fastest growing segment of the market.

We typically see a fairly consistent demand for germanium substrate used in satellite solar cells, but we have seen growth in demand throughout 2017. We believe that positive market conditions are likely to continue to provide opportunities over the coming quarters.

Finally, with regard to our partially owned raw material companies in China, revenue increased in the third quarter as many of the markets into which they supply are improving. In addition, the 3 that we consolidate into our financial statement collectively made a positive contribution to our financials. All 10 companies remain an important part of our supply chain strategy and we believe that as the demand for compound semiconductor substrate continues to increase, our holdings will provide important leverage for our operations, as well as opportunity for positive contribution to our overall financial model.

Now, before I close, I want to give you an update on the relocation of our gallium arsenide manufacturing to our new facility in the city of Dingxing, China. We are on schedule with our plans and pleased with our progress to date. As you may know, Dingxing is approximately 90-mile drive south of our current Beijing location. The site has approximately 140,000 square feet of existing manufacturing space and 50,000 square feet designated for offices and dormitories. And it also has plenty of acreage available for expansion.

Under the direct supervision of our general manager, we are actively preparing the new facility for the requirements of gallium arsenide manufacturing. And we are also in the process of installing and testing both new and relocated equipment.

In addition, we are working to complete our initial staffing and training requirement. I'm very pleased to report that the result of all these is that within only 6 months acquiring the facility, we already have our first single crystal ingots grown and extensive testing has been shown that they have a consistent quality level as those product produced at our Beijing facility.

As we have discussed, we're planning for a staged relocation of gallium arsenide throughout 2018, which means that we will produce substrates from both our current location in Beijing and the new Dingxing site for a period of time and then gradually increase production volume at the new site. We continue to expect to have initial qualification substrate late this quarter. And we're working with customers to schedule qualification beginning in Q1.

We are not required to move indium phosphide and germanium substrate production line at this line. Those substrate will continue to be manufactured at our Beijing facility for the foreseeable future. And over time, we believe we will evaluate our plan for possible relocation.

In closing, this is exciting time for our business. We see great upside potential from the applications across our product portfolio that we are already qualified into and supplying. In addition, new opportunities for incremental applications like 3D sensing, hold promises for 2018 and beyond.

In essence, the basis for our optimism about our business is that the application that we sell into have long, multiyear cycles, is a [view] that is shared by many expert and analysts. We're early in and we are well positioned for growth in these markets for many years to come.

Further, we believe that the relocation of our gallium arsenide manufacturing will allow us to expand our ability to support the increasing customer demand for our substrates. And we're pleased that our [operation notice] discipline is enabling us to drive improved profitability and cash generation.

This concludes my prepared comment. I will now turn the call back to Gary for our fourth quarter guidance. Gary?

Thank you, Morris. As we discussed, we're excited to see upside potential across our portfolio and believe we are well-positioned in many applications that will drive our business growth in 2018 and beyond. Coming off of a very strong quarter in Q3 and heading into a typically seasonally weak Q4, we expect to see revenue of between $26 million to $27 million in Q4. If we take the midpoint of our Q4 guidance, we are, therefore, expecting 2017 revenue to be approximately $99 million, which represents annual growth of 22% over the prior year 2016.

Given the expected Q4 OpEx levels that I discussed earlier, we believe our profit per share in Q4 will be in the range of $0.07 to $0.09 per share based on 40.1 million diluted common shares outstanding. Using the midpoint of this range, we're expecting an increase in our annual earnings per share in 2017 over 2016, an increase of 47% on EPS.

This concludes our prepared comments, and Morris and I would be glad to answer your questions now at this time. Operator?

Thank you. (Operator Instructions) And our first question comes from the line of Edwin Mok from Needham and Company. Edwin Mok, your line is now open.

Sorry about that. Congratulations for a great quarter. So first I have a question on the GaAs market. Morris, thanks for providing some color regarding the different opportunities and beyond (inaudible) traditional GaAs market. But just I wondered kind of maybe is there a way to kind of think about the size of these opportunities? There's a number of different application for GaAs. And I think you had mentioned IRR, for example, and obviously (inaudible). [Any way we can] kind of think about how big are these opportunities? Is it going to be as big as the traditional RF market? Or are they smaller? And any way you can kind of give us some color on that?

That's a very good question because you know gallium arsenide, of course, covers a lot of applications. Let's take the easiest question first, perhaps, is RF market. I mean, RF market was decimated by the SOI. Now we're not afraid to mention it and so the [PF] business was totally lost. But on the other hand as we move into 4.5G or 5G, gallium arsenide (inaudible) still needed. I mean, it's very much entrenched applications for wireless handsets. And so for LEDs, we mentioned that in our last quarter conference call we see a strong demand from high-end customers who take our gallium arsenide to make red LEDs as well as infrared LEDs. Although, the application of those market, we are still in the process of collecting all the data. And we even hear some horticulture, indoor farming, could potentially be a big market driver. But all of these applications, very diverse and takes a lot of time to collect all these data. But I can tell you one thing I share with you is one of our major customers using gallium arsenide semiconducting wafers is telling us they are building capacity big time. So if they're building, we should prepare ourselves to build that capacity to support it. We also talk about this drone and automobile application. I think it's really early in the [stage], although we've been working on this project for almost 7, 8 years. But finally start to show some real applications which could drive volume. So we're excited about it. But how big could it be? I really don't know, because it really depends about how many people wants to install in their fancy electric cars or to keep their car cool parked under the sun. So, but it's a very large European car manufacturer wants to put it into their car. We think this potentially could be very large market. And then, obviously, the ultimate is 3D sensing. We are very excited about 3D sensing. Although it may have a sort of slow start as we were projecting this could be a huge start. But I think all the analysts or the market study says that maybe has a slower start. But I think the future for this application is just great. I mean, I've seen many, many people telling me they're waiting to buy the next generation phone because they want to have this application feature in there. So we've been also pondering how big opportunity for the substrate is, okay. So I've done some calculation, and I hope you can correct me to see whether I'm right or wrong. I say the final device maker, if their addressable market is $100, our substrate addressable market is between 4% to 5%. So if they are $100 million, we're $4 million to $5 million. If they're $1 billion, we are $40 million to $50 million. That's for the substrates. So is that a reasonable answer, sir?

Yes, actually that's extremely helpful. Thanks very much. Just one clarification. You talk about VCSEL. You mentioned that you have a customer qualifying your substrate in the first quarter. Is that substrate being built in your Beijing facility? Can you kind of repeat that? I'm sorry. I missed your commentary on that.

Well, in our commentary said we're not playing in the 3D sensing market now. We don't have a purchase order in hand. But we are in contact with our customers. From the feedback we have with the customers, they're very pleased with our position, which is prepare ourselves for very large volume production. They're also telling us our quality of our product is great. However, some customer have a issue with us. They didn't want to qualify the substrate twice. So they said, well, why don't you supply the good material that you can produce on the new site and we will start to qualify that. And so we said our estimated new product coming from Dingxing, the new site, should be at end of Q1 of next year. That's why I'm saying we should be able to submit qualification wafers in Q1 of 2018. However, we do have enough inventory in our Beijing site of this low-EPD, the good quality material fit for this 3D sensing applications. So if the customer are willing to take the old site manufacturing to qualify into their product, that could perfectly be fit for their requirement.

Okay, great. Actually that's extremely good clarification. Second question I have is on InP. I think last quarter you guys talked about roughly around half the InP demand comes from the PON market, you think roughly around a quarter comes from the silicon photonics and a quarter come from other applications. Is there a way you can update us with that number? Is it still roughly around the same mix?

No. I think silicon photonics is outrunning PONs, especially PONs market is having its problem right now. So it's either inventory or -- so the way I view it is, is PON market is probably if you look at if there's no inventory overhead, it should be a very large size market. But the growth rate is relatively small. My estimation is maybe 10% to 15%. However, silicon photonics coming from a smaller base, it has a much greater growth potential. So, so far, as far as we can see, you know, we said 2 to 3 years ago, silicon photonics mix in our indium phosphide revenue was almost 0. Now is the biggest contributor. However, it's difficult for us to really put a number down because lot of times we don't really know exactly what are using it for. But if we only count the one we're sure about is silicon photonics, we only count the ones that we're sure is PONs market, silicon photonics definitely is bigger portion now and it shows bigger potential of growth, too.

Okay, great. Actually, that's helpful. Last question I have is on gross margin, Gary. If I did the math correctly, in the fourth quarter, you expect gross margin to moderate, this 39.5% you print on the third quarter. Is that correct? That's the first part. Second part is, as you ramp the new site, do you expect to have like some additional fixed cost absorption and, therefore, put some pressure on gross margin? If so, any way you can kind of quantify that?

Okay. In regards to Q4, we haven't really said a specific number for margin. But I don't think that we want to say that it's going to be equal to Q3. Q3 was very strong, very good. The sun, the moon, and the earth lined up and it was very, very positive. But I'm glad to get those from time to time. But I don't think it's prudent to have people that run models to be modeling at that high level. I think somewhere in the mid-30s, plus or minus, is reasonable. And Morris and I feel that we should be striving internally for something more. But it may not be achieved, so we should help all of us understand. Q3 was very good, but --

And as always, want to put the pressure on our CFO, deliver better margins.

In regards to your second question, in theory, I think there will be some duplication of cost. It's going to be hard to not have that. However, I'm not sure it will be substantive. So let's take the assets first. So the actual facility, the useful life depreciation cycle, is 27.5 years. So that on a per -- if you spent -- I mean, the extra facility costs won't hit the income statement in any harsh way. Furthermore, as we do move out of the Beijing facility, at some point then you will stop depreciating the facility because it's no longer in use. Not the entire plant, but maybe, for example, a building. So when you no longer depreciate it, then you're just depreciating at one site again. So maybe there's a short period of crossover. But then what you move to, Edwin, as you're familiar with, but for the whole listening audience, that particular building, if it's not depreciated, then it's subject to the impairment test, which would basically be is your book value greater than the market value. But given the reason for relocation, which is that the municipal government of Beijing is basically occupying Tongzhou, the region that we're in, the real estate market already is responding to that and will continue to respond, we believe, reasonably upward. So we don't think that we would be subject to an impairment write-down. So that covers the plant kind of things, 27.5 years. Maybe some quarters of crossover, but it's a relatively small number. Then let's talk about equipment. So in theory, there would be 0 equipment increase because we have enough equipment that we can move it pretty quickly. We proved that when we had the - I hate to mention it - the fire back in Q1. So our guys within days moving equipment out of that building into another building and refitting it and retooling it, and they did a marvelous job. So, however, there will be some increased equipment depreciation because the volume expectations have enlarged. So we will take advantage of old equipment, but we will also purchase new equipment, not because we're relocating, but because we need more equipment because the volume is higher. So we're already feeling pressure in indium phosphide. This past few quarters we felt pressure in germanium. We've had some slight pressures in gallium arsenide, but we expect that to become a little hotter going forward. And that's before even 3D sensing. So we remain confident we'll be playing with 3D sensing, and that's part of the reason we're planning on buying more equipment. So to summarize, the building stuff I think will be -- there is going to be some [double-off] charges of depreciation, but I think it will be for 1 or 2 or 3 quarters and it won't be very noticeable. And on an equipment, in theory, there would be no more equipment except there will be because of adding capacity, because, let's face it, we're already ahead of what we thought -- if you go back to when we planned 2017, back in December and January, we've exceeded what we expected. So that's going to put pressure on us for equipment. But I'll take advantage to give you sort of a commercial about what we think about ourselves as entrepreneurs and business people. We are very good at adding capacity. We've proven that in indium phosphide. We've gained market share well beyond what the market has allowed because we could move faster than our competitors. And so we're doing that again with indium phosphide now, and we expect to do that with gallium arsenide.

So let me add a few more comments on the increased expenses. Now we're operating 2 sites and obviously we need people to travel on the new site, for instance, (inaudible) facility managers. We have our old facility needs to be maintained and the new facility we need to hire new people. So there will be added expenses. I mean, we may need to hire new people as operators at new site and we need to train them at the old site. So that could be some we didn't see. But those increased costs are minimal, okay. And, but what can come out of it is that we will be then very well prepared if the market plays out to be what we expect it to be, a huge opportunity coming to the '18, '19, and '20, then we're the best prepared company. We have a new facility. We have increased capacity. And we have now more trained operators to help us to produce all these great products we'll be making.

And I know most of the listeners are aware of this. But when you talk about additional human resources, the price for that is significantly lower than you would imagine if it were to be additional operators in the United States. So, yes, we are going to need to hire new operators and train them, and we've already begun to do that. But the actual dollars spent are real but they're relatively small. Next question?

Thank you. And our next question comes from the line of Hamed Khorsand.

Just want to ask you here, just given the commentary you're talking about, is all the seasonal weakness you're seeing right now coming from indium phosphide?

Hamed, say that question again, please.

So your commentary outlines a lot of strength in gallium arsenide and in germanium. So I'm trying to figure out and assume that the seasonal weakness you're talking about for Q4, which has been natural in the past, is it all just coming from indium phosphide this time or is it spread out?

I think in our budget, indium phosphide will come down a bit. Germanium, I think will remain strong, although we have one customer taking a little bit less material because of Christmas. I think gallium arsenide, surprisingly, it will be not as strong as Q3. And ask me why; maybe they taking too much inventory from Q3 and then they have to take a breather, yes.

So indium phosphide will be down. Gallium arsenide will be down. Germanium will be flat. And raw materials, we're not sure about; it's harder for those guys to project this far in advance, so.

Okay. And then just trying to figure out lead times. Q1 has always been a seasonally soft quarter for you guys as well. So are you guys seeing extended lead times? Or are you seeing good enough demand where you think that it comes back and Q1 could be abnormally strong?

No, Q1, I don't think we're normally soft. I think Q1 should bounce back. Q4 has two reasons; one is the holiday season towards the end of the year, and we also have a national holiday at our factory. We close down for 10 days.

In China.

In China, yes. So the lead time-wise, I tell you, it's also depending around customer-specific product. For instance, for the gallium arsenide 6-inch product, the lead time is very long. I mean, we have to almost -- they're knocking on our doors to ask for more material to be delivered. And but then in Q4, they do tell us, hey, we got enough inventory, slow down a bit. So you have to worry about comparing to a very strong Q3, in Q4, you take your slack. But if you take that number and compare to Q2 or Q1, it's still a great increase. That's the way I would describe it.

Okay. And last question is, how long before your crystals get qualified from your new facility?

How long a crystal will be qualified, what --

For a customer qualified on the new site, how long will that take?

Yes.

Well, it all depends. I mean, some customers -- well, one of the customer, Japanese customer, by the way, they tell us that it will take 12 months. Yet, another customer, a European customer, told us 2 months.

So the smaller customers, Hamed, will do what I would call a paper qual. We will provide substantial characterization data from wafers from both sites and, presumably, we believe that they will look the same. And a smaller customer will sort of check the box and say, okay, we're done. We qualified. It could be a day. And then the customers are actually going to touch and feel and examine new substrates and compare them to old substrates. It will depend on how compelling their interest is to qualify us. If they're a fan of AXT, if they want to be in the AXT universe, they're going to be motivated to go quickly. If they're at the opposite end of that spectrum, then it might take a longer time. So it's going to be different for each customers, so.

I would agree. But, Gary, I think you also have to take into consideration we believe the market is really very hot.

Yes.

And there's not a whole -- you can't by gallium arsenide wafer like candies. If you take time to qualify AXT substrate, I mean, that's also to your detriment. Do you worry about --

If you take too long, yes.

Right, if you take too long.

Yes.

But some of them, they do have their quality standard, they're very rigorous. So we are in communication with each one of them and asking them, what's your requirement so that we can plan ahead? Some of the customer tell us it will take 12 months. So then we make extra effort to make sure that we have inventory built for them so that it will last that long until they qualify.

So I think what we're saying, too, is that it's fortuitous and nothing to our making that we [cause]. But this qualification process is going to be taking place in a market that is relatively strong, and that's going to provide compelling motivation from the customer base to do it more quickly than they might have.

Yes. Let's face it, there are only 3 supplier in the world. One of the reason I told them that you should qualify us quickly is that if you want to switch to the other supplier, you need to take that much long time to qualify them too.

Okay. Hamed, anything else?

No, that's it. Thank you.

Next question, please.

Thank you. And our next question comes from the line of John Sepenzis (sic) [Thomas]

Thanks for taking my question. Congratulations on another great quarter. I'm wondering if you could just give me any kind of update on [NGAP] and 5G, if you're testing that at all, because I know a lot of people are pulling in 5G cellular plans are to have networks up in late '18 and early '19. So I wonder if you've been talking to anybody about potentially using indium phosphide as an alternative to gallium arsenide in cellular.

Yes. I mean, that's a very good question. Actually, maybe 3 quarters ago we were very gung-ho and proposing, perhaps, the next generation cell phone will need indium phosphide to operate at 28 GHz or at high speed. But now our enthusiasm sort of cooled off a bit. And mainly because one of the large cell phone device maker said, well, it's probably not going to launch until 2022. And the reason why they said it's not going to happen sooner and it will happen later is because they say, well, the first generation cell 5G is going to emphasize on high-speed receiving end. In other words, the mobile phone all they need to do is receive that 28 GHz speed frequency signal on the mobile phone. So if it is only on receiving end, perhaps they can find a better solution other than indium phosphide. Because, you see, indium phosphide has great advantages, it not only can deliver very high power, high frequency, better linearity power signal to the base station, and also needs to operate at a very low voltage because cell phone, all operating at 1.5 volts. But if you only need for it to receive, then the urgent need for indium phosphide sort of diminishes, okay. So we hear a lot of demand for the, let's say micro cells. But those micro cells, because they are powered by AC, the house current. So they can operate those high frequency devices with silicon carbide, gallium nitride-based devises. However, so I think the indium phosphide opportunity probably will be pushed out for a year or two, okay. And so that's sort of a lost opportunity. However, I will say the other great alternative for 5G to come is really tying up all these micro cell together. See, if you have a lot of these micro cells, the micro cells could be 100x the number of base stations like what we have for cell phone tower today. And you need to collect all the cells together, and they're feeding to the central station as well as into data center. And so for indium phosphide, if you are going to use so much more data and so much -- sometimes described almost 100x or 1,000x the data rate, then you need indium phosphide on the fiber side to get all the cells connected together. And also the connecting back to the data center as well as taking the data and the [3D packing] to individual cells. So I think our play into the 5G is more than just a mobile cell phone market. Although, that market is delayed for a little bit, but we're going to be kept very busy if this 5G start to launch.

Great. Thank you for that. And then it sounded like from your commentary that the Chinese PON market kind of got a little bit softer in the quarter and is looking a little bit softer than you thought maybe. Directionally you thought at the end of Q2 it was starting to improve. So did I hear that correctly?

No. Actually, I think the way we describe it was that in 2015, let's say the PONs market, when we saw it was, let's say 100. And in 2016, it dropped down to something like 20. And in 2017, we believe that we said it's recovered to around 50 to 55. And that market we still see that's around that level. It's not improving. It's not going down. But I also saw one report saying that is all because of inventory correction. People build too much inventory, they want to digest that inventory. So it depends upon what you think 2018 is going to be. So as far as I'm concerned, I don't believe this market can be down forever, because China wants to launch 5G as well. And they're going to build a lot of data centers, which every time you connect a data, fiber to some kind of device, you need a PON.

Okay, great. And then in terms of the new facility, is that all GaAs right now? And are you going to add silicon germanium or indium phosphide any time in the near term? Or is that something that you do like a year from now once you've got the GaAs transfer?

Yes, I think three years from now we probably will consider moving a lot of things more there. But I think the first phase, Tom, is gallium arsenide only.

Trying to keep it as a single focus, gallium arsenide focus.

Got it. Thank you.

Next question?

Thank you. And our next question comes from the line of Jon Fisher from Dougherty and Company.

Two questions. One, when you look out into 2018, given the strength that the business has exhibited this year and the broad-based strength, is there anything, any end markets or specific product lines that you would expect that you're concerned about potentially slowing from a growth standpoint in 2018, that may be a strong contributor here in '17? I've been thinking in terms of risks, just everything's going really well right now and just trying to think in terms of the other side of the coin. Any potential risk or concerns that we might want to keep at the front of our mind?

We could give you a product line by product line. For instance, germanium, [several] launches are lined up.

Already scheduled, too.

Already scheduled.

Yes.

So I don't think it's going to slow down. Gallium arsenide, at least from the LED side, our customers are showing as they are expanding their capacity. So I expect they to buy more from us. But on the other hand, I also worry about are we going to lose market share to our competitors? I don't think so, but that's one potential worry. Indium phosphide, we've got a large order, so we're preparing for that. And I think the question is, is PON markets and China market is going to recover, and that should layer onto what is already into the big order we already received. And gallium arsenide wireless, maybe that's a potential negative. I mean, because gallium arsenide device maker are telling me that they are very, very busy. So some of them are telling me that they're dropping out on the low end, it is less profitable, they don't want to do it no more. Since we're selling real estate, they don't do it, it will impact our business. But on the other hand, you can say 4, 5, 6 of them, they're all growing and if so one of them is not doing so well, so be it. And 3D sensing, I really think this is big judgment. I'm hoping for the homerun, but maybe it's going to be later than sooner because the technology's so attractive, their entry to barrier is so high. And we shouldn't be looking for short-term gratitude, and we think this is going to be a long play. If we prepare ourselves for it, it should be a long growing market and I think the time for it to explosive growth maybe is 2019.

Okay. So without expecting any contribution from 3D sensing when you look out to 2018, you're going to grow --

Well, no, no. Wait. We do expect something.

Well, okay. Well, I think most people when they model, they're taking the conservative route and they haven't put anything in their models yet for 3D sensing. You're going to grow the business from a top-line standpoint about 20% this year, low 20s. Is that a comparable growth rate for '18 that you think you can hit? Or do you think there will be much of a deceleration from 20% to low 20s percent revenue growth '18 versus '17?

We are going to talk to our sales boys for our next year budget. But somewhere around 20% is what we're trying to drive. But obviously, we need to build that model from the bottom up, each [individual] customers.

And then final question. I don't know how sensitive the order is. But any extra detail you can provide on the multimillion dollar order that you received in indium phosphide? End market? Any description or any details or nuggets that you can put around that order?

Yes. First of all, we are very excited about that order. Substrate maker, we always say that we live quarter-by-quarter. They have the opportunity to cancel at any time. And if they say stop, we have to don't deliver. But this is a good indication that a customer, they're planning for their big volume increase. They worry about their supply line. So they talk to all customer, and then our customer tell us, hey, you better get yourself prepared and there's a big order coming. So I think this is -- I mean, although [somewhere] we do negotiate the annual capacity updates as well as price negotiation every year anyway, but this is a unusual (inaudible).

Yes. And it's more formal and firm than just giving projections.

No, no, no. Every year --

Yes.

But this is formal, but nothing's firm.

So I think the way we view it, Jon, is it's an interesting data point that could be suggesting that the customer base is becoming more sensitive to supply and to make sure they've got it line up, which would mean that theirs own demand and the customer's demand is driving that. So we hope we're right. We think we are. And we have to wait and see, yes.

Yes. I mean, so the question is, how do we build it into our budget? Also, how do we plan for our capacity expansion?

Right.

So we usually do conservatively in terms of budgeting ourselves, revenue protection. But we do aggressively for capacity expansion because we want to be prepared for it. And we believe that we have that entrepreneur spirit in ourselves so we can capture all this upside. So what are the consideration, I think is -- I mean, I hope I'm proven to be right, is that China market cannot be down forever, sometime it's got to come up. And there's not a whole lot of substrate supplier in the world.

So any hint or clue on the addressable end market that this order would be categorized under?

No, no hint or clue.

Okay. Thanks.

Next questioner, please?

Thank you. And our next question comes from the line of Richard Shannon from Craig-Hallum.

I'll just throw out a couple quick ones here. First of all, in indium phosphide, what's your customer base in terms of breadth look like for silicon photonics? Is there still kind of a single customer driving that? Or have you seen some breadth here recently? And/or do you expect some to come next year?

We always have multiple customers, we always have. I mean, we're also developing new silicon photonics customers. But I think, Richard, I think what's interesting, I think is some of the silicon photonics uses more substrate to achieve the same thing versus the others. So, but because we sell just substrates, so it's very difficult for us to tell who -- well, we think we know who is winning as far as we're concerned. We also developing into relationship with others, we're also [serving] to their market. So we think silicon photonics is a great alternative for indium phosphide.

Okay. Second, probably last question here on gross margins. Gary, you gave us some moving pieces on the third quarter number and you mentioned some favorable drivers in raw materials. Is that a one-time impact? Or how does that translate going forward? And specifically, is there any comments about the pricing environment for gallium as in terms of how you see that trending going forward?

Well, the raw materials have been trending down. But now we're beginning to see them tip back up. So we have a hedging model built into the business model, which is our own substrate gross margins will be detrimented if raw material prices go up. However, our partially owned companies in China that are in the raw materials business will benefit from that. And so hopefully there's some offset. But, yes, I think Morris can give some more color. But basically, raw materials generally are tipping up, we think, so.

Yes. Our JVs are doing very well. We have a JV who is just servicing to the OLED market. I mean, they are seeing big orders. Although, compare AXT as a whole, they are much smaller.

Right. That's fair enough. I think that's all the questions for me, guys. Thank you.

Thank you. And our next question comes from the line of Dave Kang from B. Riley.

The first question is just a clarification. What was the stock compensation?

$339,000.

$339,000.

Got it. And then for fourth quarter, how should we model CapEx? Is it going to be down quite a bit or --

Well, we don't have a new facility to buy yet. But we'll --

So it's going to be like low million --

I don't have a (inaudible) for that. I mean, I'm unaware of anything that's going to be as large in this quarter. But there will be capacity growth in indium phosphides. So there's going to be furnaces for that, equipment.

Yes.

But probably it will be down because we're not going to spend $15 million in furnaces, so.

Yes. More of the strategy I've been also telling investors is that we're buying extra 30% new equipment in preparation for the move. So all we want to do is buy the new equipment and have it set up in the new facility and have them quantify once they are running, then we can start moving out at 20%, 30% of the new furnaces and old equipment from Beijing site to the new site. So I think in the next year, I think the CapEx should increase. I mean, other than buying the facility.

Right. Yes.

Okay. And then just for questions on indium phosphide. So is indium phosphide bigger than GaAs right now?

Well, compared to individual, [semi-insulating] or semiconducting, yes, it is, by far, the largest.

No. I mean combined. Combined.

Combined, no. No. Combined, gallium arsenide is still big.

Got it.

But it's also, you've got to give credit to gallium arsenide, they're growing very well, too.

Yes. Yes.

Got it.

It's sort of woken up, so, yes.

And then on the multimillion dollar order, what about in terms of a shipment timeframe? Can you give us color on that?

Well, they're telling us it's between 10 to 12 months delivery.

Got it. And then my last question. In terms of a customer concentration for indium phosphide, so what will be the top customer in terms of a percentage of indium phosphide revenue? Would they be like 50%? Or any color on that?

Yes. We have many customers. But also depends upon time. I mean, some of the customers, when we've got them into big -- so I don't think we can say who's the largest and hold that [pose] for a long time.

Well, what I'm trying to understand is, whether this indium phos -- or silicon photonics is really driven by one customer or multiple customers?

We don't give that detail, David. But the answer is it's numerous customers.

Numerous customers.

There's not a huge concentration that's -- there's certainly customers that are larger than others. But as we said 2 quarters ago, we have a sort of new bucket that we talked about with indium phosphide, which is other small orders because so many companies are jumping on the optical bandwagon and doing optical development of products. So they're all buying indium phosphide, so.

Thank you. And I'm showing no further questions at this time. I would like to turn the call back to Dr. Morris Young for closing remarks.

Thank you for participating in our conference call. As always, please feel free to contact me, Gary Fischer, Leslie Green directly if you would like to meet with us. We look forward to speaking with you in the near future.

Ladies and gentlemen, thank you for your participation in today's conference. This concludes the program and you may now disconnect. Everyone have a great day.