Phison Electronics Corp. (8299) Earnings Call Transcript & Summary

Hi, everyone. My name is Jeffrey Ohlweiler. I'm Head of Research at Macquarie Research. So very pleased to be hosting Phison CEO, Khein-Seng Pua, for their third quarter earnings. So we're at a very interesting time, I think, in the semi and memory cycles, and I look forward to Khein's words wisdom. So with that, I'll pass the mic to Khein's for the overview. Khein's, please?

Thanks, Jeff. Good afternoon, good morning, ladies and gentlemen. Welcome to our third quarter's earnings call. There we start, today, I think we prepared over 30 pages, so I need to move a little bit faster. Today, we also likely to bring you some more new transformation to share with you. We're collecting the memory companies since Q1 this year. Based on their public information, the Q1 total loss, NAND DRAM most likely was $10 billion. CQ2 was 9%. CQ3 up to now is minus 6%. So we can see the gap is getting smaller. So I think this is cosigned. And you also realize for the market that NAND company, they're cutting the production and the price recently bounced back like over 70% within 3 months. Let's move to the Phison product. This is our Q3, our pie about our business. So 2 things to share with you. Controller in CQ3 increased some portion just because [indiscernible] company, our NAND company, their bank control from us, they are safe some inventory adjustments since last year's CQ4. But inventories start to move and after the Samsung announced to car production demand coming back. So start from August, September, the control demand is a little bit coming back. So our ratio in controller 3 is increasing from 26% up to 29%. And in consumer portion, actually, we increased it a bit because CQ3 is moving to the Christmas season. And also thanks to Samsung saying the production cuts, we are taking some more advantage, revenue increase in the consumer. And by the side, mainly smartphone and PCOM also, the demand coming back, we also gained a share. So moving from 9% up to 12%. The only slow portion is industrial. I think you also realize on the market, industrial is just quite slow. Maybe industrial were big by next year's CQ2 or CQ3. But overall, Phison we’re good because we have a good diversifier. So we can foresee the controller is going to increase. The modules for embedded phone gaming and for consumer also going to increase. So we are happy about our revenue growth in the Q3 and very healthy. And a total debt, I think you already get from public CQ3 revenue is close to the – the revenue, CQ3 is a $12 billion and gross profit is $3.9 billion. And the good thing is gross profit Y-o-Y and Q-o-Q, we are improving. So we also see the cycle back to the normal. I think we are able to perform much better than the last 2, 3 quarters. And this is our [indiscernible] trend. I think we are more suffered was February to May. I think we passed the most difficult period. So CQ3, we bet to our growth. I think it's around 20%. CQ3, we are working super hard. Hopefully, we can have more growth. And next year, I think we are very optimistic. This is gross margin. Our CQ3 margin is 32.2% is the third highest in a single quarter. And we have the reversal from our write-down inventory 0.7%. So CQ2 last quarter, we are still aware of inventory, but CQ3 we’re market start to bounce back. We are slightly gradually to write down our inventory route. So I think this is helping to our gross margin in the coming few quarters. This is earnings per share. Earnings per share is 4.32 in [indiscernible] we are growing, but we also like to report to shareholders what happened in this portion. First, our associate account using equity method [indiscernible] causing global, return to profit at CQ3. The CQ2 earning call, I am already saying that they are able to turn to profit. So I think it's good to us. And also in CQ4, I can see they are start to enjoying their much better profit. And one thing we like to highlight, Phison keeps saying that we are heavy investing into our R&D activity. We are tapping a new controller, PCI-Gen5 7-nanometer happened by September. And that one tape-out effect, Phison EPS 1.85%, so you can know how expensive this investment. But anyway, this happened by September. And the one is about the share-based payment actually is a treasury share. First time, we bought the treasury share by last year, I think it's a CQ4. And this CQ3, we exercised some portion of the treasury share to our executive. And this also affects the EPS is 1.36. So this is the product about our earnings on the book is 4.32%. But overall, I think we are back to really normal like year 2021, '22, first half. We also foresee see CQ4, the revenue and the profitability will get much better. This is our some income statement. We have a few highlights here. The gross margin, actually we still keep around 32%, which is the third highest in our single quarter. And the OpEx we need to highlight actual OpEx, just as I mentioned, the mass cost increasing around TWD 0.45 billion and the share-based payment was TWD 0.34 for exclusive and R&D activity. And revenue-wise, we are growing. Gross margin will keep tread and overall earnings per share is improving. For the first 3 quarters, 9 months, again, the OpEx is almost similar to last year. The reason is we have a new tap out of 7-nanometer and also, we start to hire new engineers to prepare our net growth. So overall, the earnings per share, we are growing by quarterly. And also happy to see coming few quarters, we're still able to maintain the growth trend. In the balance sheet, the cash flow wise, we still keep really healthy. Inventory is -- actually, we start to suffer some NAND flash component under supply. So waste range, 2 months ago, supplier came to us [indiscernible] to buy more. Now is the same person go to them to ask them, “Hey, please share more to us.” But anyway, we keep valuable relationship, I believe we're still able to get the more efficiency supply from our suppliers. Overall, this is our balance sheet. As we also disclosed our non-chef report. Based on the comparison, this is our operating profit. The thing I need to highlight mainly is our treasury share to our employee happen by CQ3, cause us TWD 342 million. Back to NAND [indiscernible] the operating profit, we improved from 4% to 7%, and the net profit improved from 4.22% back to the 9%. This is one -- we also agree we need to keep heavy investment into our human resources. So we start to use our treasure share as a compensation to our employees. We still maintain high portion of treasury share on hand. We only assetized 1/3 of our Hong Hanh treasure share. This is about October revenue is $5.15 billion. It's up 3% in MOM. And overall Y-o-Y is down 27%. And remember, first half Y-o-Y was minus 40-plus percent. We are doing our best to reduce the Y-o-Y to like 10-plus percent by end of this year. We keep working very hard. And the demand in the market is still very, very strong. Today, I'd like to share some concept idea with you. And also, we'd like to share the clearance of our new transformation, how to grow pricing in the future. This is -- today's actually is our key point. For so many decades, 2 decades Phison after IPO, many analysts, fund managers, investors keep complaining, "Phison you are building modules, modules very with low modules less attractive to the investor." But we made a study about what the so-called modules. Actually, Phison’s the modules, we are starting from the chip design, IC design, system design, firmware software design as our foundation. Then we turn to business mainly to the major business. Why? Because it is able to add our value, our ASP and also our gross margin as a tall amount. So we keep working hard for 2 decades. But unfortunately, we're still not able to convince most of the analyst, shareholder that we have a better value in the net industry. I think the 2 examples to share with you. These 2 companies, they also start with a chip design, but the revenue is not coming from chip design. It's coming from the products, from the modules. And this to prove that chip design, if you are in the small term of a business, the investment today just a 7-nanometer tape-out will cost Phison is close to 2 anti EPS as a expense. This is too heavy. By selling controller, I don't believe we are able to take our ROI back within 2 to 3 years. If I'm not able to tap back 2 to 3 years, then I'm not capable to keep the investment. Next year, we have several 7-nanometer chips need to be kept out. So we need to find a way how to unpack our dollar amount profit. So these 2 examples, A company and company to show you selling a module, it's not a betting, selling module is helping you to get a more ASP and gross margin in total amount, but you need to have a foundation. The foundation is chip design. So Phison, we are fully exercised our chip designs to the modules business model. What's the key value here? We try to put the value add into the system with our design capability. At the end, what we gain is ASP improved, gross profit improved, total month income improved. Later I will show you more about the time of each industry, then you may more agree what Phison, our transformation turn to the much positive direction. Phison, again, since 2 decades ago, we are saying that we are not just a pure controller house, we and a pure module house. We are the module house, with the add-on value capability in the IC design. Phison we are going to have our internal slogan to our all executive and our staff. We are going to have a foundation that accelerate innovation. What is innovation? In these 4 years, for the press release from public concession broad, you may be able to read Phison, we announced our image design service. Recently, we have a partner from a U.S. company that announced the new Gen 4 performance SSD, which get the best positive review from the average reviewer and the PC company. Net controller is a follow with so-called imaging plus design service. We also provided a cutting-edge technology. So recently, we have some new concept on the AI and adaptive they try to build the NAND into the GPU training systems, and we also build a ecosystem in the EV industry application and gaming. This is our innovation. But why Phison capable to deliver from nothing designed to the products because Phison we have very strong foundation we built in these 2 decade. We keep investing to the R&D activity. We keep working very closely and alias with our main partners, system partners, gaming partners, EV partners. So Phison is going to transform ourselves from pure design module house to the so-called value-added modules, but we have ICVs capability, which we try to learn from today's GPU company, the phone company. I believe with our keep heavy R&D investment, we can make it, not really soon but in a few years, we can show you Phison are more capable to have more service to our NAND supplier and also to the systems users. We are going to try to grow Phison to make ourselves big. Today, board meetings, we approved 2 motions. The board members [indiscernible] meetings approve Phison to make a prepayment to one of our NAND vendors to secure Phison by 2024 NAND supply. Why? Based on today's market research firms, TIFRS was in our supply since actually CQ2 2022. But based on the forecast, start from second half of '24, maybe 1.5 years, the TIFRS would be in undersupply, and the gap is not just a few percent, maybe close to the 2 digit. Phison we need to grow our business. We engage with the many system company, we need to have secure more than supply. So we approved to made a prepayment to one of our NAND partners. Second, we learned in this last down cycle, we accumulate a lot of NAND flash, but I can share with you by end of July, we still suffered the high inventory. By end of September, we have suffered every NAND component shortage. Why we shortage? Because after September, we be to the next supplier, they are not willing to ship as many as we want to Phison. So in this 2 years' experience, we learned, we need to have cash to keep our buying power and also keeping our inventory. And also, we need to have more test to support our R&D activity especially next year. We have many 7-nanometers going to cap out. This 7-nanometer are requested by a lot of system company and also by the NAND company. We are going to integrate our AI engine into our SSD controllers based on 7-nanometer. This is a heavy investment. Of course, the customers will pay some portion of [indiscernible] but we need to increase more cash portion to help us to grow Phison our business. This CB is going to apply by today after board members approval. And I believe with this CB to make the Phison stronger in finance, will help license the 2024 and 25’ business to get a strong growth. Again, as I shared with you, July, 3 months ago, 4 months ago, we suffered low inventory, September, we suffered shortage in our inventory. So we learned that we need to accumulate more inventory when it's a time cycle to help Phison grow our revenue, our profit and our influence power in the system modules business. And for many research report, almost everyone agreed second half '24 will be shortage. Phison we wish we are able to collect more inventory by coming CT1 to help us to have a growth in the 2024 and '25. [indiscernible] highlight. Today, I think this is our new segment to share with you how Phison is going to transform ourselves and what is the room Phison is going to grow. This is from some marketing data, talking about client controller, the term of controller is $1.4 billion plus because the control recently, price is shopping a lot. And the module itself, retail and PC banging is $20 billion. Phison, we made a study of Phison. We are getting like 24% of our controller market share here. Even though Phison’s going to 50%, if we kept 50%, it's only like $700 million. It's good, it's good, but I feel not enough to Phison since we invest the heavy R&D activity. And we're getting the $2 billion revenue last year, probably this year, I think we're still able to -- we are declining, but we try to reduce the gap. But in the SSD module, Phison is only 3% of total cap. So imaging one thing. Controller investment is getting so expensive, and I don't believe any single controller house are capable to invest any new controllers with 7-nanometers if they are not able to get their return from the business. But Phison, we keep investment. Again, Phison, we are not going to sell controllers to the market except NAND company and Kinston, our partner and Hongxing in China, our partner. So we only picked up a few controller customers. The rest is going to modules. So we Phison still have a big room to growth in the module business. And I think it's a simple -- if this market is a less controller house means less module house. The reason is the fresh module need controllers. So Phison, with our strongly investment and our transformation, I believe we have a much better bright future in the storage business. So this is a client SSD, the temp. Second, we'd like to share you automotive. Actual controller in automotive is small, mainly the EMC and the module itself is around 3 million plus. Phison we are good in controllers. Even though I get everything 100% but still like $200 million. But in the module wise, we are less than 1%. And the good thing is, in these 2 years, Phison, again, we work with many ecosystem partners in the automotive and EV. We already gained 15 plus of the top 20 carmeters, mainly steel controllers, but some we already start to move to the module business. So we believe we gained a few percent in this Module 10, then I think we are good to grow ourselves. And mobile storage. Controller wise, because controller many is in-house controller, we are good in the controller EMC UFS. But in the modules, we are only 2%. And since the mobile China mall makers, they are going to use fresh and even discrete to reduce the cost. High-end steel MCP by middle low end is a descript. So start from the August when market turn to strong, we are gaining a lot of order from the China mobile makers. So if we're able to grow a few more percent in this business, we are good in our future revenue and our gross profit. Gaming storage. Phison we are the key player here. But even though a key player, Phison only 37% of the total temp. So we are expanding our gaming business. We have a controller. Of course, we are trying to get a more module business to grow our revenue. This is just to show you the room Phison able to grow in the coming futures. Enterprise, taking enterprise is $25 billion. I don't know how much controlled revenue because not any single controller house can survive in the enterprise, to heavy investment, but the return is almost nothing. Phison, again, in the revenue we are totally nothing in this industry. So since already price going up, start to shortage, our design win activity start to gain the order. Recently, we get order from one of data center suppliers, Some more order is around $4 million. We believe when everything turns to normal, Phison can gain some percentage in this enterprise helped to grow our revenue and again our gross profit. Industrial storage. We also don't know how big the controllers, but Phison -- actually, we are doing quite good in this portion, but just only like 15%, of $2 billion. So we believe with our full range of control development, full range of system module development, we believe we can get more shares in the industrial storage. Space storage. Again, this market is very, very small, so don't expect that much, but this is to show Phison's technology muscle to the market. Since we work with one of our partner in the U.S., we launched our SSD. In these 10 months, we gained many, many orders from the North America, from the Japan and from Europe. Of course, the order is small, but at least to show Phison we are capable to move our products to the space. And we also gained a few projects. It's a core development with the space industrial company. Revenue is a small, but this to show our leading technology to the market and our innovation. So above is just show you in the storage, if Phison, we are still stuck in the controller business, you can imagine our revenue, the sealing is very, very low. But Phison again, we transform ourselves. We learn from the phone company, Chip company, selling the modules is able to increase ASP and gross profit. So we are going to spend more in that portion. But the foundation is Phison needs to keep with strong IC design team, firmware team, system integration team, service team, with the vision team, and we are doing that quite good. And one good news to share with you in this world other than the NAND company, only Phison built the world-class redivision. Only Phison capable to do redivision based on the NAND module system. So because of this heavy investment, we attract a lot of EV gaming, phone, PC industrial space companies come to us to work design from chips to the modules. So we believe with this design, with this capability, we are able to grow ourselves. This open our storage. Then back to our Phison totally the new business is a signal conditioning IC market. Gen 5 the driver is a very small business -- the revenue, but we gain around 50% in Gen5. It's good, but the market size, to be honest, is quite small. I don't expect this can be no growth that much. But this is a significance to the PC computing industry. We are moving our driver into the EV industry. So this will help Phison gain a more storage business in the EV retirement. The good news is, Phison we made so many design in activity in this server industrial. And our new -- the mass production IC is coming back by mid of December. We already has the enhanced order from the Tier 1 server company. We are happy to see that. I believe [indiscernible] coming back, we are going to ramp up our [indiscernible] production. In this case, next year second half, we have a high expectation in our retirement business. So we believe we are based in Taiwan. Taiwan is mainly the server companies here. We have a language advantage. We have geographic advantage. We have a relationship. So I think we are capable to gain more share in the Gen 5 retimers. And the retimers right now, we almost get -- we just get the certificate from the PCI SIT [indiscernible]. And almost we engage every server, [indiscernible] in Taiwan and U.S. We also have a heavily design services. We mentioned there the new announced performance GM4SSD is going through Phison imaging plus service activity. We are now doing the design activity to the EV company and also the phone company. We are working very close to the phone makers on the U.S. UFS 4.0 activity. We have a 2-phone company working with us together. Last conclusion is Phison, we need to build ourselves as a value-added module, but with strong foundation IC design activity. This is our foundation with a system module design, differentiation evaluate. At the end, we are cut everything to ASP and gross profit as a total amount. We need to grow Phison business. The last is, again, we keep increasing our on headcount. We keep increasing our activity in the new chamber. This year, unfortunately, the one experience is 21% of our total even is heavy, but we have to be super since CQ1/CQ2 and last CQ4, but I think we overcome the most difficult period. And I believe bright future to Phison is almost coming. This is today's presentation, and we got a few questions to answer you then we can go to the chatroom and live questions.

[Operator Instructions]. Khein, a couple of questions for me first. You mentioned a few months ago, maybe too much inventory now not enough inventory. But if I look at your balance sheet that you put up there, it looks like the second and third quarter, the inventory was about pretty similar, about $20 billion.

Okay. Let me clarify, okay. By July, we have inventory. But those inventory, we don't have customers place order. Inventory, most likely big portion already allocated to customers. I give you example, by September, the order we collect from customers, we consume like 2.5 exabyte, our NAND flash. January this year was a 1 exabyte, but October 4 exabyte. So it's a big growth. Inventory sitting in our warehouse is going to ship. So [indiscernible].

And we saw in October, a good spike in both DRAM and NAND prices for contract. When did you start to see an increase in your ASPs?

Actually, ASP start increasing by mid-September. We get the new order, of course, we bargain, we arguing with the customers after mid of August. Of course, customer compromise something, we compromise something. So new orders start to ship by mid-September. So ASP increased after mid-September. Of course, October, the order we get price increase again. So I believe MOM, keep increasing.

And if you look at your monthly sales, you had a nice increase in August to even a better increase in September and still good number for October. Do you think it's real demand picking up? Or do you think it's customers who are seeing prices start to go up just essentially one product? Or do you think end demand is actually taking up.

I answer you by 2 portions. First of all, customer side, they keep very low the inventory. Actually, since the end of one system company, retail company, they keep very low inventory because sometimes they’re saying that they are not going to cut, they can ship anything. So the customer thinking tomorrow cheaper what they are today. But after mid-August, they keep low inventory, Samsung in inventory, so they start to review their inventory. And August is coming for Christmas season, so they prepare something for Christmas. We believe 70% is a rare demand, 30% is refill their inventory. Second, Phison, I agree to you, September versus August, the venue pumps up -- I don't know, 20-plus percent. October, almost nothing. The reason is we have to manage our inventory level. Actually, we can easy to clean up inventory by one night. Any day in October but we’re to manage ourselves. If we sell too fast, we are going to suffer our next cost up from our suppliers. So I believe the demand 70% is real. But of course, economic scale is not the good. Some is going to the inventory. I believe we're still able to buy back some good inventory in December or January or February, I believe, I believe. So after that, when everything back to the phone spring model, the Nobel spring model, then the market we're going to undersupply.

And once again, you have a gross profit margin over 30%. Obviously, it's been very good, especially given the difficult year. I know things are turning up, but certainly, third quarter was still pretty difficult. So how sustainable is that very high margin that you've been putting out?

If talking about this, I need to back to the 7 years ago. Phison was 2016, 80% in the consumer. But now you can see we have a consumer like 25%. I want to share with you why July we suffer high inventory, September we suffer low inventory, why? Because Phison making so many design in, design win into what, into the gaming, into the industry. And you don’t know how many industrial systems in the market and embedded ODM, mainly smartphone and the PC. So we made a lot of design wins, but in the first half this year, demand was weak. When the market turns to good, every order coming back with the pricing increase, then we are able to keep our good gross margin rate. The answer is, Phison diversified our business to every NAND storage, we cover full range on NAND storage. So even one product line we suffer low margin, but overall, we can keep the good average. This is our value.

And next year, it seems like we're setting up for a very strong 2024. Where do you see the biggest growth if you break down your products by end market or [indiscernible]? Where do you see the biggest growth next year and maybe top 2 growth areas?

Capital growth.

Whether it's a client SSD or [indiscernible].

Embedded model for client OEM and also for smartphone. Embedded because this is mainly is a margin business. Gaming, we gain many gaming projects. So gaming will be bit close. But on the other hand, controller potentially will be a big growth because this first half, NAND company plays a less controller. But next year, we already get the focus. Next year controller they have the other big growth. And of course, the retimer, we are starting from nothing. Retimer will be.

Maybe next question on the time since you mentioned that one last year. So as a newcomer to essentially a pretty small dollar component in the server, getting designed in is one thing, but how comfortable do you that you actually get pretty good sales and actually real shipments there?

First of all, server data center, we need to get the trust from the market. We went to every server company, no matter data center enterprise ODM OEM. They ask a question, “Who use your enterprise SSD?” And the answer is no. Of course, we have a Seagate as a partner. But Seagate get many, many design wins, unfortunately, in the first half, no matter what price we call, Samsung always 10% lower. Always spoke, we are not able to get any room. But then with Samsung, they are saying the car production, they are raising the price, giving some room, Phison able to ship our products to the market to prove we are good few more months, everything doing well. We got the trust, then we believe next year, [indiscernible] will be tight. We have our LTA, we have allocations. We believe some server company will come to us to ask for supply. So this is our good start. So thanks to the market. Without the market, we are not able to get any business. But also, we invested 4 years in this GM4-GM5 systems, everything doing good, customer happy. Price now we're doing good, then I believe by 2025, 24’ will be the starting year. We start to get business to get the Trust 25’, we are able to see the potential significant growth in revenue in enterprise business.

And on client SSD, when you talk about that market, is your market in the notebook market, the desktop market or both? And what's the main difference between those 2 markets?

Phison, everything starts from consumer markets, from retail. But after we work with AMD on the Jan 2019, we gained some trust from a CPU company and a PC company. So we gradually do gain some client SSD margins from the PC OEM, Tier 3, Tier 2. Now we cover Tier 1. For Tier 1, we're starting at last year, still where there’s more portion, but we believe Phison can be one of -- of course, we cannot be strong, the bigger player, we are not the NAND manufacturer. But if we are able to service every tier customers, 2% to 4%, the share, then we can get a lot of growth from a PC OEM. Of course, the retail is still our main business.

Let me read some questions from the question box here. Do you think memory makers are able to reduce production further in fourth quarter '23 and first half '24 or they have already reduced the production aggressively so there may be less room for them to lower their production rate further along with the recovery of memory price.

Again, I'm not a memory company. I cannot represent them to answer, but I believe the price today already bounced paid from the lowest point over 70%. Even 70%, they are still losing money. So the leader of a NAND company already announced that we're going to raise more by CQI. I don't know if either one already start to reset their capacity. I don't know, but I believe some already start to vision. That is why I'm saying PC1 potentially may be a little bit slow because the system companies start to adjust inventory and production may be increased a little bit. But anyway, we think this is healthy. The price won't go down anymore because the NAND makers they are not willing to lose money. But back to the next year, I don't know, economy scale and many noise there, but based on NAND industry, we can see the big growth is still improving. The PC go to [indiscernible], automotive go to the 256, 512 gigabyte. So we believe that demand in big growth is increasing. So I don't worry much next year. Maybe CQ1 is one of potential slow season, but after that, everything will be good.

In terms of the controller competition landscape, it seems like Silicon Motion is still facing ASP and GPM pressure. Are you seeing any impact from them if they still take aggressive pricing strategy to complete the inventory? Also some Chinese IC companies also introduced more controller products in the market. How do we compete with them? Fleet China also wants to increase its controller localization rate?

Talking about the NAND controller were separate to the 2 markets. Module company, mainly in China. NAND company, PC company, whoever is other than China. We have no question in China. They have a few controller design house already mature in SATA, mature in the [indiscernible]. And I look to the accusation that the gross margin is a low 10%, low 10. The SATA controller today when the pandemic was close to $3 now, I heard in China $0.70. I told to my myself, hey, forget we are not going to sell anything in China. But the good thing is a Phison never sell controller to China company, except [indiscernible]. Our control only to the name company and [ Kinsa ]. So we are not in China controllers, but when -- if you're in China, you have to face the China low 10% gross margin. So I think the controller ASP in China now is really agri. Next year will be more wide. If NAND is undersupply, NAND is a shortage, who is going to buy more controllers. So I doubt controller next year can be good in China. And if controller is not good in China, then the gross margin may turn to minus inventory. Then talk to about the -- other than China, the main company, okay? Yes, -- my competitors signed more recently, I don't know why, I think I can know why other fail in their M&A. They need revenue, they are be aggressive in the controller coating to China. But to China, I think they cannot compete with China company, either or they tell loss. Then they go to NAND company where they're aggressive. And I'm really clear my policy, okay? We are good in today's our gross margin. Since our modules in the gaming, industrial and embedded everything, we enjoy the good profit. And by the way, we're still sitting with a good inventory. So if competitors willing to lower the price, I'm happy to follow because controller is only like 29%, but can be the 90%. So I don't worry about that. In China, we don't touch controller, -- but other than China, we are going to aggressive its competitors going to aggressive. The reason is we are enjoying modules group profits return.

I think we have a question on the line from Jason from CLSA.

Congratulations for the results. So my first question is in terms of your gross momentum in the first half next year. We see demand is improving and also we face pricing tailwinds. So which is the major catalyst or drivers, I mean, for first half next year, I mean, for shipments or for ASP growth?

I take this just to give you the explanation, okay, just a moment. Again, first, controller, we believe next year will be too strong because we made so many design wins in the NAND company, okay? So controller will be good. Consumer, I think CT4, Christmas is not okay, [indiscernible] slow again. So consumer just so solar okay. And we are not much focused to the consumer anymore because next year, NAND is shortage. We are going to move them more NAND into the industrial, gaming and embedded and enterprise. So next year, we believe the other big growth were coming from, again, from embedded smartphone and PC OEM. Gaming, yes, we get many, many design wins. So I think this is what we are expecting. We design win not only in the console, but also in the PC base and also the retail upgrade. Industrial module, I think it's way slow. We will read the industrial PC makers, right? Were slow in industrial don't expect first half. First half, I think industrial steel may suffer, but were okay, crazy shortage. We move fresh from our industrial consumer, mainly to the embedded and gaming and enterprise.

So my second question is in terms of your gross margin. So we do see we potentially have a reversal from our inventory write-off. And we also see the pricing tailwind. So can you give more colors on maybe gross margins outlook for the following quarters?

First of all, ASP increasing from our inventory, so helping our gross margin. The inventory write-down reversal. I think in CQ4, CQ1, CQ2 next year, we're gradually back to our profit, helping our gross margin. Retirement start to ramp up by April next year, after wafer-out so helping our gross margin. Controller increasing, also helping our gross margin. But in module wise, I think now inventory is doing good, but our coming waiver the price is increasing, okay? The portion may impact some of our gross margin but the dollar amount of gross profit increasing. So overall, by average, I think we're still able to keep our current average gross margin, okay?

So my last question is in terms of your NRE or design service business. Can you give us more color on it? How do you -- maybe to get the design win or who is application or end market is?

Actually, every quarter, Phison recognized our no income from most men company and a system company, okay? So we are going to expenditure -- the design service will be Phison our main business in the coming future, okay? And next year, since AI is in popular, many system companies come to us to ask, “Hey, how about putting AI engine something customization into the SSD controller? How about putting something into the enterprise controller? And so how put our engine into the mobile controllers whatever.” And also in the EV. EV is the most heavy requirement in the specification. So overall, I think I'm not able to answer you which Poland we are able to enjoy but any system in EV, in mobile, in computing, in IoT, in server, they're asking some customization. So this is helping us to gain a more and only through their business service. We need this in imaging class. So this portion is going to improve in the coming 8 quarters.

So I have one question. I'll read from the Q&A box, and then we'll go to Sebastian next. So the question online is the NAND companies are developing own controllers by own internal teams and by paying design houses, how much of man-controller marks for SSP will be in-house in NAND companies versus merchant controllers?

Again, I cannot answer in this question that much. But some companies still insist using now, but some companies start to outsource okay. We have one Korean company that have been outsourced to one of a controller company, but the theaters policy side back to in-house forever. So I don't know how to comment. But I still believe the man company will more focus their control them to the high pickup pipe module spaces such as enterprise SSD gradually. Gradually, they are outsourcing, they are mobile controllers. They're outsourcing their Gen 4 controller Gen5 controllers. And some land company, if they are weak in finance, they also consider, hey, are they going to tape out their own 7-nanometer fumed controller, -- this costs too much. So they may talk to the outside party, how about sharing the [indiscernible]. So the answer is, we believe outsourced will increase other than Korean company, they suffer from third party, so policy in-house, no more outsourced on in-house. I think this is at [indiscernible].

Sebastian, do you have a question?

Yes. I have two questions. The first one is I can't -- maybe my memory is bad, but you can remind us that where is the last time you do this prepayment? I don't remember you did that before. Did you?

We did. We did 2,000 beds to Korean [indiscernible] to Japan partner.

Maybe I try questions. You didn't do anything like this in the past decade. So it sounds like that was the time that before the -- more than a decade ago when the last time you did it, right?

Yes.

So we have been through a couple up and down cycle in the past decade. But even in some of the very bullish cycle in the past decade and you didn't do that. And why this time? What are you seeing that you're going to do this, you must see something that could do this prepared because it definitely will put a lot of burden on your working capital. And we understand that your business model and a lot of that depends on the working capital as a module company. So you're betting on this -- we definitely see something coming that you're willing to take the risk of doing that. Can you share with us what are you seeing?

Again, I keep emphasizing in July, we still suffer in our high inventories. To be honest, the demand is very slow, inventory is so high, supplier keep asking rate to buy, we keep buying it. But by end of September, most inventory allocated to our customers. So I just realized the good and betting is we made 2 many design activity. We don't know where the demand is coming. But when market tend to good, everybody come back to us supply -- and unfortunately, inventory one can be gone in one night. But of course, we manage well. Next year, when you go to this chart, okay, this chart, I think no doubt, second half next year, the market will be turned to very, very tight. Actually, no need to wait until next second half now today. We call to U.S. suppliers, Korea, Japan, even China supplier. You asked them item 2 to 3, okay? So if next year, definitely I want to grow my business because I think we have a much better position, okay? We got so many system design wins. We have every control design win. But if we're not enough flat supply, how can I turn to my design to my income. So I need money, okay? And next year, no matter what… yes. And second is we gained so many OEM Tier 1 design win, which they're asking for certain petabyte, we may not be able to fulfill if we are not secure our supply next.

So if I interpret what you're sharing correctly, the 2 elements, one is that what the company and you're doing for this prepayment is actually signaling 2 things combined. First is that Phison is actually calling for probably one of the biggest up cycle in their flash industry in the past decade. Second is that piss entering into a new phase where you have much more design win that you didn't have before that you need a lot more density of the NAND flash. Am I right?

Yes.

Second question is on your gross margin. I think the -- in the past 2 quarters, particularly in the recent third quarter, you end up with a margin of 32.5%. If I look at your history, this is probably one of the highest margin in the past decade. You did a 33% something about 2013, something like that. But even in one of -- in some of those up cycles, where you probably have a good margin, but you didn't -- but still lower than the third quarter. But third quarter, you -- previously, you mentioned that pricing only started to move in mid-September. So which means [Technical Difficulty]. I'm just curious about it because the third quarter is only the second half of September, you started to enjoy some of the pricing benefit, but you actually have probably one of the highest margin admitted down cycle, still in a down cycle for most of the part of the 3Q. And now if we are going into the up cycle where pricing become more favorable, where you actually have some local inventory plus some of the margin tailwind you just mentioned about the controllers or re-timer -- are we going to see Phison moving up to like mid-30 or high 30 or even higher gross margin profile as a new level going forward?

Of course, this is my goal. This is my goal. But before we achieve this growth as a public company, I am not able to comment that much. But what I'd like to say is we just show you, right, Phison is a business model, again, we transform ourselves and to so many -- you can define a niche application, which we be able to enjoy in the much higher gross margin. But that to Phison, I prefer to improve my top line first, then second will be bottom line. The reason is we need to grow our term. Once we grow up term, then we can enjoy our gross profit and our gross margin, okay? And again, Phison, we already have a 3,000-plus engineer, maybe the end of next year, we still need to hire the other some portion. Why? In these 6 months, we got so many name controller came to us to ask Phison design something for them. We've got so many season companies come to us. System company come to Phison to pay [indiscernible] design something for them. So we smell that the market is changing in the past, name companies, Samsung gives a standard module. You just use it. Don't complaint on us, okay? But today, every system company, they're asking differentiation. And the good thing is net company may not take it. I give you one example, okay? It's an EV company. EV, if you're selling 1 million EV is a big amount, right? But 1 minute EV using 1 million SSD. 1 minute SSD they say to [indiscernible]. You are only using the 8 million piece of time. In order to get this MMP business, you need to hire a lot of engineers, too expensive. But Phison, our niche is we have everything. We met a new record. We designed one thing for EV, we think 7 months from nothing to [indiscernible]. Why? Because we have a big foundation, good skill engineers, good trend engineers, they keep sustainability in Phison -- the reason is we may have investments. So yes, the answer you, I want to grow my gross margin in red, but instead of that, I prefer to gain the revenue and share first to improve our influence power. I believe again oaths is my work, okay. Controller House will be getting much difficult if moving to the 7-nanometer [ fanometer ]. Even Phison, we made a good gross profit as you heavily suffer very painful to invest to the advanced technology. So other than controller house, I think would be really difficult. So if no controlled house means no merger house. Phison is going to take the merger business. Like today, I show you in this picture, we cannot name that much. But this is the model that Phison is going to lead. Okay.

Yes. I think the next question is coming from Simon.

Very quickly, sir, any color volume increase versus price change for Q3 results?

Sorry, so can you repeat, sorry?

Q3 revenue of right, but any rough inert portion based on the -- your sales volume increase versus the price hike?

Actually, I'm controlling my volume shipments, because price will be higher, right? So many is coming from some ASP improvement.

Yes. So what's the offer idea that quarter-on-quarter, your product price changed low single digit, mid-single digit, industry data showing very minimal price increase for September quarter, but presumably, your Q3 revenue seems to be based on the price hike, right?

No, let me correct some portion price high, but some portion is a new product such as enterprise SSD is a new product.

Mix improvement?

Yes.

But the December quarter, should we expect your -- the competitors try to increase for your module product mostly?

So answer you this way, okay. We follow Samsung. I can answer Samsung is the industry, right? So we follow Samsung. If Samsung is going to raise a price, we just follow.

And then your inventory level remained very high, TWD 21 billion. It's good when the industry shows over the trend. But you are issuing the new convertible bond, the TWD 6 billion, this bond to be used to purchase more of the landed chips, right, with the long-term agreement.

No, let me answer this, okay. We keep TWD 21 billion inventory, but among this TWD 21 billion big portion reallocated to customers. So we know how this is -- of course, on the book inventory after 2 months gone, most portion gone, okay? We need to turn to our revenue and our income. And talking about the convertible bond, we have 2 main reason. One, we believe next year will be a big undersupplied. We need to call it whatever supplier willing to ship to us. The reason is we get so many design win next year with our commitment to where to ship it. Second, next year, I think we are still going to improve our R&D activity, mainly in the 7- and 6-nanometer controller tape-out. So I think at this moment, if we improve our cash flow will help us to grow our revenue and gross profit by 2024 and '25.

And very quickly, 7-Nano note, would you share some details on when you will be commercialized meaningfully? And then which application can be used.

It's mainly for PC computing. The IC will be ready by end of this year. so commercialized by second half -- end of first half next year.

In the end of first half next year.

Probably May to June time frame.

It sounds like what client SSD, PC EGV? And then UMC is the supplier, right?

Does UMC have a 7-nanometer?

No. TSMC, sir. Yes, but you are using the legacy order from UMC and more events to node from TSMC, right? very balanced and still partially SMIC, right? SMIC.

No, never. Phison never SMIC. It's only in UMC and TSMC. Now we increased the global property for our re-driver.

And then on just the last one is, sir, for the long term, how about if you sell your controller to the team makers more actively? What's the color status, your controller service, 100% internal use or 30% for external...

We posted, of course, SNUS, we made many design wins from a nan company.

So what's the rough, let's say, you manufacture then to control the 100 units for this quarter and the percentage...

More than 2/3 will sell controllers. More than 2/3.

[indiscernible] customers.

Yes.

I see the customers which use your own grade controller, the majority should be 2 makers, right? And makers or…

Phison controller only sell to name company plus Kinston plus [indiscernible] global.

Hong means YMTC?

No Phison subsidiary in China.

Sorry, you are saying the 2/3 of net controller for the NAND makers, right? 2/3.

Yes.

And then the -- but you are not active almost very minimal for the OEMs.

Okay, define what's OEM.

OEM means another maybe model makers.

No, no. Please clarify. OEM is OEM. OEM is not module house. Module house, most likely it's a trading company, okay. We have no selling control in company. So I said we are not selling any controller to module house other than Kinston and Housing Global.

Yes, sounds good. So overall, this really means on the opportunities for more customer diversification with the NAND makers.

So Simon, let me ask your question. Is that this GPU company selling GPU or the selling models?

In media for example as GPU and also these days turbos well.

So with portion bigger single GPU bigger or the modules bigger in portion wise.

I mean the NVIDIA GPUs.

No, there are many selling modules. They are not willing to sell more single GPU than selling modules.

So in that case, they need your controller.

We are not selling controllers. We are not telling a module house. We build module. We don't need mutual house. Any controller rely on module house at the end is die, die.

And then another maybe some questions from the investors. I know the $60 billion convertible bond is important for your future growth. But your balance it shows $15 billion cash. And also you can bottle of course, the interest rate very high these days. But you don't worry about low to mid-single-digit potential dilution from the CB issuance or not a big deal, you think.

First of all, personally, personally, I believe NAND market will turn to good. Phison potential is going to enjoy. And we are able to turn to the good revenue, maybe good profit. So if investors support the dilution will be very small.

We have a couple of questions from online here. The first one is, today, NAND supply tight, but NAND companies have mothballed capacity as prices are too low with negative gross profit margin. If price increases can they quickly turn capacity back on?

Now the whether in and out at least 3 months, at least 3 months. So I don't know how quick.

And the other question is, how is your outlook on OpEx in fourth quarter or in 2024? OpEx Q4.

I think we are going to increase the reason is when we turn to group profit, we need to run off a certain percentage to the employee bonus, okay? So we still need to take care of our employees. So OpEx definitely Q4 increased because I believe gross profit increased. Next year, again, I believe our -- we got in the revenue because it's a many macro business re-timer and the control business. And the R&D activity next year, we still have several a nanometer, 6-centimeter is planning to tap out, mainly is a design service from some name company and system company. Of course, they are going to pay some annoy. They pay out any as a gross margin, 100%. But on the other hand, OpEx is going to increase. but the voting is OpEx increased but turned to our ASP and gross profit were going to increase. This is my key point.

I don't see any more questions online. Any last questions? Anyone please put in the -- raise your hand or put it into the Q&A box. I think that does it for the questions. I'll turn it back over to you for closing statements.

So thanks for your time. I think we over like 10 minutes, but again, we are working super, super hard try to gain more market share next year. And also, again, I still believe doing controller is so difficult, big investment. So we transform ourselves mainly to our potential value at the module. We also show you in Phison’s share in this time is still nothing. So coming years, we try to improve our R&D activity. We try to improve our capital finance cash flow, then we are going to try to capture more business opportunity.

Okay. Great. Thank you, Khein. Thank you, everyone, for joining the call. Everyone, have a great night.