Good afternoon and good morning, everyone. Thanks for everyone to attend the meeting. I'm Ray for Morgan Stanley. Today we are pleased to have Phison CEO K.S. with us. He will talk about the financial results and also the company outlook and go over and we'll have some Q&As. Yeah. Without further ado, let me pass it to Phison to K.S. Thank you.
Good morning. Good afternoon. Good evening, ladies and gentlemen. Thank you for joining Phison's our 2023 Q4 earnings call. So, let's restart the call, okay? Okay. We made a summary regarding the memory company, the Q4 2023's OP gain or loss. We read that SK Hynix is going to deliver profit just because of HBM memory due to the AI's crazy demand. The loss is narrowing around -TWD 3 billion. We believe by Q1 most likely will turn to positive. So this shows the NAND market is overcome the most difficult period. This is Phison our Q4 the distribution. You can read the consumer portion increase quite a big number just because the Q4 is original retail season. And also the NAND price turned to the price increasing helping us to gain the revenue and some profit.
Controller-wise, from this table looks like declining but actually not declining. This is just a ratio. Controller itself is increased 12%. Our NAND company controller demand slowed down since the second half 2022 due to the COVID issue. 2023, they are digesting the inventory. Second half of 2023, we can see the demand sliding back and the controller turned to strong, from Q4 and carried to Q1, Q2. So the controller business, we made a lot of design win into the NAND company. We can see the big growth in 2024. The rest of gaming module, embedded module, industrial module, by dollar amount actually also increased, okay? So the Q4 financial result, revenue, you already know from our announcement around close to TWD 16 billion. The gross profit, we hit the TWD 5.6 billion. Actually, the YoY is 64% QoQ 42%.
Q4 2023, the gross margin itself, we hit the all-time high. So it's a little bit surprising but I think this is just to deflate. Phison, of course, NAND price bounced back with the inventory help. Of course, we gained a little bit gross margin. But when you compare to the Q3, Q2 last year, we still keep kind of a mid 30%, actually low 30% of the gross margin. Just, deflated Phison. We've made a lot of design win into the system, which we're able to gain the better gross margin. Even though during the first half of 2023, the market was still bad, we're still able to keep low 30% of gross margin. So we believe the Q4 last year is the all-time high in the gross profit.
This is going to carry to Q1 and Q2 because the design win activity just turned to the revenue. So we are proud and happy that we are able to hit the all-time high at last year's Q4 even though the market just started to bounce back. So the full year the full year revenue is TWD 48 billion, declining YoY 20%. But the gross profit declining 7%. On the other hand, what we can say is we gained the gross margin rate in our business. So this is our revenue trend. You can see this can be a V-shape, okay? We hit the all-time high in the Q1 2022, then slid down, touched the bottom, last year's Q1. And after that, the revenue starting sliding back but big growth in the Q4 last year.
So this is a V-shape and we believe this growth will still carry to CQ1, CQ2 this year and hopefully will be CQ3. Okay. This is a gross margin rate, 36%, which is also outperformed than our internal expectations. So this 36%, there's a 1.4% is a reversal from our inventory write-off, okay? So we take out this 1.4%, the real gross margin, 34.6% still in the all-time high. Again, this is just deflate. Of course, inventory price up. But as I said, even though in the first half of 2024, the market's still tough, the NAND's still ugly, blurry, terrible low, we keep better gross margin just because we win a lot of design-in activities. This is also to telling you, Phison, we transform ourselves. I'm actually talking for many years.
We transform from retail to embedded to design-in to design service and to a lot of full turnkey. So this to show that Phison is totally going to the high-value business model. So with 36%, we are good and we have confidence in the CQ1 this year. We're still able to keep high 30% of our gross profit. So earning per share in this CQ4 we are 10.3, close to 10.4 earning per share. And the Hosin, our China investment company, basically first three quarters, they keep big minus in their business. But the whole year, they turned to profitable. Means CQ4, they have a big earning from the business. So in the last four quarters since CQ4 2022, this four quarters, Phison suffered quite big amount of the investment loss due to the Hosin loss in the China business.
But we overcome and we also believe by 2024, on the whole the year will keep quite reasonable or quite attractive profit. Phison able to recognize. So means we don't have any burden on our China investment. So with this, TWD 10.4 EPS, the whole years of our Phison earning per share is 18.5 something. Also over our expectations. So again, we are proud. Phison people, we work very hard. We won many activities. This is to show to our shareholders we hit our target. We overhit our target. Okay. This is some highlights in our the P&L, our income statement. The margin itself is 36%, which is a record high. The OpEx also higher because we have our employee compensation. And also we have to keep investing to our technology.
Okay. Just a moment. Can we mute?
Okay. Sorry. The non-OP gain, actually we have a quite big of a non-OP gain due to just SK Hynix's turn to the profit in the quarters, okay? And the whole year's income statement, 2023, the full year's gross margin 33.4% is, all-time high. And the expense also all-time high due to the we have a 7 nm tape-out. We have, also preparing employee compensation. We also have a restricted, share, recognized, exercised to employee at the last year's Q3. So and the full year's, net operating gain and loss, actually returned to the, gain versus, 2022, again due to the SK Hynix turn to profitable.
So overall, 2023, even though revenue, dropping 20%, but we hit the gross margin rate 33.4% all-time high and we're still able to achieve, two digits of earning close to 20, I think we are happy to tell to shareholders we really work very hard.
We are very optimistic about 2024. About the balance sheet, okay? Inventory-wise, we keep like TWD 24 billion something. In cash flows in the end Q4, we're still in the cash rich in the operations. The overall ROE, ROA, we are improved a lot versus 2022. So we are keep building our inventory because we foresee the supply is getting tight in the Q2 and Q3. Okay. This is regarding T-IFRS and non-T-IFRS differentiation, okay? So this is about our Q4, okay? To be honest, the gap here is quite small. This is just because of we have a stock option write-off every month. This is regular. So this amount is a kind of a small. So the impact here is small, only like 0.something%.
But the whole year, the whole year actually is around non-T-IFRS EPS increased TWD 2.2 due to the treasury stock exercised in the Q3. And this impact around TWD 1.1 happened in the Q3, okay? So this, as we explained to shareholders, we are like to follow what the U.S. game rule. We have a T-IFRS and non-T-IFRS, this crucial that you can make your own decision and your own study. Okay. Consolidated revenue of February is TWD 4.6 billion. Revenue YoY 42%. And the consolidated two months is close to TWD 10 billion, YoY 58%. So February we have around one week, more than one week's Chinese New Year holiday. So QOQ is declining a little bit. But overall, I think we are happy to see the business with this revenue. Okay.
Again, since many years, the investor, the analyst, the fund manager keeps saying that systems, company system company value is low. PE should be low. But again, I need to define what is a system company. There are two kind of system company. One, they buy component, assembly as a system, or they design their own IC but revenue coming from the systems. So these two companies are ready to show you how they create the value. They own the in-house IP, IC technology but they are not gaining revenue from IC. They're gaining the revenue from the modules, which they're able to to increase huge amount in their revenue. On the other hand, they also gain big amount dollars as a gross profit, which Phison business model, we start this model since day one when the company start.
So Phison, I keep telling to analysts, when you're talking about the value of the company, you need to study. If they own the foundation as a chip design, IC design, the IPs, the value should be much different than the traditional system house, okay? So Phison, we are in this model. Okay. What I'm saying that we are the activity is a chip design but revenue coming from the modules. So better ASP and a much better gross profit. And we show you even though Q4, market just a little bit bounced back but we hit all-time high in the gross profit because we are exercised with the gross margin with the modules. And this is what according to the TSMC chairman, Mark Liu, they're saying that in the coming years, system company is going to grow, okay? But the design house is growth a little bit.
Fabless is 10%. The IDM is 4% but system company is 17%. And MediaTek also saying that MediaTek already transformed to the system company. So Phison, we don't mind to keep saying we are system company but we need the analysts to know what is the value of my definition in the system company. Okay. So this is what we are doing. Foundation with the IC design. Then revenue coming from the modules. This improves our value in the gross profit. Okay. We are proud to get the award of our very close partner, Micron. This award happened by last year November. We have the outstanding supplier in the category of semiconductors. So they have only one, winner in every category. And in the candidate, they have a still much bigger size than Phison but we won that. Why we make it?
Because any request from Micron in the chips design, system design, Phison fully deliver on time. So they're happy about Phison. Due to Phison, we keep investing our strong enough resources, human resources, equipment resources in Phison. We are able to react the request from Micron and to make their products to the market on time. Their products also won a lot of, top-ranking award from other, you know, reviewers just because Phison, we contribute our IPs, our know-how, and our design to them. We also won the KIOXIA award last year December due to they're happy to see Phison helping them to pull in their new generation of a flash memory, which is a big step. We are putting that into their coming, high-performance, SSD. So this is just to show you Phison's, our R&D capability already recognized by the NAND company.
Through the NAND company, they also bring us to the tier one system company. So we believe by 2024, Phison, we are able to get engaged with a more bigger system company, which including their server, data center, AI, automotive, aerospace. So we are happy to tell the shareholders we work very hard. We invest that much but we have a good achievement. Okay. This to show you, we try to split the segment in the revenue. In the client SSD controller, this is amount $10 from the Semico Research Center. Phison, we are like 20% of the share. I think we're good because Samsung still in-house, Hynix still in-house. But we hit our good market share in the controller segment but you go to the module, Phison's only like 3% of total $10.
Means Phison still have a big room to grow our revenue and our share in the coming years. With Phison's in-house, IPs and our customization service, I think we are able to work with the more tier one PCs, servers, and such, application to help them make the difference of the products. Then we can gain the revenue from 3% to some certain double digit. So we are potential to growth. In automotive, we are already doing hard in the controller side. You can see, in eMMC, in the auto, we are 40% but the TAM is only like TWD 200 million. In the full turnkey, they have a TWD 3+ billion of, TAM but Phison is less than 1%. So from 2023, Phison start to work very closely with a lot of, the tier one, the brand company.
We are here and we're happy also we have a very strong partnership with those brands and we are start to gain the business from them. We believe by 2025, Phison revenue in their auto modules will go to the big growth. I believe we like to go to the double digit in the total 10. In the mobile, Phison also have a, I mean, 50% in their controller share I think is quite good because Samsung, Hynix still in-house, Micron, UFS still in-house. But we gain some share from the NAND company and also we are partnered by ourselves and also in China to make the modules. But again, go to the modules, we are only like 2% of total 10. So again, we work hard. If we can gain this close to the 10%, then we can enjoy the big revenue.
This, we have to directly work with the phone company in China and in Korea. We are trying very hard on that. Talking about gaming storage, Phison didn't sell controllers but in this gaming camp, Phison like 7%. We are really doing good. I hope we can go to more than 10% in the coming years. The reason we can make it because we have many in-house and unique technology. We are able to work with the console makers, with the gaming machine makers to customize the products what they need. So this, we also potential to growth. Enterprise, which is the biggest of the consumption in NAND, Phison, unfortunately, we are less than 1%, much lower than 1%. But from last year, our products ready to the market, we believe 2024, the portion of enterprise SSD in Phison will go to double digit in Phison revenue.
So we are going to grow these products into the market with our not only tier one, we have a SI partner, we have an ODM, OEM, EMS partner. So this also can be one very potential growth in Phison in coming years. Industrial storage, I think Phison doing good, okay? We're doing good. This is very slow such as, automotive is very slow. But we keep engaging with many, many local partners to gain the business. And we also put this into the aerospace. We gain a lot of aerospace design win and a lot of new applications. We are helping them to make the customization. So this is still growing but a little bit slow but we are happy about that. Talking about the High-Speed interface, the Retimer, the Retimer is a very small market. Phison gain over 50%, okay?
We are newcomer but we gain. But unfortunately, it's a small business. But it's okay. We show Phison's, our capability to the customers. Retimer, we are ready. We are shipping samples. We are doing the POC. We are doing AVL. We already have a two-digit number of AVL with the server company. We believe second half we can see the revenue coming from the Retimers. So we are newcomer but we can see the revenue in the first year when products ready. Okay. Yes, Phison commitment. We keep telling our shareholders we are going to keep investing into the R&D, okay? And when you look to the Phison's many achievements in the business just because we have a strong R&D fundamental people, equipment, resources, know-how, and accumulation. So 2023, due to the revenue declining but OpEx increasing. So we also hit the all-time high in the expenses, R&D expense, 22%.
Unbelievable. But this number is going to be controlled. It's going to be controlled. We are start to use Phison invented aiDAPTIV+ AI solution to design our own firmware, software, chips, documentation. So by 2025, a lot of R&D designing will use machines. The machine is developed, invented by Phison. So we are going to control our headcount. Means we are going to control our expense. Okay. So this is today's earnings call material. So we collect some Q&A. Yes, we finish Q&A when we open online. Okay. Dividend. This morning shareholders meeting, we decide the dividend formula is here, okay? Based on our finance book, EPS is 14.7% in the second half. 55% is our principal. So we are going to define the TWD 8.09 per share is our dividend of second half.
But as a commitment, last year Q3, we exercised our treasury share to employee, which the expense is around TWD 1.1. We committed this TWD 1.1 to cover back to the profit as a company profit. We put this TWD 1.1, 55%, then we got the 0.75%. And this we also pay to shareholder as our dividend. So overall, the total amount of their dividend will be the 8.84%, which is close to last year's finance book, 60%. So shareholder may ask, why not increase their dividend? I like to express that Phison, we are still growing. And I foresee, I foresee 2024 and 2025, we are going to growth much rapidly than what happened at the past. So we still need a lot of cash flow to help to growth our business. So we like to take care of our shareholders. We try to maximize the dividend.
But on the other hand, we also need the cash flow to growth our 2024, 2025. So this can be a good balance between the company and the shareholders. So TWD 8.84 dividend by last second half. Okay. You may read from the media, from report, Phison, we are launching our aiDAPTIV+ solutions. So why is that? I don't think I have a time here but I try to summarize why is that. Basically, AI is crazy, okay? AI is crazy. You can see the U.S. stock market. Those are big guys. They are doing a lot of pre-trained LLM models, okay? As users, small, medium-sized users, we need to have a fine-tuning. But fine-tuning, the equipment is so expensive. Means the users can either go to Microsoft or Google or Amazon asking for service or buy the high amount dollars of equipment.
A lot of our government, financial company, tech company they're not allowed to share the data to the cloud. They are not allowed. So but they're also not capable to buy that expensive equipment. So stuck. They have nothing to do. Phison, we launched our aiDAPTIV+, which is a low-entry ticket in dollar amount to help any individual company or single person, you can fine-tuning your data by yourself in your home, which is a low entry. Means if you go to their today's, AMD or NVIDIA model, you need a few million dollars. Using Phison invented solution, a few $10,000, you can start to have your own fine-tuning. This is like magic but it's okay. We have no time to talk about this. Next week, GTC, NVIDIA GTC, Phison partners is going to launch the server in the in the booth.
Then I think the world will know what happened on aiDAPTIV+. So this is example. We need to do some fine-tuning but with the—I mean, okay, sorry. We have. Personally, I have a nickname of aiDAPTIV+. I name it a poor man's AI. Poor man, poor man. We don't have money. So, if you have only TWD 1.5 million, you can buy only this GPU, can only fine-tune the model 13B. But 13B may not good enough. So even to go to the 70B, you need to put more money there, okay? But with Phison solution, with the extra TWD 200,000-TWD 1 million, you can expand your memory size. Then you can run much, much bigger the LLM model to have a better achievement accuracy in your inference. So this is what Phison invention.
We just able to make the fine-tuning to go to the popular in the market in second half. So who need this fine-tuning? I back to the first page. A lot of professional lawyer, doctor, engineer, the, the, the call system, the government, anyone you have a data, you need trend, then potentially you will need this, service and this equipment. So again, this is Phison's invention. Okay. So Phison, we provide this one is a we call aiDAPTIV+ edge, okay, which is a lowest entry ticket to go to the fine-tuning. We also have an aiDAPTIV+ cloud helping the rich people. When they adopt Phison aiDAPTIV+ solution, they're able to improve the performance. Based on our study and testing, maximum can go to the up to 30% in computing, okay? Means you save time, you save power, okay? So this is Phison introduced to the market.
By GTC, we are going to launch with our partner. By Computex, we are going to show everything in Taiwan. We believe when Computex happen, these things is going to be very popular. Okay. So this is going to benefit a lot of professional users, okay? Again, low entry ticket in the dollar amount, good to use, easy to use, and you no need to put your data to the cloud. All the data, you just keep in your home, okay? So this is about aiDAPTIV+. Okay. This is saying that we see the narrowing, the loss in the NAND company. The demand actually is doing good. We also keep hearing the data center start to ask big amount of SSD to expand their data center.
So the demand I think is doing good in the industrial but consumer-wise is a question. The price going up too much also impact the end user to buy the storage. And also the end user may not buy the high capacity. They buy the half size of the capacity. Question is retail is a questionable but for AI system, data center, automotive, AI PC, AI phone, IoT, blah, blah, blah, the demand in the end still very, very strong. So the question is, NAND price is going to rise? Answer is yes. We keep hearing our supplier going to raise another 25% by Q2, to raise another 30% by Q2. We're happy to see the NAND company turn to profit. But if the price keeps raising, I don't see this is a good symbol to the industry, okay?
So I got so many calls from shareholders, analysts, fund manager this week. The rumor saying that KIOXIA is considering to resume their capacity. Basically, I asked to KIOXIA. They said no comment. But I believe definitely they are start to increase their capacity in their factory. Definitely they have to. Why? I was in Tokyo two days ago to ask for 30% extra supply from them. I need extra 30% because my design win. My design win. Customer keep complaining they got not enough supply, which impact their systems shipment is a big amount of their revenue. So I need my supplier to improve their output. They're able to have a better allocation to me. Yesterday, I got the other, message from my supplier. They're saying their second half, they are going to cut my 70% allocation. 70%, not 17%. 70%. Why?
Because they got so strong demand from the data center. Of course, my response to them is totally unacceptable. The reason is when market was slow, we're helping them to accumulate big inventory, helping them to use their NAND to go to the design-in activity. Finally, design win happening since the last second half. The business is growing. If they cut my 70%, who are capable to help my customers? So answer is no. So then the question is, my analyst saying that when KIOXIA increasing the NAND supply, their market will crash. Again, this is not going to happen. If the NAND price keep going crazily, what happened? The system company eventually they have to cut half size of the capacity because the price too high. If they cut half size of the capacity, then the NAND market will turn to the bubble, which will kill the industry.
This is totally not acceptable. The best scenario is NAND company right now, I believe with the selling price, they start to have a much reasonable gross margin with a reasonable operating profit, net profit. We hope they can increase the capacity to keep continuous supply with a reasonable price. Then Phison will be more than happy because we got so many design win business. So we're happy to see Japanese, Korean start to resume their capacity. This, it will be healthy to the market. So price stable, everyone happy. System company happy, supplier happy, we are happy. Okay. The gross margin, last Q4 all-time high. We believe first half this year, we can keep the high 30% of gross margin, okay? So this is not just because of inventory gain but we have a lot of design win, which customers they're willing to pay premium to Phison. Okay.
Talking about inventory, since Q1 last year, we start to reverse our write-off, okay? So we believe, because the market price is going up. So 2024 Q1, Q2, we believe we will still reverse our write-off. And based on their Q4, we still the write-off portion still have like 7-something%, 7.00%, 7.00% of the inventory. So means the inventory start turn to healthy. Okay. ESG anyway, Phison, we keep doing a lot, doing all our best to in the ESG. So every ranking, we are improving, gradually improving by every quarter. So this is to show our commitment to our environment, to governance, and also to the social activity. Okay. So above is my presentation today. We have like 25 minutes left. We can go to our online. Phison? Hello, Ray? Ray. Hello, Ray? Ray. Can you hear me? Sorry.
I was on mute. Yes. So yeah. So, like before having questions from investors, maybe can I follow up some questions? Yes. First is that, like in the past quarters, we also have some like very good contribution from NREs. Can you also share about how is the NRE contributions in not only last quarter but also maybe ongoing quarters? What do you think about those customers' feedback and also NRE contributions? Okay. NRE-wise, Phison, we have several NRE. Mainly is design service to the NAND company and making customization to the system company. NRE-wise, Q4 we keep flat versus Q3. And this year, we expect NRE increasing due to we get some requests from the CSP company, storage company, NAND company, and some more industrial partner asking for customization. So NRE-wise, it's stable. I don't expect NRE can grow rapidly, but stable.
This year, my goal is to improve our revenue in our products, both controller and modules, okay?
Got it. Thank you. And also, like, it was pretty impressive, when we share about the AI, aiDAPTIV+, solutions. So, can we share more about like the potential clients, and how it works with the current like AI system, and maybe how can we price this, and what's the potential revenue contribution to Phison?
Okay. Talking about the AI, right, we have to define end user, the user, they need to have a server, which server including the GPU card, GPU card with the Phison aiDAPTIV+. This is Phison aiDAPTIV+. This is server with the GPU card. And they also need to have some software platform, okay? Then end user can start to train by themself. We are engaging server.
We have more than five server companies, branded or ODM, doing the POC. Some of them will launch their server next week in the GTC or week after in the GTC. Next week, we are also going to have a partner to jointly develop the Phison aiDAPTIV+ with one of the key players in the ecosystems. They are going to provide a service. Means they have a user interface. They have a software to let the end user easy to use in fine-tuning. Of course, we are working with the GPU supplier to optimize our aiDAPTIV+ into the GPU systems to make better performance in computing, to reduce a much dollar amount in the systems. This is ongoing. After GTC, by end of the March, another GPU company will also launch the architecture with Phison aiDAPTIV+. We believe this is very new concept.
We need to prove many, many POC. And the good thing is Phison, we donate few set of a server to the National Yang Ming Chiao Tung University. Many AI professors use this. And next Friday, they are going to present their final fine-tuning result. And this is going to be public. So I believe the products will start to go to popular after second half of this year. And Phison revenue will come from our SSD we name as AI100 with the Phison's middleware. So revenue will happen by second half of this year, okay?
Got it. Thank you. That's very clear. And, maybe lastly, on the NAND supply related, so, we did, I think, share about like maybe some suppliers are cutting supplies by like big amount, 70%. But as such, like some suppliers like KIOXIA will be more supportive.
So can you share about, like, what's your view on, like, into the second half? Like, like, are we seeing some potential shortage on the NAND supply or like we can still have a pretty good, secure on the supply chain?
Personally. Okay. Personally, okay? I cannot represent the NAND company, okay? Personally, I believe the NAND company already start to resume their capacity, I believe. Of course, in the public, somebody said no. Somebody said yes. Somebody said no comment. But I believe they are starting. But since you start to resume your capacity, this is not easier. Equipment need to warm up, need to, you know, final adjustment. And wafer in and out, you may need 3-4 months, okay? So even though the wafer in, the output will be after 3-4 months.
So that's why now I'm facing a big, big challenge is I have a demand. I have an order, but I don't have flash. So that's why I need my supplier hopefully they start to resume their capacity, okay?
Yes. Got it. That's very clear. Thank you. Yeah. So, let me check if, like, investors can please hold your, please, place the raised hand and we can, put you into question. Thank you. The, KGI, Tony Hsu. Tony. Thank you, Tony.
Hello. This is, KGI, Tony. May I know more about the, mask cost, in terms of, fourth quarter and, first quarter this year? Will there be any, quarter-on-quarter increase or decrease impact? Thank you.
Okay. This is hard to say, okay? Phison, we are, count their mask as expense. So when we tape-out the chip, then happen by the day, okay? This quarter, we have a few 12 nm tape-outs.
12 nm is not that much expensive. But again, the mask cost is there, right? This year, we are going to have a 2 to 3 piece of a 7 nm, which is super expensive. May happen by second half this year. So means Q1 and Q2, the mask cost is not as heavy as last year Q3, okay?
Okay. That's clear. Thank you.
Do you have follow-up questions? Okay. I also see, Jason from CLSA, did you raise your hand?
Yes. So yeah. So can I start my questions?
Yes. Yes. Okay. Go ahead.
Okay. Thank you. Thank you. Thanks for taking my questions. So my first question is in terms of your growth momentum. So, just wonders, what kind of growth growth drivers for Phison in following quarters? Should we expect that the shipments can be the major drivers or the pricing? Thank you.
Okay.
Of course, when you go to the client controllers, we gain a few new component NAND companies as controllers. But unfortunately, 2023, they are adjusting their inventory. So we see the new demand, new focus. So in controller-wise, we're doing good. In the client SSD, we are working with a few Tier 1 PC makers. We customize our controller module to them. And this revenue coming from Q4 this year. So this is a big amount of modules business. We also win also some SSD modules to the Tier 2 SSD, Tier 2 PC makers. So we and we also believe the NAND supply will get very tight Q2 and Q3. Since we have allocation, we are able to support our ecosystem partners. So this can also turn to our growth. So go to the auto. Auto growth is slow. It's slow.
But we are gradually getting a lot of AVL. This growth will happen second half this year. But main growth will happen by 2025 in automotive, okay? So very soon, you can see a lot of press, a lot of joint development, a lot of activity with our auto partners in the market. Back to the mobile. Actually, mobile in the last two, I mean, Q4, you can see the China demand is good. We gain a lot of controller demand. Also by the eMMC UFS itself, we gain the number. So this is the other, actually, this is the other growth to Phison. In the coming years, we are going to launch several UFS controllers. This will help us to gain middle, low, end phone makers in China. And Phison also going to have joint venture with our India partner by end Q2.
Then Phison can take the India mobile storage business through our India joint venture. You can count like our China joint venture, we are copying one into India because India hope they can develop by themselves. So mobile, we are going to gain not only in China but also in India. Gaming. Gaming, I think due to the inflection due to their interest, gaming is a kind of a slow but looks like Phison is very unique players. So we only 7% of attempt. But second half, we have a new model working with the console company. We believe this can turn to a big revenue and go to the enterprise. If you ask me what is a big growth for Phison this year, will be enterprise because last year, we almost nothing, okay? But we are ready. So this year, this revenue-wise from enterprise will be a big jump.
Next, industrial, again, I'm saying that is a gradually improved, increased. But this is slow. So don't expect too much. If you're asking about growth, industrial is stable but no big growth. Next is enterprise. It's totally new. We can see some forecasts already happen by Q3 and Q4. But turn to revenue will be 2025, big growth. Then go to the aiDAPTIV+. We believe revenue will start to happen by second half, most likely by July, August, okay?
Okay. Thank you. So yeah. So, my second question is in terms of your, enterprise business. So it seems like you had, several qualifications, especially, in enterprise clients. So, when do you expect that we can have, contribution coming from maybe CSP clients or, or data centers?
And another question is, can we provide a total solution including NAND storage or SSD controller or even retimers to provide a bundle or a total solution to our maybe data center clients so with it we can expect more contributions or design wins or you know any potentials upsides coming from this kind of segment? Thank you.
Okay. First question. We believe we are working with a lot of AVL, not only with the CSP but also with their server OEM and server ODM. We believe that revenue will start to come slightly from Q3. Q4, we can see significant number. And a lot of activity ongoing. Unfortunately, I cannot show you what we are doing there. But we have confidence Q4, significant growth on retimer. Second question. The answer is yes and no.
Basically, those CSP or server company who in charge of retimer and the storage is a different group, totally different group, okay? Different person. But we have a good, you know, the synergy. When we talk to their data center supplier about SSD, we also bring in our retimer. Since they have a good, they are comfortable about Phison technology, service, flexibility, blah, blah, blah, in SSD, they are happy to test Phison's retimer. And we also see some CSP server company, their server current retimer, maybe price, maybe service, maybe technical issue, Phison show very hardworking, fast react to them. They're happy about Phison retimer. They put Phison test and go to AVL. On the other hand, when they're happy about Phison retimer, we also say, "Hey, how about bringing our storage?" So this can be a big leverage. But it's not 100% combined, okay?
Okay. Thank you.
So my last question is in terms of your controller business. So I think precisely you mentioned that we see the trend that all the NAND manufacturing players are outsourcing the data centers controller to the professional design house like Phison. So do we expect this kind of trend or demand will continue in the future? So, how's our outlook in terms of the?
Okay. I believe the answer is yes. But this is not going to happen tomorrow, not going to happen next month, not going to happen next quarter. Most likely, this may start to talk second half this year and happen by 2025. We have confidence, okay?
Okay. Okay. Thanks, Jason, for your question. Thanks. Jason, do you still have questions? Sorry.
This is like, let me repeat my question.
So my last question is in terms of the, your SSD controller business. So precisely you mentioned that, the.
You're talking about outsourcing, right? Outsourcing, right? I just asked already. I answered already. I said yes. But not happen tomorrow, next quarter. But most likely, start to actually, we start to talk something already. May happen by end of this year. And we can see the result in the 2025.
Okay. Thank you. Oh, I have no more questions. Thank you.
Okay. Thank you. Then, let me pass the question, to, Nomura Donnie.
Oh, hi, K.S. Thank you for taking my question.
Yep. Yeah.
My first question is regarding to you have, qualitative guidance about that business will be growing rapidly in 2024 and 2025. And you have answered some of the progresses in different, products.
But just curious if you, considering the strong NAND price as well in the near term, just wondering if you could give us some rough idea about the rapid growth in 2024 and in 2025. Is that what kind of percentage will be driven by NAND price? And what kind of percentage will be driven by all the efforts we've been investing into this new product? And also you also mentioned about the supply resumption timeline could be a couple of months lagging behind at moving the equipment. Wondering if you could give us some color on, you know, those NAND makers' production resumption time period in the coming quarters. And when do you think their production will be back to normal level? Thank you.
My perspective, based on today's price offer from their NAND, every NAND supplier, they already overcome their cash flow problem.
They're also going to their OP is positive. I believe this happening. So of course, a NAND company, they wish the price going double again. But this is not really going to happen as this I'm showing here. If price keep going high, the system company will cut half capacity, half capacity into the system. So this then will kill the market. So I believe with a reasonable NAND company growth margin rate like 35%-40%, it's really healthy company is healthy business. In this case, I believe the flash price already going to touch the top of their selling price. And we believe we also hope that NAND company start to increase their capacity because the market need the more component. The market cannot accept the price going too much high.
If this is going to happen, demand increase with a reasonable profit, then NAND company start to accumulate profit. Then by Q2, they may announce to start build the new fab. Then the industry of NAND start back to their healthy cycle. We hope this is going to happen, okay, by Q2. Okay. So, yeah. So back to the first one. So rapid growth in 2024 and 2025. So sounds like, 2024, for sure, there could be some support from the NAND price recovery from very low base in 2023, right? In 2025, what kind of, growth in 2025 you are seeing is I think that's not just purely depends on, NAND price but more like the, the new products growth, right? Yes. We have a big confidence on aiDAPTIV+ products. We talk to every server makers, ecosystem users, government officers.
They all agree aiDAPTIV+ will help the human to have a low entry order amount, start to have their own fine-tuning. If this happened, this equipment will go popular very, very rapidly to the market. Phison, we invent this solution. We have our software, our patent. This will help Phison to grow our revenue. No doubt, we need to have support from the GPU company. We wish after we launch the products in GTC in end of March by one of GPU makers and Computex, we hope GPU company to endorse these solutions to help the human able to use a small amount of dollars, start to their fine-tuning. Adaptive will be one of big growth to Phison, and no doubt. Retimer this year, I think Q3, small revenue coming, Q4, gradually every quarter improve. I cannot tell you how much I can make it.
But everything start from nothing. We have good confidence. We are going to gain share. Industry, again, industry is very slow, okay? But it's our good, fundamental revenue. Enterprise. If you're asking me again, enterprise will be the much, much replica because this is a high terabyte. Every piece can be a few thousand dollars. I just give you a number. I just got a 1,500-piece SSD value is a $3 million. So this will helping us rapidly growth in their revenue. Then gaming. First half usually, first half gaming is slow. But go to the second half, Christmas season, summer season, gaming can helping us to gain our revenue. Mobile. Mobile, not only controllers but also module in our China partner, our company. India, we are going to have a joint venture. This can be a new growth.
So in the mobile module, we are very optimistic. Auto. Again, auto slow. But we can see the growth next year. Last is the SSD for PC, for retail, for aiDAPTIV+. So this is the other big growth. Hopefully, I answered your question.
Yeah. Understood. I just have two follow-ups here. So the first one is that, I understand that there are strong pipelines in terms of the new products. But, considering you also mentioned about NAND makers will resume some production or rebuild the fabs, just curious what's your view on the supply demand situation into 2025, regardless of the strong new product pipeline we are seeing at this time point? And, another follow-up is, would you elaborate more on what kind of system companies we are working with on the AI aiDAPTIV+ system?
And also, can you give us some quantitative guidance in terms of the sales contribution from AI aiDAPTIV+ system as well as retimers in this year? Thank you.
Okay. If second half this year, if no any NAND makers announce to build a new fab, 2025 second half, there will be crazy shortage, okay? So we wish NAND company start making money, Q2 start to announce to make the new fab, start to order equipment. NAND equipment, factory can be ready within a year. Then start to ramp up. Then second half 2025 will be healthy. If not, 2025 second half, I think everyone have to suffer the high price on NAND. So we wish they start to build a new fab by Q2 or Q3. Talking about their aiDAPTIV+, yes, we have many activity with, again, with their ODM, OEM, SI.
But unfortunately, I cannot tell you how big the market, the dollar amount. But I think everything can be more clear after Computex. But again, we are very optimistic.
Okay. Thank you, K.S.
Donnie, do you still have questions?
Oh, no. I'm fine. Thank you, Ray.
Okay. Thanks, Donnie. Next, we'll have Simon Woo from BOA. Simon, are you with us?
Looks like he muted off.
Okay. Maybe let me ask a question from the line, and then I'll come back to Simon later. So, there's a question from the chat box, like SMI said that it expects revenue from NAND supply to increase by 50% this year. Did we see the same?
Okay. SMI saying that. I'm not SMI, right? So I'm not have to comment about. I don't know if SMI counting the NAND supplier from China or not. So this is a question.
But Phison, we believe, again, Phison's revenue, when you look to my distribution, right, is like 30 ± % of my revenue. So I'm I cannot telling you, you know, our, our controllers portion. But Phison company policy controller only sell to their NAND company and Kingston and our China company. That's all. We didn't sell controller to any Shenzhen module house, traders, brokers. We never. So if you ask me, most likely, Phison controller go to the NAND company. So this is totally different with the SMI's business model. So when SMI saying that, they gain the share from I mean, the revenue come from NAND 50%. On the other hand, they lost their China, Shenzhen module house controller. Why? Because the price is broadly in there, okay? Yep. This is my answer.
Okay. Thank you. Back to Simon. Are you with us? Okay.
It seems like there's a line issue. So yeah. So maybe these are all the questions for today.
No. I saw 1 question about the PCIe Gen 5 design win. Yes. We have a few. But I cannot disclose who. But we have a few, more than two. Okay?
Okay. So yeah. So I think that's all the questions we have for today. Yeah. So thank you, everyone, for participation. Sorry? Is there any?
Hello, Simon?
O kay. So.
Yeah. Sorry. Can I ask 1 question quickly? Oh, great. Can you hear me?
Yeah. Sure, sure. Please go ahead.
Yeah. So yeah. K.S., your presentation great. But yeah, I think a lot of the global NAND capacity based on the still 112 or 128 layers, not the 176 or 200 layers above.
So if the NAND makers increase their utilization ratio, their concern is, they are having lots of the legacy NAND chips rather than fresh, new, high-density NAND. So do you still believe maybe Japanese, Korean memory chip makers can increase easily to their NAND fab utilization ratio? Again, we do see the global NAND capacity still very, very obsolete, old.
No. Actually, I don't care, right? We use fresh. I don't think the old technology or new technology, we just use the memory, right? So I believe I can see some of my NAND vendors start to migrate to the 200-layer above and quite smoothly. And of course, the U.S. player already ramp up in the 232, China already 232. So I no much worry. But what I can see is the Korean supplier, basically, they are a little bit behind.
But I believe they are going to catch up.
Yep. That's a great point. So maybe very quickly, your aiDAPTIV+ business model really makes sense with AI. But the, it sounds not much relevant to the cloud storage. So that means you believe lots of the B2B, even the B2C customers, they're going to establish standalone SSD or storage device rather than relying on the hyperscalers, cloud?
Okay. I lost your voice. But I think I catch your question here. Basically, Phison, we have two solutions. One is aiDAPTIV+ edge. Mainly is for their B2B2C not B2C. B2 small, medium-sized company, I think this is mainly lower the entrance ticket, entrance amount. This is going to be popular. And the big unit is going to ship by second half. And we also have to prove. We prove.
With aiDAPTIV+, we are able to help the cloud to utilize their GPU to improve performance. So this we need to make many POC. But this we need to endorse by the GPU makers. So this may happen after one or two quarters.
Yeah. But these guys, I do remember, they prefer maybe a large NAND chip makers, SSD solutions. So I used to confident to compete with a big NAND chip makers like Samsung to sell your SSD solution.
Okay. First of all, Phison, we never compete with the NAND player because we are too small to compete. But I would like to tell you, Phison is a complementary with the NAND company. It's easier, right? NAND company, if they invest their resources, human resource, to build this solution, they are not able to sell as SI because the demand in this is not storage. It's in computing.
So they cannot sell as many NAND they wish. But fortunately or unfortunately, Phison already, many patents. We have many patents in our aiDAPTIV+. Yes. There's the NAND. Every NAND vendors very close with Phison. They are my customers. They use my design controllers. I'm also their very close customer, helping them in the last year. Especially Korean supplier last year was tough with helping them carry the inventory. This year, we're helping them to make a lot of design win. I agree to you. The large cloud, they prefer to buy from NAND company. But NAND company, if they think they're going to invest, the return is small because of low NAND. And Phison also will feel uncomfortable with our patents. So most likely, we are going to work complementary with the NAND company. We provide our design service. We believe.
Yeah. That's great. Yeah. Great point.
Maybe, lastly, I know it's very late. But the one last question. Yes. Everybody saying memory cycle, semiconductor cycle getting better. But because of, you know, NVIDIA, NVIDIA or AI theme, we guess, when we look at the traditional, even the China smartphone, PC, the even the conventional servers, we still see quite weak sales through or timeline growth. So how do you assess the Q1, maybe Q2 overall, the net chip demand except AI-related things? Thank you.
Okay. Again, I know the question. But to be honest, my answer is this is not my business because I cannot manage, right? But I believe with today's agree, consumer very weak. But for data center, actually turn to very strong. You can go to call the Korean supplier, turn to very strong, okay? A lot of AI system built from nothing is strong.
So I don't know overall, there can be just a little bit shortage. But Phison will grow too much because with the new application. So if you ask me, I wish the demand is not that much strong. Then Phison able to get the more supply with the much reasonable price. I wish. If the NAND going to strong like 2021, crazy, then I die. All my investment gone because I cannot get flash. I cannot turn to revenue. So I hope what you say is right. Then Phison able to get the more enough flash with a reasonable price.
Yeah. That's a great point. Yeah. I think already, over 6:00 P.M. But, Ray, it's okay. One question from Europe, one European investor is asking one question through me. So it's okay if I read it.
Sure, sure. Yeah. Please go ahead.
The long-term investor's question is your business model with a working capital burden. According to your here the presentation material, December quarter inventory, TWD 24 billion, right? However, your COGS number is about TWD 10 billion, which means already more than two times of the COGS number. I mean, the cost of goods number is inventory. So this is a normal inventory amount going forward versus previously much lower than this. So what's the reason? And then you're going to retain this kind of the high inventory level through the rest of 2024? Thank you, sir.
Okay. 2022 and 2023, I have a high inventory. Not because I wish. I have no choice because NAND company came to call help. It's a long-term partnership. I help. So I carry high inventory. But this year, I need high inventory because the visibility of supply is unclear.
But my demand is very clear, okay? So I don't care about whatever you call the percentage healthy or not healthy. I need component. I have a real demand. So one thing I want to share with you, if you saying that $24 billion is high, but Q1, I'm much higher because visibility is low. Simon, I showed to you, yesterday when I left Taiwan from Tokyo, I got a call from Korea. They said, "I'm sorry. I cut you 70% second half." I said, "What the hell? You cut me 70%?" Okay?
Yeah.
So the smart people now is to keep high inventory for my next quarter business.
All right. So clear, K.S. And thanks again on your great, you know, presentation today. And also, congrats on the 2023 great result. Thank you.
Yeah. Thank you. Ray?
Okay. Yeah. Thank you, everyone. Thank you, everyone, every investors.
Thanks, Morgan Stanley team, K.S., for the great presentation and Q&A. Yeah. We will conclude the call here. Thank you.
Thank you. Have a good day. Thank you.