Alchip Technologies, Limited (3661) Earnings Call Transcript & Summary

Dear investors, portfolio managers, analysts, welcome to our fourth quarter '21 Institutional Investor Conference Meeting. Today's meeting will be hosted by me Daniel, CFO of Alchip; and our CEO, Johnny Shen. We will quickly go over our fourth quarter operations and the financial results and, of course, the 2022 full year outlook, and then we will go into the Q&A session meeting. Now as usual, this is safe harbor disclaimer. And this meeting will be in English. If you need the Chinese presentation slides, please go to [Foreign Language] to download the Chinese version. And for the Q&A session, we also welcome the Mandarine questions. Johnny and I are all good with Mandarin. And please, participants, please, write down your question to host through room's message function if you don't want to speak up your questions. And please leave your company's name. [Operator Instructions] And the video and audio content will upload to last maybe after 2 hours of the end of this meeting. So let's start. Johnny?

Right. Good afternoon, ladies and gentlemen. Yes, this is Johnny Shen, President and CEO of Alchip Technologies. Once again, thanks for attending investor conference meeting. We appreciate the opportunity to share our 2021 results and also provide the guidance for future business. In case you are not too familiar with Alchip, allow me to make a company update and identification. Our company is founded in 2003 and IPO in Taiwan Stock Exchange in 2014. The current employee is a little bit over 480 people. 3/4 of our employees are engineers. And since we found the company, we have been successfully tape-out more than 480 designs in leading-edge technology. Revenue, yes, last year, we achieved $372 million. in terms of capacity last year, we concurrently support more than 30 design activities. More than 80% of our revenue coming from HPC and AI area in leading-edge technology. We are one of the VCA member value chain aggregator for TSMC.

[Foreign Language]

Please mute your speaker.

Yes, next page, please. On 2021 recap in summary. We can say we had outstanding years with record-breaking numbers in all categories: revenue, $372 million; operating income, $65.3 million; net income, $53.2 million. EPS for NTD around $21 (sic) [ TWD 21 ]. Now we are the -- our market cap is more than -- is approximately USD 3 billion. Overall, 2021 yearly results outperformed the '20 number we made at the beginning of last year, despite the incident happened to 2 of our customers. All detailed number analysis and comparison will be provided by Daniel in a later section. 2021 achievement, not only just in number, we also established many new business in different applications, including cellphone image sensors, smart mix speakers, automotive and cryptocurrency. We have multiple tape-out and design wins, mainly in 6, 7 or 5 nanometers. Of course, HPC and AI, including CPU-GPU, GPGPU, CoWoS-related business demand remain very strong from existing customers and also from many newcomers. As for the mass production forecast, we received incredible volume forecast from key customers. MP revenue well start to surge from late Q2, all the way last for at least 8 quarters. The capacity limitation is our major challenge. Since the business demands too high, because our resource shortage, wafer, CoWoS and substrate shortage limits the MP volume. Fortunately, with the current allocation from our major suppliers, we can deliver a reasonable amount of MP number this year. Of course, we are still working aggressively on the upside for this year, also demand for the next year. Lastly, a quick update for U.S.-China trade war tension. Yes, as you know, there are hundreds China customer on the so-called entity list, including 2 of our customers. After working with the U.S., China and Taiwan lawyer, suppliers and customers representatives, we figure out a rule and procedure to handle the situation. We have an internal SOP to minimize and anticipate potential impact in case there is any customer or partner on the list. Once again, thanks for your trust and patience, we have confidence to continuously grow our company and deliver good results. Thank you.

Okay. Let's get into the financial part. For fourth quarter last year, the revenue, actually -- we already announced the revenue and the bottom line numbers. So this is just a recap. For the revenue last quarter, our total revenue is $89.3 million, which is 2.8% quarter-on-quarter decline, but 34.4% year-on-year growth. For the net income in fourth quarter last year is [Technical Difficulty] please mute your speaker. Okay. For the net income last quarter, it's $12.7 million. It's a 1.7% quarter-on-quarter decline but 55.2% year-on-year growth. For the EPS translating into NT dollars for fourth quarter '21 is $5. For the yearly comparison for 2021, as Johnny mentioned, the numbers in every category is all record high. For the revenue -- total revenue last year is $372.3 million, up 55.4% year-on-year, and the operating income is $65.3 million, up 95.4% year-on-year. For net income, it's $53.2 million, up 88.2% year-on-year. For the EPS for last year is TWD 21.3. And for the revenue breakdown by application, you may see that the HPC remains the majority part of our total revenue. For the fourth quarter last year, HPC accounts for 78% of our total revenue. And the niche market accounted for 7% and the network account for 9%. Consumer sector accounts for 6% of our total revenue for the first quarter last year. And therefore, the 2021 as a whole, so the HPC revenue attributes about 81% of our total revenue, and the niche market accounted for 10%. And there was a little bit networking, and 7% of the revenue came from the consumer sector. For the process note, it's the company's goal that we are aiming to becoming the leader of the leading-edge technology node design. So for our revenue mix in terms of the process node, for fourth quarter last year, for 7 -- revenue coming from 7-nanometer project or even 6- or 5-nanometer project accounted for 61% of our total revenue, 27% comes from 16 and 12 and 28% -- and the 29-nanometer contributed about 6% of our total revenue in fourth quarter last year. You may see, over 94% of our revenue are in 28 or advanced technology nodes. For the yearly breakdown, for the 7-nanometer, 6-nanometer, 5-nanometer accounted for 60% of our total revenue and let's say 16- and the 12-nanometer accounted for 28% of our total revenue. We can proudly say that in terms of the process node in our industry, Alchip is, no doubt, the leader. And therefore, the revenue breakdown by region. For fourth quarter last year, the revenue from China accounts for 58% of our total revenue. But you may notice that the others accounted for 24% of our total revenue last quarter. The majority of the others came from the North America region. In this year, we believe this number will go up quite a bit quarter-by-quarter. For the yearly numbers, the 71% of our total revenue came from China, and 17% comes from other regions, which also referred to mainly North America region. And for the business review, I guess many of you may know that the fatal incident, the entity list thing. So in our view, despite these, there was a geopolitical obstacle. We're still see a very promising result last year. And for both -- the reason behind, I think, from a big picture perspective, both China and the U.S. markets showed very strong physical design demand for the HPC applications, which include CPU, GPU, AI, DPU, many, many applications within the HPC area. And we won several high-volume production projects, and the delivery will start to happen sequentially in 2022 and 2023. And for the profit front, last year, our profit margin hiked a little bit on multiple reasons. First of all, the 2021 gross margin was 34.2% versus 32.6% in 2020. The reason behind this, a little bit hiked gross margin is, first of all, the design revenue, the design business was very strong. As I mentioned many, many times to investors that for the whole industry, actually, the demand is much, much, much higher than the current supply, especially within the leading-edge technology node and the complicated design. So that's the major reason for the margin last year. And in the same time, the technology are still moving towards the more advanced technology nodes. We think for this year, 2022, this trend will continue. And for last year, another reason for our gross margin is because the higher revenue exposure to high-margin production projects. Some of you may know that for the first half last year, our production revenue shipped to certain customers was relatively high gross margins compared to the other production revenue to the other customers. So that's also one of the reason for the -- for our profit margin performance. And for the business outlook, revenue for this year, 2022, we believe the revenue to the U.S. customers is the main growth driver for the whole year. The production shipment for AI chip to our U.S. service customers is expect to surge in the middle of second quarter this year. The demand is very strong. And because of the supply shortage, we are unable to fulfill all the demand from this customer. And secondly, on the demand from the AI application of U.S. customers keep on increasing. I would say, the visibility for now already extending to 2023 next year, and the orders from this customer is greater than the orders this year. And again, for the industry wide and the physical design turnkey demand for HPC, and that's stronger and stronger. I knew that some of the investors are thinking if the HPC trend already into the peak phase. But in our view, we still see the overall demand for HPC is still climbing. And actually, I would say it's keep on accelerating in 2022 or even in 2023. So the project pipeline of Alchip is stronger than ever right now. Mostly most of the projects are within the HPC application. And of course, the majority of them are 7-nanometer or even 6- or 5-nanometer projects. For this year because of the significantly rising production revenue of production shipment to the U.S. customers, the 7-nanometer will remain the mainstream of our revenue source. In the meantime, we have multiple 5-nanometer projects to enter the design phase in 2022. And we will have our first full masked 5-nanometer budget tape-out in late -- in late second quarter or the early third quarter. And yes, again, as Johnny mentioned, the supply chain management is the major factor for our 2023 top line -- for our 2022 top performance. Currently, we are positive and optimistic for the supply. The CoWoS and the substrate capacity are currently very critical factors for our growth. And for these 2 projects, we are seeking every way to ease or to increase the supply for our customers. So far, we have some CoWoS and we are still optimistic to the second half this year or the whole year of 2023. Hopefully, we can get a very good support from our suppliers. And the current capacity allocations are far short from our customers' orders but even, in this case, we are still confident we can deliver a very promising growth in 2022. And I guess that's the presentation for our investor conference meeting today. And thank you for your participation, and we are going to enter the Q&A session. Thank you.

[Operator Instructions] Ask please.

Okay. Johnny, Daniel, congratulations on the good results. My first question is regarding your business outlook. Could you provide your sales growth expectations for this year? And maybe you can also try to provide more detail on the major projects for your design and also turnkey business respectively.

Okay. Haas again, because we -- okay, sorry, the TWSE gives us a very clear instruction that we cannot disclose the numerous guidance for you. So I would say this way. I noticed that the current consensus of our total revenue in 2022 has been moved from $460 million to $480 million to about $480 million to $500 million. For the consensus, there's the higher end of $500 million. We have confidence that this number is not very difficult to achieve. That's the guidance for this year. And of course, I have to emphasize that how many capacity of CoWoS and ABF substrate we can get dictates the upside of our total revenue. And of course, the major growth driver is the instance chip shipment to our U.S. customers. The demand is very, very strong. I told some investors that the total -- the scale of the order from our customer is far beyond the supply we can find. So -- but even in this case, the shipment volume in terms of the -- and the -- in terms of the total value to this customer for this year is very significant. And let me give you another hit that for last year, the revenue to North America accounts for about 14% of our total revenue. But this year, we do believe the revenue to U.S. -- the revenue from North America region will account for more than 40% of our total revenue. And that the service cost -- the revenue to the service customer will account for the majority of this 40%.

Okay. My second question is about your current progress on the node migration. As one of your U.S. peers overnight on its past conference call mentioned that ASIC will stay strong in the next couple of years, and they're migrating 3-nanometer. Could you discuss your expectation for the mid- to long-term growth outlook for the ASIC market? And as you mentioned, 7-nanometer will be the major growth driver for this year. But looking beyond this year, when do you think 5-nanometer will start contributing more than 7-nanometer for your business? And when do you think 3-nanometer contribution will start?

Okay. Let me give you the answer. And I guess for this part, Johnny can give you more color. This year, of course, because of the production revenue to the U.S. customer, 7-nanometer will still be our mainstream technology node for our company's revenue. But for the 5-nanometer, the revenue contribution from 5-nanometer or even more advanced technology node will increase significantly. For last year, we have 2 5, 7 -- 5 -- look, we have 2 5-nanometer projects. One is the test chip to U.S. customer list and another one is the full-mask project to the Middle East customers. But for this year, there are multiple 5-nanometer projects from U.S. customers and the China customers. And of course, there will be 6-nanometer projects, and there will be -- most likely, we will have our first 3-nanometer budget kicking off this year. Johnny, can you -- do you want to add in -- add something?

Sure. Let me add more. Just like Daniel mentioned, the major revenue driver will be 7-nanometer. I think that's mainly because of production. But in terms of NRE, 5-nanometer for sure will be the major driver for NRE. Right now, most of our customer, new customer, they are pursuing 5. The existing customer, the 7-nanometer already tape-out. The next project definitely will be 5- or 4-nanometer. So the 5-nanometer NRE contribution for sure will be the highest amount of design nodes. Also 3-nanometer, we're going to have a test chip tape-out by June time frame. But in fact, some customers are quoting us for 3-nanometer full turnkey project. Most likely, we will engage a few, but tape-out will not be this year or even next year. I think based on our estimation, the 3-nanometer tape-out will be early 2024. The test chip will be -- will come out will come back sometimes next year. But full-mask production won't start until 2024.

Okay. And a follow-up. Could you quantify how the wafer price hike and substrate supply constraint is impacting your sales and margins outlook? Do you have the capability to fully pass on the growing cost to your customers?

For the price hike on the materials like the wafer, like substrate or the OSATs, we can easily pass through this cost increase to our customer without a problem. And as I mentioned to investors that some of our revenues, the pricing scheme is cost plus margin. So there's no problem to pass through the increase in cost. And for the substrate shortage, I said that we are trying to do something to increase our substrate supply. And all these things are in progress. And currently, we are very positive of it. And that's good for 2023, I think 2022 is almost that. It is very legal to find the large portion of extra support. But we do have some room for improvement. But for 2023, since the orders from customers are still very strong, and for this time, we have time. And we already engaged with the suppliers for the 2023 supply. I believe -- I personally believe that the supply situation for both CoWoS and substrate will be better in 2023 than in 2022 this year.

Okay. Let me try to add more. Yes, the substrate shortage, I think this is everybody's problem. Yes, I think -- but to us, I'm thinking about there's upside for Alchip. As you know, the substrate supplier, in usual case, they won't talk to each of the individual customer, unless they are very big. Even our #1 customer in U.S. is very difficult to get their allocation. But they would like to talk to us and discuss with us because we are representing more than 30 customers. So the support we get from the ABF for substrate house we can say is very good, and they even talk to us for the further cooperation. So in terms of price, I don't think that's an issue. All our customers are willing to pay the substrate price as long as we can get the volume. So far so good, yes. As I mentioned, substrate shortage will become well for our weapon to secure our customers.

Okay. Next one, Szeho.

I have 2 questions from my side. So first one also again on the back end, on the back-end CoWoS and also substrate. It seems like you are starting a little bit more positive, right, in terms of securing the capacity. Can you share with us maybe for this year and next year, what would be the fulfillment ratio relative to the forecast you gave for those back-end suppliers?

Okay. For the substrate supply, I will say, for this year, the major customers in U.S., the orders fulfillment rates, for now, it's about 42% to 43%. But the confidence level for us to achieve about 5% to 50% is getting higher and higher since we have getting some commitment from the suppliers, but it is impossible for us to fulfill 100% of customers' needs for this year. That's the current situation.

Still getting better compared to let's say 3 months ago.

Could you say -- your voice is a little...

It's still getting better let's say compared to 3 months ago.

The substrate -- I can tell you, everybody in this industry is looking for substrates, of course.

True. True. Okay. Sounds good. And second question, definitely, going forward, you will target on projects with high volume and production volume, right? So I just want to check with you guys, let's say, going into the 5-nano project, what would be the revenue split between, let's say, NRE and also the production for any specific, let's say, 5-nano project after mode.

It is very difficult to tell because we have so many projects in different applications. Different application derives different ratio. So yes, it is very hard to answer you. But again, the larger is if you -- when you invest more upfront in the NRE, reasonably, we will expect higher production revenue from this project because the customer can always make money on production. They cannot make the money back from the NRE. So that's the logic behind. So for the ratio, I'm sorry, I don't have an answer for you.

Let's say a ballpark let's say, the production revenue will be, let's say, 5x or 10x bigger than the NRE revenue.

So our expectation is in terms of ratio, we'll be discussing internally over and over again. So the -- Yes, in the past year, our production NRE ratio is close to 1:1. But ideally, I think starting from 5-nanometer, 7-nanometer, the ratio, hopefully, will be 1:3, 1:3 ratio. But for this particular U.S. customer, the ratio will be -- will be much better than 1:3. So it depends on the situation. Our goal and expectation is 1:3.

Okay. Yes. Sounds good, yes. And going forward, let's say, into 3-nano, that ratio would go even higher, right? I believe.

I believe because the investment getting higher and higher, they need a bigger revenue to get better that money they invest.

And next 1 is [ Britney Len ].

I want to ask a little bit about China. Since you mentioned the U.S. is growing very fast, I'm curious to know what the expected contribution will be from China this year, the application pipeline and in particular, Phytium.

Okay. For China, this year, I think the growth mainly comes from NRE. There are so many design opportunities out, so many these opportunities out there. And since we have very good position in the China market, we can grab the best projects on the market, for sure. But for the production, as mentioned, we have entered the listed thing last year. So although everything is in progress, but at this moment to talk about the substrate supply with the substrate suppliers, it's really, really difficult. Hopefully, we can get a good result from a substrate supply. And I think the production revenue from China market, we expect the strong growth would occur in 2023. But it is hard to tell because, let's say, the reason why we don't have so -- we don't have very firm answer for you, it's because some of the projects we have done, we have taped out are for the start-up companies in China. For start-up companies, it is very, very difficult to predict the future shipment. It could be big. But again, it could be small. So for certain projects in China, we do have high expectations. For example, we did the smart speaker project with China major service providers, and we believe these projects can generate a certain amount of production volume this year. But honestly, we don't have a very good idea of how high it will go.

Okay. Let me try to add to that. So in terms of NRE, I don't worry at all because in China, to be honest, we have a very good reputation. Any new customer who want to do the leading-edge technology, they all come to Alchip. So we can say for the past year, every month, every -- even every week, we have a new customer come to us. But unfortunately, our resource shortage are limiting us to take many projects. But in terms of production, people doing the GPU, CPU, they also have a high expectation, but they also have a certain uncertainty. And because of a little bit sensitivity for that area, we're thinking about penetrating other applications. Just like Daniel mentioned, the smart speaker from the Tier 1 service provider, yes, we already won. And also the cell phone image sensor using for the multiple Tier 1 cellphone maker in China, that project, we also won. And go back to the FEITIAN, yes, we also have -- still have a good expectation on this account. And based on their forecast, I think it's still very, very s***.

I see. Just one more follow-up question that you mentioned earlier about the shortage on not just substrate the CoWoS given the technology and also the high cost of it, what's your strategy to get supply? And can quantify how much you actually need?

Okay. Go ahead, Daniel.

Okay. If you are asking for volume, I cannot give you a number because we have agreement with the suppliers. It is not a good thing to disclose the numbers because almost every customer is asking for capacity allocation from substrate vendors. But you may think this way that for substrate vendors, they want to do business with Alchip. Because by doing business with Alchip, as Johnny mentioned, they -- you can say they connect indirectly to multiple good customers with very good upside potential in the future. So even though for these 2 years or 3 years, it is the golden period for substrate suppliers, they also want to diversify their customer portfolio. Through us, it is a very good way for them to do so. So because of that, the substrate supplier, no matter you name it, Kingses, Nanya, UMTC, Tasera, they owed us very, very good support for our customers.

Hello, Charlie?

So now since you have getting a very heavy exposure to a U.S. regional customer, so my question is about how sticky this is going to be in the long term and also some potential competition from the U.S. localization? So for example, now for the U.S., AWS, right? They also have several other projects like Graviton, their own CPU designed by themselves. So what circumstance do you think this kind of key account will decide to design chips on their own since they have this kind of capability? And are we going to address this long-term risk?

Okay. Let me answer you first. First of all, we -- I cannot speak for our customers. But in our thing, I think, as you mentioned, for example, AWS, actually, the reason why they do the ASIC is to gain more control and to differentiate their services from their competitors. For them, essentially, they are still service providers. They have many, many applications and they have many, many design needs. So for now, they still put the majority of their engineering resources to the front end to the applications. For the back end, first of all, for the leading-edge technology node, they don't have too much choice. I think, in the future, there will be a hybrid structure. They may choose 1 or 2 chips to do the whole design portion by their home. But in the meantime, there are still many design opportunities there for the outsiders. And as you mentioned, we did the [ Inverse Chip ], and they did Graviton. But if you are asking about the stickiness or the sustainability of their project, I would say we already won another project from this customer. And they are -- we are almost won a third project from them, I guess, in this quarter. So we don't doubt this relationship with customers will last shortly. I think it's a long-term partnership kind of relationship with this customer for the chip design.

Okay. Yes, let me try to add to that. I agree with Daniel. I think working with this particular customer, we consider it's a win-win. To be honest, they are not so price sensitive. We are not charging them too much. I think the relationship during the design and also during the production is getting better and better. So that's why instead of -- in additional to this product line, they being referred another product line to us and fully endorse us to winning another project, which we already did. So in the future, I think we want to respect customers' decision. And we will try our best because, in terms of design, we have much more experience than them. They can -- they're only doing like a 2 to 3 tape-out a year at most. We have 30 tape-outs. We accumulate enough experience before next generation. So working with us, I think, is -- we always consider that's the best way. I told them, please consider our team as part of your company. I think so far, they are agreed of this concept. I think so far, so good, the relationship.

Yes. So from the U.S. localization perspective, right now you have a super high exposure to HPC, high exposure to the U.S. region. But the Intel foundry service, their [ PH ] is really to localize their HPC project in the U.S., right? And IFS, they are funding like USD 1 billion to enable ecosystem. So do you think there is a kind of long-term competition for you in the U.S.?

Yes, let me turn to take that. I think for U.S., if they want to establish this back-end related resources, they will take a long time. As you know, for the past 20 years, not too many engineers willing to do the hardware. So even right now, the demand is very high. But in a short time, if they want to build out their own thing, I think it's -- I consider it's very difficult. The company you mentioned about, they also have a very ambitious plan. In fact, they also come to us, yes, for this desire opportunity. Also, the localization, I'm a little bit worried about the geopolitical issue. Maybe later on, they may have some restriction using the China desire resource. And that's why starting from last year, we did start to build up a big amount of resource in Japan and also in Taiwan in order to anticipate this. But so far, among all our 40 customer, only 1 customer has a particular request, do not use the PRC citizen apply on their design. But so far, I don't receive too much restriction on that area.

Okay. And my next question is about the 3-nanometer, right? Because it seems to me, right now, you can see that TSMC's first wave customers are all huge companies, right? And I'm not sure going forward, given a very expensive wafer cost sort of design fee, photomask, et cetera. Do you expect your customer, because you're kind of competing edge [ E3 ], those are smaller volume niche projects for smaller customers, right? Do you think your customers will migrate to 3-nanometer? I know you have a 1 project, right, just in terms of the cadence or the scale, do you think a 3-nanometer is going to be big revenue stream to you in the coming years?

Okay. For 3 nanometer, basically it's a little bit too early to tell. But in fact, we worry about the similar situation for 5-nanometer 2 years ago. But after rise, we have so many 5-nanometer customer, yes. So looking forward, as long as the technology improvement and performance improvement, yield improvement, I think yield up to a certain stage, I don't worry about the business opportunity. So right now, it's already 2 to 3 potential customers talking to us about 3. But right now, 3-nanometer for me is a little bit too early. I don't think the production will be kicked off starting from 2024. And starting from next year, I expect we have a few 3-nanometer design win. And by the end of next year will be some tape-out. This year, our goal is to take a 2. I think that's on the design phase. I think that's good enough. And for this year, our focus will be 5-nanometer NRE plus the 7-nanometer production.

Charlie, let me answer you that I think the story repeats again and again. For 16-nanometer, many people said the NRE is too high. So the number of the customers will decrease. It turns out for 16-nanometer, the customer is flooding to this node. And again, these things are also happening for 7-nanometer and the 5-nanometer. To us, our view is, as Johnny mentioned, it's too early to tell. But to us, we will think within the HPC area, because the majority of our customers or the majority of the chip makers, actually, they are not selling these, and they are building up systems to provide services to their customers. So the NRE investment will become more and more reasonable. They don't -- no, I won't say they don't have to. The concerns for the end product price is not a concern anymore. So I think the situation for the HPC area comparing to the consumer products is a little bit different.

I see, yes. Yes, so I have 1 or 2 smaller questions, but it depends on you guys, whether I can continue or you want to take other audience question.

Okay. So if you have questions, please use the Raise Hand function.

Charlie, yes, okay. Just one more question, no problem, yes.

Okay. Yes. So I do this quickly, right, about China. It seems like lots of China company say they still introduce HPC chips, right, namely, for example, Pingo or Baidu do whatever or some GPU start up. So for those China customers, they don't rely on design service, how is the execution? So I just want to get a sense what exactly your value add to China accounts, whether they can really survive without your support?

Okay. Charlie, let me answer you first, and I believe, Johnny is located in Shanghai. He can give you more color about it. For my observation, the energy of China's chip design business actually right now is actually in the startup area. They get very good funding, and they -- because of the funding, they can attract very -- they can attract many engineers with good incentives. So the speed of doing chips actually, for the staff, they are quicker than the existing giants. I knew that you mentioned that for the service providers like Ali, like Tencent or like Baidu, they are all trying to do their own chip, and those chips are in the most leading-edge technology nodes. But based on our knowledge that those chips are to [ lend ] a kind of experimental kind of stuff. So when they want to do the full-mask production version of the chips, I will say we have very high confidence. They, at least, will discuss with us. Since if you want to do the mass production version of chips, if you have -- if you don't have experience, it will delay your schedule. It -- there will be many obstacles or problems going to happen. So we don't worry this too much.

Yes, in China, I think other than this major service provider, so many startup company, high-profile startup company, they have design resource from the Tier 1 U.S. company and former team to do the design. We have at least 8 to 10 customer like this. So design experience, I don't -- front-end design the experience, I don't worry about too much. They also have enough for funding. They also have very fruitful design pipeline. We finished the 7-nanometer. Now they are doing 6 and 5 altogether. But in terms of production volume, I think that's a major challenge for us and also TSMC. Yes, we're working with TSMC very closely because the support effort is very high for both of us. Production expectation, I think, is very important. So we're working with the related supply, including the foundry, including the substrate house and interviewing those customers. I think, overall, we have a consensus of new customer may have cushion to hit a certain amount of production. I'm looking forward to see some production happen late this year or early next year.

Yes. This is a little bit important information to me. So what we're saying is that front-end design is not a big issue for the staff or even, but is that the right way to think about the agent places in raise of the key bottleneck for entire China's HPC?

Exactly. Exactly. Yes, also no, front-end design is not a process technology dependency. They are doing the -- they're just doing the algorithm and coding. They don't worry about the process technology too much. They have experience because they're all coming from the Tier 1 company like A&D, like Nvidia or others. Yes. But back-end is purely experience driven. You need to have a certain experience to understand the process technology for current technology before next -- before go to next generation. So you can see so many bright young engineer in front-end, but you can hardly to find a young engineer are very good for the back end. Back end is purely experience driven. You need to learn from the hard way algorithm. If you are good, if you are smart enough and you can pick up a good algorithm very, very, very fast. But back-end stuff is need to be built by experience. So that's why the start-up company, I think it's very difficult to find a good back-end engineer and back-end thing, but they can find some guru from the product company to form a front-end team.

Okay. Here is a question from message. How much NRE from N5 would account for 2022 NRE revenue? And how are the nodes going as well? I would say, first of all, Magpie, I didn't do the calculation. But for sure, the revenue from N5 this year could be much, much higher than we had in 2021, since we have multiple N5 projects either in design phase or we will tape out within this year. So roughly, I would say, if we assume like 5 N5 projects, roughly $40 million, $50 million, yes, that may be the amount from NRE. And for the production, we don't know because we are going to tape out 1 N5 full-mask project in the middle of this year. And we -- even there is production contribution, the contribution will be very small because of the said turnaround time. And for other technology nodes, for last year, the 7-nanometer -- the revenue from 7-nanometer already accounts for 60% of our total revenue. For this year, I believe this number will be even higher. Okay. Szeho, please.

Two quick follow-ups. In the second half year, when we ramp up the production business, what sort of gross margin should we be expecting?

You see the overall gross margin or...

Overall, yes.

Overall gross margin, because the NRE full front is also strong this year, even with expanded production revenue percent -- in terms of percentage of other revenue, we feel very positive that our overall blended gross margin could be above 30%.

Okay. All right, but it would not be materially different from -- let's say, from Q4 last year's level, right?

No. Because last year, the NRE percentage is higher. And for the production, as I mentioned in our slides for the production in the first half, the gross margin for those products were high.

Okay. Got you. And then for the R&D headcount or hiring this year, what are you internally projecting?

More than 100. We issued more than 100 of the letter already. And of course, there will be people come and people go. But we are still targeting to expand our total employee headcount to 600 at the end of this year.

600. Okay. From last year's, year-end, it's 480, right?

Actually close to 500.

500, okay. So basically adding 20%, yes. And then in terms of the operating expense guidance this year, I'm not sure if you can share with us on the call?

On that, we'll be double digit. For last year, the operating expense is about $62 million. For this year, I would say the operating expense range will be at about 73, 74. Okay. Now it's 3:30. We take one more questions, and we can end up this meeting. Is there any questions? [Operator Instructions] Charlie, please.

Yes. So just a quick follow-up. Do you expect any revenue contribution from FEITIAN first half or second half this year? I know you cannot talk too much, right? But just in terms of, do you still see some registering from FEITIAN?

We hope we can have revenue from FEITIAN this year. And in our forecast, there is revenue contribution from this customer.

Is it more from NRE or turnkey?

Charlie, I cannot discuss it in detail. Okay. And I guess that's it. Thank you for your participation, and thank you for your support to Alchip. Thank you.

All right. Thank you very much. Thank you. Thank you.

Bye.

Bye-bye.