Okay, it's 2:30 P.M. Ladies and gentlemen, analysts, portfolio managers, welcome to our second quarter institutional investor conference meeting. Prior to start the meeting, I will request you all to mute in order to keep from any noises. Today's meeting will include the second quarter operation review and, of course, financials. We will have the operation outlook for the second half of this year, and a little bit about the outlook in 2023. Okay, again, this is the safe harbor disclaimer. This meeting will be English and if you need Chinese presentation slides, please go to the MOPS to download the Chinese version.
When we do the Q&A session, you can raise your hand by using the Zoom Raise Hand button, or you can write down your questions through the Zoom's message function. We will repeat your question and then answer your question as well. The video and audio content of this meeting, we'll upload to MOPS. It may take about two to three hours for my colleagues to do so. Please be patient. You may download the video and the audio from the MOPS around like 5:00 A.M., 6:00 A.M. this morning. Thank you. In the beginning, I would like to invite our CEO, Johnny, to do the opening and very short introduction and his message to the investors.
Good afternoon, ladies and gentlemen. This is Johnny Shen, President and CEO of Alchip Technologies. Thank you for joining our investor conference meeting. We appreciate the opportunity to share our Q2 result and provide a guidance for future business. In case some of you are not very familiar with Alchip Technologies, allow me to provide a brief company update. Our company is founded in 2003, and IPO in Taiwan Stock Exchange in 2014. Our current headcount is approximately 600 people. Three-quarters of us are engineers. Since we founded the company, we've been successfully tape out more than 400 ASIC designs. Revenue number, 2020, we are achieving TWD 240 million. Last year, TWD 372 million. Current capacity, we can support 30+ designs in parallel. So far, majority of our revenue are coming from HPC and AI area.
We are one of the TSMC VCA members and can distribute latest technology from TSMC to our end customer. Quick recap for the first half numbers. We can say we have a fairly acceptable Q2 and first half revenue. Even though our revenue is kind of flat from year-over-year standpoint, the good news is we have a record-breaking number in all categories, including revenue, net income, and EPS. Revenue number is $195 million. Net income, TWD 30.5 million. EPS, TWD 13.3. Our HPC mass production demand remain very strong. The main cause of a relatively flat revenue is due to capacity limitations. As I mentioned before, our major product line are in a transition period from Flip Chip BGA type of design to 2.5D advanced packaging.
The longer assembly and final test turnaround time, plus the shortage of CoWoS interposer and ABF substrate, limits our near-term mass production revenue. For the past several months, Alchip work with partner very closely to improve the situation. For future production, on the wafer and interposer side, we received a reasonable commitment from our key partner. On the substrate side, we are seeing more stable support and commitment from multiple suppliers. Overall, we expect to have a substantial revenue growth from Q4 this year all the way to the middle of 2024. Alchip continuously holds great position in leading-edge technology business. We have received the most worldwide business inquiry this year. Almost all of them are in leading-edge technology, 7 nm or below. Many demand are 5 nm or even 4 nm.
With many design win this year, we expect healthy NRE revenue growth from Q4 through next year. Another noticeable highlight is our automotive application penetration. As you may know, many electric car makers start to develop their own ASIC. We have been approached by many car maker recently. Some of the design will be kick off very soon. We truly believe, starting from 2024, automotive application will be another key revenue driver for us in addition to current HPC and AI market. Lastly, I would like to emphasize our neutral position and diversified business condition. Similar to our key pure-play foundry partner, Alchip will never make our own product to compete with any customer. We never make any direct investment to any customer as well. We have a fully diversified and well-balanced business from all region, including North America, Asia-Pacific, Japan and Europe.
In terms of head count, we grow quite a bit recently. In addition to China, we also implemented a very aggressive hiring plan to increase our engineering and supporting resources in Taiwan, Japan and U.S. region in order to provide more efficient local support and meet customers' expectation. Thank you. Back to you, Daniel.
Okay, let me do the financial part. Actually, you guys may already have the revenue numbers, and we announced the bottom line, the net income numbers last week. Let me do a quick recap for the numbers. The total revenue for the second quarter is TWD 101.2 million, which is 8.2% quarter-on-quarter and 3.9% year-on-year.
The operating income for the second quarter is TWD 19.3 million, which is - 1.2% quarter-on-quarter decrease. The net income for the second quarter is TWD 14.5 million, translating into EPS is TWD 6. For the breakdown, actually, the overall application breakdown, you may see that Alchip is still concentrate into the HPC application. In the second quarter, the HPC related revenue accounted for 81% of our total revenue, and followed by 8% of niche market and 7% of networking, and 5% for the consumer products. In 2022, the first half this year, the distribution is pretty similar.
82% revenue from HPC related application and 7% from niche market, and 5% from networking, and 5% from the consumer electronics. For the process node breakdown, we can say that we are still very proud in terms of the process technology. Alchip should be the industry leader among the IC backend design service industry. In second quarter, 7 nm or even the advanced nodes like 5 nm, 6 nm, totally accounts for 69% of our total revenue. The 16 nm, 12 nm accounted for 18% of our total revenue, and followed by 28 nm revenue account for 10% of our total.
The rest is 40 nm and the more legacy nodes combined accounted for 3% of our total revenue. In the first half, 7 nm and higher nodes revenue account for 65% of our total revenue. 16 nm and 12 nm related revenue accounted for 23% of our total revenue. We can probably say more than close to 90% of our revenue this year are within the advanced technology nodes. As for the regional revenue breakdown, in order to more clearly demonstrate our geographic breakdown. Previously, we have Japan, China, U.S., and others to do the categorization.
From now on, we will use Japan, Asia-Pacific, which include China, Taiwan, and Korea as a category alone. For the second quarter, Japan accounts for 16% of our total revenue, while the Asia Pacific revenue from Asia Pacific region accounts for 39% of our total revenue. For U.S. market, you can notice that the revenue from U.S. market grow quite significantly and quickly. In the second quarter, revenue from U.S. accounted for 40% of our total revenue already. The others include the Middle East, Europe, accounts for 5% of our total revenue. For this year, first half U.S. market accounts for 35% of our total revenue, while Asia Pacific accounted for 41% of our total revenue.
We will say that our revenue breakdown regionally will become more and more diversified sites in the second half this year and going forward. For the second quarter business review, honestly, the second quarter revenue came as mildly lower than expected. Actually, we as the company management were expecting higher revenue for the second quarter. Due to the constraint of ABF substrate supply caused by the EUV issue. The second quarter we posted limited quarter-on-quarter growth. The design demand has no problem. The design demand remains very strong and as expected. We don't see any weakening demand on site even though the macro environment outlook is worsening. The HPC industry, the demand for HPC industry remain very strong.
For the profit margin, in the second quarter, the gross margin went down to 34.5% from the first quarter, 38-something%. That's because, in the second quarter we have relatively fewer design revenue exposure. The second quarter, the design revenue accounted for 40%-50% of the total revenue, comparing with the first quarter. In the first quarter, design revenue accounted for close to 60% of our total revenue. We will soon upload and announce the first half financial statement very soon. Actually, you can have it later.
We want to address that the inventory level on our book went up because of the piling of top die for the second half in the 2023 shipment to the U.S. customers. I guess people who follow us should know that these customers keep on placing the top die orders to us, and we place it to TSMC. The inventory level in second quarter end rose to $100 million from about $38 million by the first quarter end. The North American customer keep on placing orders for the top die. First purpose is to do the preparation for expecting surging shipment in late fourth quarter in the 2023.
The second reason is we have lower than expected substrate supply in the second quarter and cannot pair up with the top die to finish the goods. That's the two reasons for the inventory number went up by the end of the second quarter. For the second half 2022 business outlook, first, the first point is we expect the impact from substrate shortage to persist throughout the third quarter 2022. Which means the third quarter revenue outlook should still be influenced by the substrate shortage.
We believe, according to our schedule and how working with the substrate supplier, we expect the ABF substrate supply shortage to gradually ease starting from the fourth quarter 2022. More precisely, we expect the revenue to trending up mildly in October and then relevant month-on-month growth in November and December. Our current solution for the substrate supply is first of all, we will add new substrate supplier starting from late third quarter. But since the new supply needs take a long time, so that's why we said the impact on substrate shortage will continue throughout the second quarter.
The improving industry-wide substrate supply, I guess maybe all you're aware that, given the weak consumer electronics devices demand, we do feel some alleviation from the substrate capacity allocation from the current top substrate suppliers. It will help our future supply, but not immediately. The high-layered, which means 20-layered ABF substrate supply industry-wide, we think is still tight, but we work very hard to solve this issue. Last thing is overall design demand remains strong. The HPC related design demand remain very strong in the second half, for both China and the North American market. As Johnny, our CEO, Johnny mentioned, we are engaging in several project discussion with very sound customers.
Hopefully we can nail down to win those projects in the second half this year and the first half next year. The industry trend continues, especially for the HPC application. We expect our customers, our projects, the process node technology will keep on migrating from 7 nm to 6 nm or 5 nm. This trend, we expect to continue at least for two years going forward. That's our view for the overall design demand. I guess that's the information we are sharing with the investors and the fund managers. Next session, we will go into the Q&A session. If you want to ask questions, please use the raise hand function of Zoom.
When we call your name, you can unmute your Zoom and ask questions. Thank you. Dennis, please. Dennis Chang, please.
Yes. Yeah.
Oh. Okay. Hereby I would like to announce a thing that, by discussion with the suppliers and our customers, we believe from now on we are not going to talk about any single customer's detail and project detail and the scheduling to the outsiders to protect our customer and our supplier. From now on, for the China business, we will try our best to answer your question, but when questions involve a single customer or any particular customer or any particular project, we will try our best, but most likely we will not answer the details for those questions. Thank you.
第二个问题是有关ABF的问题，ABF今年很紧，很多大家都会知道，但是其实最近一些法说会也都说明它已经有一点比较松了，那这个，是不是会比较好？然后是客户是不是会影响未来的案子，比如说今年你们因为ABF它有些紧没，那，那人家会不会影响客户对你们的信任度？
关于这个问题，就是刚刚法说会的slide也已经有提到了，就是说的确我们有感觉到，就是整个industry的这个ABF substrate的supply，它有一点比较情况比较好，因为这个consumer包括PC、包括handset，目前大家预期在second half，可能整个市场的情况都不会太好。但是呢，我们必须要讲说我们的主要的产品用的是二十层的ABF，通常CPU、GPU的这类，或者是application process这类的产品用的这个ABF的层数在十二层到十六层。那相对于良率来讲，这个ABF的substrate的良率一过十五层之后它会掉得很厉害。所以说在二十层的ABF这边的supply，整个industry我们感觉还是紧的，但是情况有没有比六个月前好？绝对有。所以说在这个部分呢，刚刚也已经有讲了，我们在很积极地跟我们的supplier沟通。第一个希望他们能够在第四季能够持续地在把能够给我们的这个量上面再往上加。那第二个就是说在明年2023年，在他们目前我们能够掌握的commitment的allocation上面，能够有更多、更多的commitment，这个substrate的基板的供应商，这个是我们目前在做的事情。
最后就是客户的黏着度的问题，就是现在客户找S公司开样板，开完以后呢，就会直接后面有量的时候，可能会拿回去自己投片，或者直接这个NRE直接转成turnkey的这个比例大概是怎么样？就好像我刚才提到你的某不大客户，那后来其实也有一个替代方案，就是去SMIC投片，那如果拿到一些许可的话，是不是还会继续透过你们，还是会直接自己去投片也好，那如果说客户每个都自己去投片的话，那我是不是就变成一个做苦工，赚辛苦钱很辛苦。那现在的景气不太好，所以很多那个design house都扩大ASIC业务，是不是竞争有越来越激烈？另外中国其实那边人力比较充沛，也都有成立很多design service公司，甚至于美国的大厂商，像是Synopsys、Cadence也都在做这些业务，那这样子的话，这个竞争态势对我们会不会比较压力会比较大一点？
第一个我必须要讲，就是你一开始的描述是不对的，就是作为做量大的客户把它拿回去做，这个 reality 不是这个样子的。当在谈 design 一个 turnkey 一个案子的时候，其实一开始它是一个 partial turnkey 还是一个 full turnkey，它就已经确定，如果一开始签的合约是 full turnkey 的话，其实客户并没有办法说做到一半，我把这个 production 拿回去做，这个在合约里面都是蛮详细蛮清楚的规定，所以说这个事情不会发生。另外您讲到就是说，是不是会越来越多客户做 project 的时候是 production 拿回去自己做。我们自己觉得说这个全是在于你能够 provide 多少 value 去给我们的客户。那这个代线 design industry 已经存在很久的时间了，那经过一番竞争下来之后，在尤其特别在 advance technology node，我相信这并不是已经仅仅是价钱的问题，问题就是说在于客户 go direct，他要付出什么，他会得到什么好处，客户来找我们做 turnkey，他能够拿到什么好处，他愿不愿意付出一些利润来达到这些好处，我觉得这个是会比较在客户方决定他是要自己做，或者是拿出来交给我们做 full turnkey service 的一个关键，这个并不是仅仅是价格来决定的。
那也越来越竞争了，像这些什么MediaTek什么全部都来做ASIC业务这样，压力会...
他不是全部都来做ASIC。MediaTek、Marvell跟Broadcom，也就是Avago，他们其实做ASIC的business已经做了非常的久了，那他们在竞争的市场呢，跟我们跟GUC所在的市场有一点点不一样，他们在compete的比较是所谓在业界叫stacking的市场，就是包括前端、后端及turnkey，他们全部从头包到尾，包括里面的IP。那在这个领域呢，其实基本上我们目前啊是比较缺乏这个能力去碰的，因为我们在前端的能力并不如这些所谓的stacking的厂商，但是对于这些厂商来讲，它的成本结构跟我们的成本结构不一样。所以说对于会去找这个所谓stacking的厂商的project，通常第一个它要量大，量太小的project对于这些厂商来讲，他不会接。第二个，他的NRE的金额是非常非常高，跟做所谓的这个backend的turnkey service的这个NRE的single project的量其实是不太comparable。
Okay. Let me try to add more input for Dennis question. Yeah, first of all, the China customer. Yeah, just like Daniel mentioned before. Yeah, due to the variety of reason, yeah, we've been instructed not talking about any individual customer in China. I think that's a good practice to protect and customer as a major supplier. But it doesn't mean our China business is not doing well. You know, in fact, our China business, I think Asia Pacific business are doing very well. Yeah, so maybe people will expect next year we have a substantial growth in U.S., only in U.S., but it's not true. I think our expectation from China and Asia Pacific customer sell a lot. We will see.
You will see the number eventually during our quarterly report. During our investor conference meeting, you will find out the revenue from AP still very, very strong. Second question for Dennis is actually third question. Actually, this design industry has been in existence for a long time, but to be honest, not too many competitors are eligible to do the leading-edge technology. Yeah, as long as technology keeps improving, I think our business opportunity is getting more and more. As you know, NRE for leading-edge technology is getting higher and higher, very high. You need to spend like TWD 20 million-TWD 30 million to do a 5 nm design. Yeah, for 3 nm, 1 nm, I think the NRE will be much higher. Entry barrier getting higher and higher.
The reason we get more and more business is because of track record. As I mentioned before, we've been enable more than 30 design a year. Even customer decide to do it by themselves. In usual case, each product company only have a one tape out a year. The knowledge accumulation and risk management, I think, is no match to us. Your second question is about MediaTek, Broadcom, and maybe Marvell. Those are major supplier. In terms of size, they are much, much bigger than us, but their main focus is still their own product. ASIC is just their side business. To be honest, the gross margin for ASIC from ASIC are much, much lower than the product itself. For Marvell, for MediaTek, their average margin is more than 50%, sometimes even more.
For ASIC industry, it's very difficult to get more over 40%. Overtaking ASIC business will be no good for them. I think that will bring down the entire overall gross margin. That will not be their major focus. Their major focus is to sell their own product line.
Okay. I hope we answered your question correctly. Si Ho, please.
Oh, hello, Johnny, Daniel. Thanks for your time.
Hello.
Two questions from my side. Yeah, yeah. Hello. Yeah. For N5 NRE, we should expect to kickstart next year. How about N3 NRE? Any timeline that you can share with us?
Johnny.
Yeah, N3, we haven't have a actual design yet. We only have a few test chip. In usual case, the beginning of each generation, you can consider the NRE is almost double at very beginning. I think that's a similar situation generation over generations. Yeah, based on our calculation, we will tape out our first 3 nm in 2024. Near that time, I think we have a more clear picture how much it is.
Oh, okay. Okay. All right. Other question, I think it's a little bit early at this stage, but I still want to ask. For the N2, right? NRE. Should we expect a jump in R&D? Yeah, because the architecture will be very different from the previous nodes.
Yeah. I think, Si Ho, this question, I think it's too early to ask.
Okay.
Right now, I think the major preparation from us is 3 nm plus the package solution. Yeah, as you know, the UCIe related 3D package, I think that will be the another important factor for 3 nm. Yeah, so, 2 nm, we haven't. Yeah. Even our foundry partner, yeah, haven't shared any detailed information to us yet. We do have s ome preparation for 3 nm.
I see. Fair enough. Yeah. Then, at the beginning, of the presentation, you mentioned the companies are breaking into the automotive market, right? I just wonder, what would be the chips we will be serving at the beginning? Is it the core chip or the companion chips?
Yeah, for automotive application, I think people approaching us for two different area. First of all, is the ADAS related autonomous driving chip. That one, I think, will use the most leading-edge technology. Actually, the industry also changed the term. We call automotive HPC. Yeah. The architecture is quite similar to the existing HPC, but the support effort will be much higher in terms of quality, in terms of all the logistics. Yeah. Another automotive demand is from data center. As you know, in order to provide like more sufficient driving instruction, each of a car maker have their own data center. Yeah. Like Tesla, like Rivian, they have. I think we have a opportunity for both area.
All right. It's all compute related chips, right?
Yes.
I see. Okay, great. Last one. For this year's revenue guidance, any update? Are we still sticking to the TWD 500 million revenue target?
Daniel?
Previously we did guide a TWD 500 million revenue target for this year. On that time, we are very, very confident that because I put in many buffer into this target. For now, to be honest, because of the substrate supply shortage in the second quarter, third quarter, I would say for the TWD 500 million guidance, I'm not giving up this target now. If we can achieve the target or not depends on, first of all, the substrate supply in the fourth quarter. Secondly, the CoWoS turnaround time. We are trying very hard to pull in the schedule and the output for substrate, and we are also trying to work with TSMC to shorten the CoWoS turnaround time from currently six weeks to four weeks.
If everything goes very well, I think we still have chance to achieve TWD 500 million. Even if we could not achieve it, I believe we will be not far from these numbers.
Okay. All right. Thanks for your color. Yeah.
All right. Thank you, Si Ho.
Yeah.
Charlie, please. Charlie, you can unmute.
Oh, yes. Thanks. Hi. Hi, Johnny. Hi. Hi, Daniel.
Hi, Charlie.
My first question is about your automotive HPC opportunity. I'm not sure, you know, whether the turnkey, you know, lead time will be much longer, right? Because the auto chip usually needs a very long time to do the qualification. Is the company has this related experience for the auto chip?
Yes. Charlie, you are right. The turnaround time and also the qualification procedure are much, much more complicated than the regular HPC. That's why I say the starting from 2024, we will see more significant revenue contribution during that time. For the design we are about to kick off, the tape-out will be in the middle of next year. Yeah, 2023. So the real volume ramping up will be a year later. Yeah, your calculation is right, but I think it's the right time for us to do all the preparation in parallel with the design.
Okay. Yeah. I'm wondering why Alchip? I mean for you know your design services competitor for example Broadcom probably they have a related experience. If you look at the it's public information right? Tesla they use the, Samsung for 7 nm and also a 5nm ASIC service. What is Alchip's advantage to compete with those design service peers for the automotive market?
Okay. First of all, like I mentioned before, because of our neutral position, we never make any product to compete with any customer. Broadcom, Samsung, Avago, I think, they make their own product. Sometimes they compete with each other. If you want to do the ASIC, go through those partners, I think they may have a concern. Eventually, this kind of information will leak to their own product or other customers. Within Alchip, I think we don't have this kind of concern. Customer won't have this kind of concern. Also we are much smaller and flexible. The supporting ASIC will not be their main focus. Going through Alchip, I think that for sure they are our tier 1 customer.
Yeah, our support effort, I think, will be much better than them. I think that's from the existing customers are saying.
Okay.
Also, our yeah profit margin as an expectation is lower than them as well.
Okay. Charlie, to me, I would say the feeling right now for the automotive application is pretty. It's a little bit like the beginning of the HPC trend. Those car makers usually don't know too much about making chips, but in the same time, they want to develop their own technology, similar with those service providers previously. I believe in the same time, they will build up their front-end design capability by hiring or acquiring company from the market. They have various needs and trying to do various ASIC to enable their service or the functionality of their products. I guess to gain control on technology is a reason for them to do the ASIC by their own.
At the same time, in order to quickly come to the market to work with the back-end service provider will be definitely a consideration for them. Within this kind of solution, I would say Alchip has pretty good industry position to win projects. That's my feeling right now.
Okay. Thanks, Daniel, for your comments as well. Yeah, I'm surprised that for those new EV companies, right, aside from the big company, they wanted to own the front-end design team. That really surprised me. Do you think, going for the design services, you know, the industry will becomes a more back-end or place and route oriented instead of, you know, running as a stacking business model? Do you think this is going to be the future trend?
Yeah, I think it is. I think, okay, for everybody's experience, including you and me, that, we all believe that, in the future, the technology feature of a car is one of the major competitive, things for them to win customers, and especially for those EV makers. Because for the EV market, even though now Tesla is the biggest, but there are a lot of new competitors. For the new competitors, they knew that they have to be at least the technology leading or to provide, new technology features, to their customers or to their potential, customers. I think based on our, feeling or discussion, almost, every carmaker, EV carmaker are very aggressively to do similar things, which is basic.
Okay, I see. Next is about your China business, right? You know, we see some news about the funding issue, you know, of some of your Chinese customers. Do you think that is going to impact your business next year? On the other hand, it sounds like the automotive HPC could be also a big opportunity for China market as well, right? Because China EV industry is booming. Do you see any opportunity from China for automotive HPC as well?
Okay. In general, I would say yes. The funding environment in China right now is not as good as like six months or eight months ago. To us, I think the good thing for us is the startup customers we secured usually is the number one, number two within their market. Funding for them is more difficult, but not unachievable. Based on our experience by doing business with those startups, when the funding situation is getting more difficult, usually it affects the project scheduling, which means the schedule of project design will become longer. It only happens in a very few number of our customers in China.
The majority of our customers in China are whose positions are pretty good and their funding situation is relatively much better than their smaller competitors. So far, we don't worry too much because overall demand is still strong. Even though one or two customers, maybe, fail to follow the original schedule, we can quickly take other projects from the waiting list to fully utilize our design capacity. I think that's the situation.
Yeah. Let me try to add to that. Yes, Daniel is right. In China, actually, we have so many pipeline for us to pick. Yeah. In fact, I think China customer is very difficult to get our resource. We have a plenty room to choose our own customer. Yeah, in terms of customer selection, we will pick the one with good funding, and also the team member has a better experience, a good credential. I think that's for sure. Also, in terms of financial, everything is a prepaid. Yeah. We don't have a, we are not suffer any financial risk, yeah, for start-up company. Charlie, you are right.
I think we expect, since the funding situation getting tight, certain consolidation will start, maybe starting from early next year. Yeah. I know a lot of company already start to discuss on that topic.
Yeah, that's very insightful. Yeah, so lastly, maybe I ask this one question from some investors. So you also have some exposure to Japan, right? You are also extending on the size there. Now do you ever hear this company called Socionext? I think it's a joint venture of Fujitsu and Panasonic. When I take a look of the website, it seems to have some exposure to automotive and also the consumer TV, et cetera. Would that be kind of your competition in Japan?
Yeah. Let me ask, yes, we're familiar with Socionext very much. Yeah. To be honest, they are doing quite well, especially on the automotive area. Yeah, they are not our competitor. They are not only the competitor in Japan, also in other regions, yeah, including U.S. and China. I think this company has been for a long time. That's not a new company. Overall, their company is not profitable. They are suffer quite a bit. And they do have their advantage because Japanese company, in terms of quality, they have a much higher reputation. So that's why they have a good penetration of automotive area. For other area, I think we are doing much better than them.
And I think in the automotive area, Socionext will be our on e of the head-to-head competitors, c ompared to existing competitors. This market is very big. Like I mentioned before, all the electric car makers decide to make their own ASIC. Even we try to take as much as we can. I think in a short period, we can only take about three to four maximum. There are more than 10 opportunities out there.
Do they use TSMC as well?
They use both. For both, f or the leading edge technology, they do work with TSMC and through our placement order from TSMC Japan. They do use for the mainstream stuff, and they use a Japanese IDM supplier as well.
I see. Okay. That is super helpful. Okay. Thanks, Johnny and Daniel.
You're welcome.
Thank you.
Okay. Calvin. Calvin Wong.
Hi, Daniel and Johnny. Thanks very much for doing this call. I have two questions on geopolitical tensions. My first would be, I think there was news on the U.S. banning China from using Gate-All-Around EDA tools. I understand that's probably not a near-term issue, but maybe say beyond 2025, for example. Do you see risk towards a lot of your Chinese customers, you know, them not being able to do the front-end design, so there might not be back-end design work for you? I'm wondering what you all think about that. Yeah.
Okay. Johnny, you want to answer his question?
Sure. Yes, Calvin, just like you mentioned, geopolitical tension will be always there. Yeah. So far to ban the GAA, and also, if I remember right, MEMS sensor will have no direct impact to us. GAA for Samsung, they may adopt in 3 nm, but for TSMC, that will be 2 nm. Based on current calculation, that will be beyond 2025. Right? I think about that as just a warning, yeah, from the BIS. I don't worry about GAA at this moment, but we do anticipate the future more challenging from BIS. I don't want to leave out, but I think to anticipate and also prepare the geopolitical tension, I think that will be one of the topic everybody need to pay attention. So far, no GAA related has no impact to us.
Sure. It's not a short-term risk, but there's the risk longer-term. Is that how you're thinking beyond 2025?
Right. I think those are GAA related. EDA is mainly for foundry to use. As you know, the China foundry, their technology are still sitting at a 14 nm. They try to make a 7 nm. I think it take a much longer time for them to go into the GAA technology for design, and we don't so far, either U.S. company or Chinese companies that design EDA tool do not cover the GAA. We are not into that stage yet. Even the tier 1 EDA maker, the design kit is not to that level yet.
I see. Thanks. My second question would be on whether or not you're seeing, especially your U.S. customers or non-China customers, are they increasingly asking to move project designs done in China outside of China? Can I also get a sense of how much of your design headcount is in China versus outside China?
Let me try to answer that. I think so far we have 30 customers. Only one customer has this kind of requirement. They are not allowed PRC citizens to touch their database. Only one out of 30. But we do have a few customer requests, database not sitting in China. On the previous conference call, we say we already have a solution two years ago. Right now, our database are sitting in Japan, Taiwan, and China. Not moving the database to China, I think that one is for us very easy to do. In terms of PRC citizens, so far we don't have, we don't see so many requirements, but we do some preparation.
Like I mentioned earlier in the meeting, we're doing more aggressive hiring in Taiwan and Japan. These two region has been stay with us for more than 15 years, so it's no problem for us to increase the design resource in Japan and Taiwan, which we already do it.
Okay. To answer your question, in terms of the headcount, the majority of the designer is still located in China region. We have many subsidiaries in China and of course, LT is born with the design center in China. It will not change. For Japan and Taiwan, in Japan we are aggressively doing the hiring. We have been aggressively doing the hiring and our Japan design engineers, who usually are very senior designers, the headcount almost doubled this year. For Taiwan, we already have a team who can do the project independently. We are trying to expand one team to two teams in the near future. That's the current headcount situation for the design team.
Sure. Thanks. Just one follow-up. Of the one customer that had a requirement not to have PRC citizens touch the database, that customer, is that a big customer to you all?
No, it's not. It's relatively small.
Okay.
Yeah.
Understand. Great. Thank you very much, Johnny and Daniel.
Thank you. Thank you.
Okay. Dennis asking questions through message. The question is, if a customer with a very successful first project, the first generation project, the second generation project, they will go. They will do it by themselves and become the direct account of TSMC. That's the first question. The second question is, Synopsys indicates expand their design service, human resources a lot. MediaTek is also adding people in ASIC. Do we feel pressure from those competitors? That's the two questions asked by Dennis.
For the first question, the direct account thing, I knew there are some sayings or rumors on the market. I will say this way. A direct account of TSMC doesn't mean they are not doing business with us. As you may know that, Alchip originally did a lot of business with Sony. At that time, Sony is the direct account of TSMC. So they were still doing business with us. So like I said, to do business with us or not, it is not related to if our customer is the direct account or not direct account of TSMC. In order to do the back-end design by themselves, they have to build up certain level of design capability, back-end.
Otherwise, it is not that simple that if I'm on a direct account, I can ask TSMC, "Do all the back-end thing for us." That is not true. The second thing is the Synopsys and Cadence question.
Yeah. Let me try to answer the Synopsys and Cadence design service. Yeah, first of all, Synopsys and Cadence would not, would never focus on design service for sure. As you know, their current business model, IP and EDA, gross margin is 80%-90%. They build out a design service team is only to enable their EDA tool and IP cell. So I don't think they are they are focused on the design service area since the gross margin are much, much lower and turnaround time are much longer as well. Synopsys and Cadence would never do turnkey.
They can only do design service. They won't touch the production as well, because the production margin are even lower. They never make any production business. Recently, we've been working with these two company very closely. I think they instead of they compete with us, I think they are willing to work with us to enable all the business.
Okay. Actually, the same thing, the similar thing for MediaTek. Actually, MediaTek is our customer. In terms of in contrast of competition, I think we work as a partner to work with TSMC. For example, for TSMC. Well, not TSMC. For MediaTek, projects production volume may be lower than 100 million per year. It is not a good mix business for them, but it is a good business for us.
When the industry-wide design capacity is not enough to fulfill the whole market demand, we think cooperation or work together is a better it is the current situation and a good approach for us to do the business. Okay. I think I hope we answered your question well. Brian, please, you can unmute.
All right. Thank you. My question here is on your NRE business in the second half. Can you just provide us a bit of color on how what we could see there in terms of scale vis-à-vis what we saw in the first half?
Excuse me. Your voice is a little bit noisy. Can you repeat your question?
Right. My question was on the NRE business in the second half. Can you provide us some color on the sort of scale, the sort of new projects that we are going to be working on in the second half in comparison to what we booked in the first half?
I would say for the third quarter, I expect similar revenue mix to second quarter. With similar revenue mix, I would say the gross margin will be similar. For the project, I think we currently, like Johnny mentioned, we can do 30+ projects in parallel. We are pretty diversified but concentrated in the HPC area. There are a lot of high-performance computing like CPU, GPU, AI inference chip, training chip, DPU, something like that. That's what we are doing right now. Also, like I mentioned before, we enabled so many designs early this year in the first half. Most of the designs will be taped out on Q4 and sometimes next year.
NRE growth, I think is highly expected. I think the mass production increasing speed will be faster than NRE. Mass production revenue can be, we expect exponentially increase. The NRE is, we can say it's only linear type of momentum. Okay. Brian, are we answering your question?
Yes. Got it. My follow-up here is on the pickup in the ABF substrate supply in November, December. It does sound like if you are still hopeful you get to TWD 500 million, there is going to be a very, very substantial increase in ABF supply when it comes to book revenues in an increment of about TWD 30 million-TWD 40 million, if I'm just on a ballpark. Is that the sort of ballpark we're talking about to get there? Does that also feed through to what we think we can be booking incremental best production in 2023?
Yes. Let me answer your question. First of all, I would say the revenue, the substrate output that can be translated into revenue, which means anything after November. Not anything, any substrate output after November will not translate into revenue this year. The fact is, when we guide that our monthly revenue likely to gradually pick up starting from October, which means the substrate output increase happens in October and September.
Otherwise, we will not have them translate transforming to revenue in November and December. I think it is a trend, but how fast we can pull in the schedule or the, like I said, the CoWoS turn around time. That's why I said we try very hard. If everything follow the expectation in good form. We may have the chance to achieve 500. If not, we will not be far from it. That's the situation right now.
If we're doing a little bit easy backwards calculation, our Q3 revenue increase is not so significant. That means, our Q4 incremental will be a lot in order to meet the TWD 500 million target. Q4 revenue growth, I think, will be very big. But that's just a beginning. I think, the momentum will last all the way to 2024. You can expect, we can expect 2023 will be another great year for us in terms of revenue.
Okay. For PGIM WD, ask questions through message. This question is pretty similar to the questions Brian just asked. Could you give us the third quarter and the fourth quarter seasonality and the gross margin trend? If your ABF capacity turn to fluid situation, how much would Alchip grow in first half 2023 year- on- year compared to 2022 target? Okay, to answer your question for the third and fourth quarter, I think the relevant quarter-on-quarter growth will happen in fourth quarter if the ABF substrate supply will meet our expectation. Another critical point is, we have to have pretty substrate supply improvement to fulfill the significant demand from our customer next year.
By comparing the demand this year and the next year from the North American customer, the demand is more than double or triple. ABF is really the key. That's why we mentioned, even put on the slide that first of all, we diversify the supplier. Until now, we only have one supplier, which is Sony. In the future, we will have three suppliers in fourth quarter and 2023 whole year. I haven't done the number forecast for 2023, so I cannot give you a year-on-year guidance for the first half 2023. As long as the substrate issue, the substrate shortage easing, we believe the revenue growth momentum will be very significant. Because first of all, the forecast from our customer is really high.
Secondly, when you look at our inventory level, in the end of second quarter, low top die is all for the preparation for the future shipment in fourth quarter and 2023. So I think you can consider the inventory pile up as a indicator to the future revenue growth. The second question is, 2022 U.S. revenue contribution grow a lot. Could you give us the U.S. contribution breakdown in terms of meaningful customer or ongoing product? Sorry, I cannot give you a detailed breakdown. I believe no company will give you a very detailed breakdown for each customer. I can just say the majority contributor, in the second quarter for the mass production revenue, is the service provider customer in U.S.
Casey, Brian, ask a quick question. Why is the NRE only single-digit % on the financial report versus the 40%-50% that company planned? What is the difference? Thanks. I think for this question, we already answered many times. When IPO, the categorization required by the authority, which is TWSE, they asked us to do the categorization, similar, follow our peers, which is GUC. So the NRE on our financial book, which means is the project only has design revenue without potential production revenue. When a complete turnkey project, which means NRE plus production, it will go to ASIC. So to us, of course, we think it is not reasonable, but on that time, we have to follow the instruction by the authority.
That's the reason why in every earnings call we tell investors the real NRE percentage and the production percentage. Because to us, NRE is totally different, has totally different cost structure from production revenue. That's the reason. I think it answer your question well. Zhongguoxingtuo Chougu ask questions about can you share Alchip's other customer for the data center AI chip? Johnny, you want to share?
Yeah, we do have a few other accounts, but it's not appropriate to specify their names. Yeah, we know some Israeli company acquired by a U.S. company. This company does give us a certain order. A few companies in China and a few companies in Japan, yeah, also give us some AI chip-related orders. Yeah, in addition to the North American accounts.
Hope it answer your question well. Is there any questions? Please use the Zoom raise hand function, that we can answer your question. I guess no. It is about the time to close today's meeting. Thank you for your participation in our second quarter earnings call meeting. Thanks for your support and have a good weekend. Thank you.
All right. Thank you very much. Thank you.