There is money for 7 nm, 5 nm, and also package related R&D investment. We are one of the TSMC VCA member, value chain aggregator. We are only a few VCA can do the service worldwide for TSMC. Next page, please. Alchip is doing the ASIC business since we found the company. ASIC market is very big and unique. I don't think I need to explain into detail anymore. Currently, company is focused HPC and AI area. We consider the entry barrier for HPC and AI are higher than most of our applications. Since the design are huge and complicated. Technology. Currently, HPC and AI using 16 nm or 7 nm solution. Eventually, they will use a 5 nm or 3 nm, or even below.
For the first half of this year, HPC and AI occupy more than 70% of our total revenue. We truly believe HPC and AI application will continuously grow and sustain, even with very reasonable margins. Next page. Okay. Let me give you a very quick recap for the first half. First half, I think we have a record-breaking first half in all numbers. Revenue, net income, and EPS. Compared to last year, we have a significant growth. Revenue is $105 million, 65% YoY. Both design and production pose a very strong momentum. We tape out 7 nm new design last quarter. The HPC demand from China customer is getting stronger. We did receive more order. Every week we receive additional order from China customer. First half gross margins, 34.1%.
I think the revenue, the gross margin still maintain to a very reasonable number, even the production revenue grow out quite a bit. For the Q2, our revenue is record high, quarterly record high, $55 million, yeah, up to 8.8% QoQ. Company doing the very detailed, the revenue recognition, and we have a record-breaking YoY QoQ revenue. Even the MoM, each of months, we're showing the steady growth. Operating income reach $8.7 million. Yeah, compared to last year, increased quite a bit. The revenue weight between NRE and mass production, I think pretty much half and half, similar to the first half. Yeah, another drawback is to be conservative, we've been write off $400,000 from one of our investment. I think later on, Daniel can explain into detail. That's a quick recap for the first half.
The rest of the section, detailed number and tape out, I will let Daniel to explain, and later on, I will come back for the future outlook.
Okay, for this page, this is the second quarter result and the first half numbers. For the second quarter, the total revenue, as you already know, the total revenue is a single quarter record high. The revenue is $54.9 million in the second quarter. The gross margin, as Johnny mentioned, compared to the second quarter last year, the gross margin is relatively lower. That's because of the product mix. For last year, the second quarter's NRE percentage accounts for more than close to 60% of our total revenue. For this year, the second quarter, the NRE revenue accounts for about 45%-50% of our total revenue. As a result, the operating income for the second quarter alone is $8.7 million. The net income is $6.4 million.
By breaking down into detail, for the non-operating income, you may see that our non-operating income for this quarter is -$180,000. That's because we write off one of our bond portfolio, the Virgin Airline. The total position is about $400,000. According to the news and according to the private banker, Virgin Australia now is taken over by Bain Capital. The future outlook now looks better right now. If the bond price goes up or if there is any debt restructuring plan, the write-off amount will reverse to our non-operating income in the future. For the first half, the revenue is $105 million, up 65% year-on-year, and the operating income is $15.7 million, up 165% year-on-year. In terms of the EPS, the first half EPS is TWD 6.01.
For the revenue breakdown, you can see that the HPC and the networking, actually, most of the networking are still within the HPC field. Still account for the majority part of our total revenue. In the second quarter, HPC accounts for 66% of our total revenue, and the networking accounts for 14%. For the niche market, our product is mainly the P achinko machine. In second quarter, it accounts for 14% of our total revenue. In first half of this year, HPC total accounts for 72% of our total revenue, and the networking is 9%, and the niche market is 13%. The consumer has been keep on decreasing to 6% for the first half this year. For the technology breakdown, you may see it's getting more and more obvious that the 7 nm technology now has already become our mainstream technology.
For the second quarter, due to some DMP, design revenue recognition, of course, there are some 7 nm tape outs. The 7 nm revenue accounts for 38% of our total revenue in the second quarter. If you combine the 7 nm, the 15 nm, and the 28 nm, these three technology nodes combined account for more than 90% of our total revenue for both second quarter and for the first half this year. For the regional breakdown, in second quarter, you see the Japan from 14% to 21% because we realized some DMP milestone and, of course, the Pachinko shipment keeps on doing very well. For China, accounts for 51%, mainly contribute by the HPC products. For the others, you may see the others keep on growing. I would say North America is the major region for this item to grow.
In the first half, we think our regional mix is quite reasonable. The China account for 56%, and Japan account for 18%, and Taiwan accounts for 5%, while the other region accounts for total 21% of our revenue. For industry and the company outlook. Again, I guess many of you have already read it from the report, or you already have conversation with me that the China HPC IC demand is still strong. The China IC localization trend keeps on going, with increasing demand from the HPC customers. Our biggest and the second-largest customer, we see picking up momentum from their demand. Therefore, we expect the HPC-related IC projects, coupled with the 7 nm shipment to start in the fourth quarter, will drive our second quarter revenue.
All in all, for this year, HPC, for the processor project, they are the major growth driver for the whole year. In addition to existing customers, we see many opportunities from new customers. As Johnny mentioned, we almost receive RFQ every week. We are now in close discussion with our potential customers to pick up the projects, and, of course, we have our priorities for those projects, and we don't see the demand to be weak in the near future. For AI demand in North America, it still remains good. Given the coronavirus pandemic, the business development pace is kind of slow. We are trying very hard to get more business opportunities and to speed up the pace of our business.
The good news is we expect our first 7-nm project from new customers in North America region to tape out in the second half this year. As confirmed, we won a 5-nm design for one big North American customer. That's the quick briefing about our second quarter. Now, I would like to enter into the second set Q&A session. Please use the raise hand function and I can unmute you, or you can unmute yourself. Please, because it's an online conference meeting, please use the raise hand. Okay, Szeho, please.
Oh, hi gentlemen. Two questions from my side. For the first one, regarding your overseas ASIC customers, I think a very strong NRE momentum. Could you share with us the volume prospect for those customers?
Okay. For the volume, I would say, for the quarterly guidance, I told that we will have slightly to mild quarter-on-quarter growth for the first quarter to third quarter. This guidance remains unchanged. Our third quarter total revenue will still post reasonable growth, mass production demand is quite strong, especially for the server. I would say the momentum in second quarter is still maintained. For the fourth quarter, we may see a relatively significant quarter-on-quarter growth in the fourth quarter, because our second largest customer will start its shipment, the 7 nm processor shipment, in the fourth quarter. The momentum is good.
Right, okay. Are you going to raise your full-year guidance? Because it looks very conservative, right, based on the first half momentum, and then we just extrapolate and could easily achieve your full-year target.
Yeah.
Okay.
Yeah. Actually, we would like to, but they still have some dependency. Actually, the most challenging is 7 nm design capacity. As soon as we can continuously receive reasonable capacity from TSMC, we love to increase guidance drastically. Right now, I think that's the biggest uncertainty and drawback for us. Luckily, even with the 16 nm demand, we still met our guidance, the number of guidance, yeah. Recently, TSMC gradually give us some capacity on 7 nm. I think that's a good news. Once we confirm receive, I think I will work with Daniel, maybe we can increase the guidance accordingly.
Yeah. Of course, I understand that $ 200 million yearly revenue now it seems very conservative.
Oh.
According to our guidance, we are very confident that we.
Overachieve.
Yeah, we will overachieve the $ 200 million. How high we can achieve? I would say 10% is a reasonable guess, but it has dependency. We work with Johnny to set up a new guidance for investors.
Okay, great. Yeah, and for the 5 nm design win, could you give us more color? What type of products are you working on, and when we should see the NRE kicking start?
Okay, Johnny.
Okay. 5 nm, as you know, whoever doing the 5 nm are very big company. Yeah, we are under strictly NDA. Do not disclose their project spec and also schedule and those kind of things. I apologize for that. For this North America 5 nm design won, we have to keep for ourselves for a while. Just share another good news with you. I think we are about to win another 5 nm. This customer is relatively open. Yeah, once we won, we can share more about that design, too. 5 nm, I think in terms of NRE compared to 7 nm, yeah, we're seeing the significant amount increase.
Oh, you mean the project value, right?
Right.
Okay, great. For the second 5 nm project you're talking about, is it a Chinese customer or also?
No. It's in the EU region. European customer.
Oh, okay. What type of product can you share?
I think it's HPC and AI also.
Oh, okay. All right. Okay, great. Congratulations, yeah.
Thank you. China customer, I think our current 7 nm customer all has a roadmap go to 5 nm. For sure not this year.
Oh, okay. Next year, any chance?
Yeah, next year, I think they have to kick off.
Oh, okay. Do we have enough engineering in place, yeah? It seems like that's one of our bottleneck, right?
Yes. You mentioned about the two biggest challenge for the company, capacity and capacity. One is a TSMC wafer capacity, the other is our design capacity. Yes, we try to increase our capacity as much as we can. Right now, I think the design successful, I think supporting customer well is the number one priority. Increasing design efficiency and try to utilize our design capacity, I think that's the thing we are doing almost every day. Also build the contest for whatever the project sales bring in, we have to select wisely right now because our capacity is limited. You are right.
Yeah, true. Okay. Now go back to the queue. Thank you very much. Yeah.
All right. Thank you very much. Mm-hmm.
Okay, Charlie, please.
Yes. Hi, good afternoon, gentlemen. Congratulations for a very strong result and execution as well.
Thank you very much.
Yeah, I guess you are now getting some good problem, right? I mean, customers start knocking your door, but you may not be able to service all of them. My question is that where can they go? I mean, do you see any other industry peers, competitors that can get these projects? If that's the case, does that mean that your industry peers will also grow and learn from those HPC projects in the future? Can you comment on this front?
Yes, to answer your question, yes, capacity we are facing very challenging limitation. I agree with you. I don't think customer were waiting for us. If we cannot take, they have to find other solutions. At this moment, our priority will be the same. Serving the existing customer is the number one priority. Serving the big NAM, whoever has a bright future, sustainable business is our second priority. Other than that, I think if a customer want to, we do not have enough capacity for them. In usual case, we will refer, we consider is a more reliable partner or a competitor for them.
Yeah. The ASIC market grow quite a bit. For sure, I think our competitor or other partner will receive some business, yeah, flow over from us.
Okay.
That's the reality. Yes. Mm-hmm.
Okay, thank you.
Okay.
Next question, I guess, for new investors that are all very concerned about your customer concentration. I guess now your top customer may account for maybe 40%, even 50% of your revenue. I guess two part of question. First of all, how does company convince investors that those local CPU or supercomputer projects wouldn't break the rule of the U.S. export regulation? Second question is more long-term, right, when that China CPU customer become larger and larger, whether they want to build their own design team and become TSMC's direct accounts? Vice versa, TSMC really want to open to them to service this account directly and skip the design service. Can you address those concerns? Thank you.
Okay. Daniel, do you want to go first?
Okay, sure. For the customer concentration, of course you know that Phytium is now our biggest customer and is the most important customer, accounts for around 40% of our total revenue. This customer, we expect in the second half or in next year, their revenue contribution will still be strong. Our second-largest customers, as long as the 7 nm start to begin the shipment, we believe the percentage for the second-largest customer will quickly pick up. If you consider the first one and the second one, I think the percentage-wise, we don't worry about too much the concentration to Phytium. You may also think that for the AI customers, as I mentioned many times, that we believe the 7 nm will be the major battlefield for the AI application.
We have multiple AI application customers and their 7 nm projects most likely will start to ship in the second half next year. As long as those AI customers' shipment begin, I think our customer mix will be more balanced in terms of the growing pace for each of them.
Okay, thanks. I guess, Johnny, please go ahead. Yep.
All right. I think Daniel is right. I see stronger and stronger forecast from this particular customer, but we're also doing the calculation starting from next year. Even this customer's demand increase, but the weight, overall percentage concentration will decrease starting from next year. Yeah, in the other words, just like Daniel mentioned about other account, production peak up I think is also very significant. Yeah. Another few customer in China and also a few customer in other region, I think peak up quite a bit. Next year, you can see the concentration for a specific customer will decrease even their revenue increase. To answer your second question is U.S. export control and also TSMC's concern. I think This company focus on the data center and also the CPU is a pure consumer-related product. We've been confirmed over and over again, it's very clean.
Everybody know the trade war tension keep increasing. To be honest, that's better for the company. From our point of view, and also from customer point of view, also from TSMC point of view, the current cooperation scheme is much better. Yeah. We draw less attention to the trade war. I think that the balance, Phytium plate go through Alchip to do the physical design, and then we place the order to TSMC. This kind of relationship, I think, for sure will sustain. I don't think customer at this stage will try to approach TSMC directly.
Okay, fair enough. Last question, I will get back to the queue. Johnny, we never seen the ASIC defensive business model can grow so fast. Right? Long term, what was the kind of the revenue scale you can envision based on the current trends? What would be a key risk for you to get there? I know this year company is shooting for maybe slightly above $200 million. Let's talk about what need to happen for you to get to $1 billion revenue scale. What would be the key risk? Thank you.
Okay. Yeah, I think right now, when we talk to customer, we are always thinking about the win-win situation. To be honest, customer also doing the very detailed calculation. They're only doing one or two chip per year. If they want to build out a very strong team and acquire very expensive EDA tool, also try to sustain all the employee, it's very costly. They also agree. Also in terms of design capability and outcome, our company takes about 20- 30 design every year. Compared to a customer doing by themself, they only do capability and also knowledge accumulation, it's totally a match. Yeah.
Okay.
As long as our margins are reasonable, customers in a safe zone, I believe our business will sustain and grow. Because HPC and AI, unlike a consumer, most of our customers are highly profitable. Even with certain margins to Alchip, they still have maintained a significant growth on their side. Obviously whoever doing HPC and AI are giant companies. Social media companies or the very big companies in China, I think even with our margins, they're still thinking about the value we provide. It's justified.
If we can keep accumulate the customer, maintaining the existing customer. That's why I say number one priority is existing customer. If we can continuously adding more customer, revenue growth, I think, is expected.
Okay. Mm-hmm.
Yeah. HPC and AI demand, we can say, is getting higher and higher. Hopefully. I'm not saying shooting $1 billion very soon, but we can say year-over-year, we have a reasonable growth.
Okay. Fair enough. Thank you.
Thank you, Charlie.
Okay, Haas from Credit Suisse. Please.
Hello, can you hear me?
Yes, I can hear you.
Yeah. Hi, Johnny and Daniel. Congratulations on the good results. I just have a couple of questions. The first one is regarding your existing customer in China for the CPU business. Recently, they have the product launch event and mentioned that they have the goal to launch on 14 nm products by the end of this year. Will those projects outside TSMC still be your business with this customer? Or do you think there is any risk you could lose share in this customer? Thank you.
Okay. Yeah, Haas, I think you're doing your homework. You're doing very good homework. Yes. This customer has recently have a big announcement on their future roadmaps. 14 nm, yes, it's not belong to TSMC. As you know, due to the trade war tension, most of our China customers thinking about their local supplier chain. I think most of them, they have to consider the local foundry. For this particular customer, we have a very tight relationship of more than 10 years. They've been approach us many times. Based on our suggestion, in terms of PPA or even in terms of cost, using TSMC solution is the best. They understand. If due to any reason, they have to go to other foundry, for sure, that will be also our business. We will recommend the best solution for our customer.
So far, we always consider TSMC in terms of PPA and price. If they insist go to other foundry for other due to certain reason, I think serving the customer is our number one priority. I hope that answered your question.
Okay. I would say in this way, that I know, Phytium announced its roadmap. For the 14 nm, they announced the four projects for 14 nm. Two of them are for notebook PC, and two of them are for embedded use, embedded application. For embedded application, I would say, that is also a new area for Phytium, and due to some reasons, we may not consider to take this project. For the 14 nm within the PC and the notebook category, we believe because of the technology, because maybe of the yield rate, because maybe of the chip efficiency, their bottom line in TSMC will still be their major bottom line. There will also be the next generation chip to tape out in TSMC for 16 nm and 7 nm . We don't worry too much about the current chip announced by our customer.
In the future, as long as SMIC's technology picks up, I think, of course, we have the ability and we have good relationship with our customers, and of course, we have opportunity to get involved in this business.
Yeah. That's very clear. With Chinese foundries limited access in the advanced nodes, at this point of time, you don't think too much concern that Phytium will shift mainstream server and PC business to other foundries. Even in the future, that SMIC can pick up the advanced capacity, you still have the capability to design in that foundry and working with Phytium on other projects as well, right?
Yes.
Yeah.
We have the ability. Our stance, we are not in a good position to comment on the foundries. I can just tell you that TSMC is a very respectable company. Their support, their flexibility, all of those categories are the best in the industry. Of course, TSMC still will be our major foundry in the future.
Okay, that's very clear.
All right.
My follow-up question on the China customer would be your new opportunity. Could you discuss more in detail on the new customer opportunity, in addition to the current CPU customer in China? Which applications and which nodes? Could you also share what is the timeframe for those new products? Thank you.
Johnny, you go first.
This particular customer, I think they are doing the data center and HPC related application. For sure, it's 100%, it's very clean, but due to the trade war tension, they try to keep a low profile. Please bear with us. I know the volume will increase. We are under NDA, we cannot disclose too much about this customer. Since the number pick up quite a bit, and so Daniel and I have to share this existence of this customer, because that become our number two very quick. If 7 nm capacity is not the issue, then starting from next year, I think they are quite closer to Phytium. Unfortunately, I cannot share too much about this customer.
Okay, Haas, for your question, I will say, you may pay a little bit attention to the trend that, in addition to HPC, the processor, the CPUs and the networking ASIC. You may think about what China wants the most. First of all, the CPU and maybe the storage IP and, of course, the DRAM. Those ICs accounts for the majority part of the IC consumption. There is another missing part, the GPU. I would say, maybe in the near future, there are some opportunity for us to get into the GPU supply chain.
My follow-up question is on the NRE content, the value you can get from a 5 nm project versus 7 nm project. Could you quantify it or just like a % increase for 5 nm versus 7 nm? Thank you.
Okay. To be honestly, at this moment, the old VCA member hasn't doing any full mask table for 5 nm yet. TSMC also working on the program with us on the five 5 nm NRE. We don't know the precise number, but based on the trend, from 17-7, from 28- 16, initially, 5 nm will be very expensive. I'm not surprised it will be 2x compared to 7 nm.
Okay. When it turns into mainstream, how much content gain would you think you can get, compared with the historical experience? Thank you.
TSMC, we are doing the NRE reduction every year. The precise number, it's not appropriate to share at this stage.
Okay. I can offer you a range, but the range is very wide. I would say 7 nm, the total NRE contract will be around $ 20 million-$40 million.
Okay.
Yeah. That, including the IP and also how many layer of a mask customer need to use.
Okay. $ 20 million-$ 40 million for project run.
7 nm. Yes.
No, 5 nm.
5 nm.
Okay. Yeah. Just a few more follow-ups. On the margin outlook, could you share more detail on the NRE versus turnkey trend for the full year in the second half? For your full year gross margin outlook, could you also provide an update, since it seems that you will have multiple NRE on 7 nm reaching milestone, and you can recognize more NRE revenue while the server shipment is also going to ramp in fourth quarter. I'm just wondering if you're updating any margins guidance. Thank you.
Okay. For margins, the guidance remains the same. This year, most likely, our gross margin will be the mid-30%s. You can see the first quarter and the second quarter. For the third quarter, the NRE percentage will be a little bit higher. We expect the third quarter gross margin will be higher compared to the second quarter. For the fourth quarter, as mentioned, the 7-nm shipment begins, and the production revenue percentage will account for even higher. All in all, I will say the gross margin percentage will peak in the third quarter. For fourth quarter, will be a little bit lower. For all in all, the mixed gross margin for the whole year will be the mid-30%s.
Yeah, that's very clear. My last question would be on the investment write-off. Could you share more detail on that? Thank you.
Investment-
Investment, the bond write-off.
The bond is Virgin Australia, the portfolio that the private banking provide to us. As you know, because of the coronavirus, all the airline companies are now in bad shape. For Virgin Australia, actually, we invest in a very conservative approach. For this bond, usually we bought a bond for $ 1 million each, but for this bond, we only bought for $ 450,000. In a conservative purpose, we write down all the amount we hold on Virgin Australia. Yeah, that's the detail. Yeah.
That's very helpful. I'll be back in the queue. Thank you, and come back again.
Thank you, Haas.
Okay, Charlie. Please.
Yeah, hi. Which private banking? No, I'm just kidding. My question is actually about your supercomputer project. I know a company won several awards for your customer. Those are TOP500 supercomputer competition. I'm not sure how we can predict those supercomputing revenue stream, because it seems like comes every one or two year, or two to three year. In coming to the three years, do you expect more of this supercomputer project to come? Which region do you see more demand? Thank you.
Okay. This company receiving the Green500 award is called Preferred Networks. The reason I can mention about the company name because we have a press release we sent months ago. Yeah, they are doing the HPC and also AI with very talented people. Preferred Networks right now is the number one Japan AI company. To be honest, even they are receiving number one, I don't think the volume will be too big. Because of the chip size and also the Japan market, currently we are working with this customer on the next generation 7 nm.
Because of this winning. A lot of our customers in the U.S. and also in Europe, and also China approach us because press release, they know we can do. I think winning this Green500 or TOP500 does bring us a lot of new opportunity. Yeah, right now we have more chance to select the right customer, I think, fitting our goal. Yeah. Preferred Networks. If you have a chance to go through the detail, I think that's a very unique design.
Okay. Thanks, Johnny. I think you mentioned you want to prioritize the existing customer. For Preferred Networks, since you are more open to talk about this customer publicly, when do you think 7 nm can start to kick in, contribute your NRE?
Yeah, in fact, 7 nm already started. We plan to tape out sometimes next year.
Okay. Follow up on this question. I know your China opportunity is much bigger as you just described it. Revenue size could be very huge next year. How about those Japan customers, turnkey or mass production revenue? Why it cannot be as big as those China customers?
I think because of market demand. We frequently see some opportunity on supercomputer in Japan, but based on the track record, the volume is not so significant. Japan, we have other approaches. I think later, if we have a chance, I can explain into detail for you. Last time, we announced one of the game console win. I think that's a significant win from Japan. Japan, there's two area, I think, it produce a very reasonable market share and volume. One is camera, the other is multifunctional printer. Even they are not HPC and AI, but in fact, we do receive certain RFQ on these two areas. Hopefully, we can win.
Okay, thanks. Lastly, also on this supercomputer project. I know, in the past, you also did some project for a Phytium customer. Next year, I would assume the second-largest customer is also related to supercomputer. From what we can see over the past two years, U.S., they put those supercomputer as one of the item they will watch closely for the technology export. I think those are including AI, surveillance, supercomputer, also quantum computer. Since that supercomputer is in that list, do you concern it could be a big risk for you into next year?
Yeah, for sure, there's a risk. I think the volume and also based on this trend, majority of our revenue is from data center and also CPU type of design. Even for those application can give us revenue growth into certain degree. The supercomputer, so far, I think is R&D. We did not see the tight restriction from supplier and also, for example, from TSMC. Every time we receive the RFQ, we will sit down with major foundry, with TSMC, and make sure we can take those kind of project. They also police our design win. I don't think in the short period, there's a certain tension on that area.
Okay.
Yeah, Daniel, do you have anything to add?
Yeah. I mentioned many times that I would say TSMC is the gatekeeper. Until now, we don't hear anything about concerning the projects for our customer.
Mm-hmm. Yeah.
Yeah.
In the real business practice, would you ask your foundry partner or lawyer to consult the U.S. Department of Commerce proactively to have some clearance?
Actually, TSMC did that.
Okay.
Yeah, we will.
Okay. That's all I need to know. Thank you very much.
Thanks, Charlie.
Okay, Haas, are you trying to ask questions? No. Okay, is anyone here wants to ask questions about our company? Okay.
Szeho Ng.
Szeho Ng. Please.
Hello. Can you hear me? Yes.
Yes.
Yes. I have a follow-up. Actually, would the company consider to start building up the internal IP library?
Yeah. Let me try to answer your question. I think we have everything. Team size is not small, but again, my principle remains the same. We are only doing the IP. As you know, a lot of leading-edge technology on the packaging side or even on the driver side, it's very difficult to sourcing the IP from existing partners. We have to do it, and customize it for customer. We're doing the IP is to enable the business, but company will not get into the pure IP business.
Okay.
Yeah. We are highly rely on partner. As I mentioned many times, leading-edge technology requires tons of IP. It's no way for the company like us to prepare all IP for our customer. We need to rely on partners. We don't want to compete with the partners.
Okay. True. Yeah. You will be very selective even if you really want to build some IPs. Yeah.
Right.
Okay. Got you. Yeah. Okay, thank you.
Thank you.
Okay, Charlie, please.
Yes. I want to follow Szeho 's question.
Sure.
Yeah, because I think you guys also hear that there is some concern about your HBM2 implementation. That is not something that EDA tool vendor, they can supply now. It depends on the design service capability to implement. How confident are you on this HBM2, this kind of IP, this kind of design service and also can you take this question first?
Okay. Yeah, I know there's some saying about our HBM design capability. Talking to the customer, we never encounter this kind of problem. We have a few existing customer using HBM. Now, one of them will be taped out soon. We have at least four to five pipeline customer who are using the HBM solution. I consider all of them are satisfied with our solution and proposal. I think once the HBM design hitting volume, I think that we will ease all the concern from the outside saying. Right now, I don't think any showstopper from our HBM2 solution.
Okay, thanks.
China customer, Euro customer, or U.S. customer, we all have HBM2 customer there.
Okay. Another follow-up is about your AI ASIC. I still remember over the past three years, there has been some delay, not just at your company, but also your industry peers as well, because those AI ASIC need a lot of time to verify to do those field test.
Do you think those AI system customers, they will also see this delay risk in the coming years, or they already figured it out? Going forward, those AI projects will be on track. Thank you.
Okay. To be honest, we're doing a lot of AI, but in terms of production, AI does contribute just a little bit. Even though AI story and AI dream is not gone, we see more and more demand, even without volume, but we see more and more demand appear. Starting from next year, I think the situation will be much better because one of our AI company in U.S. already in volume production. I think their volume is reasonable. Another Europe company, recently, they just sell to the U.S. customer. I think their volume will be peak out next year. Hopefully, the AI contribution starting from next year will increase a lot, hopefully.
Okay. Got you. Thank you.
Thank you.
Okay. Here is a question from Eunice Baritz . Could you please briefly comment on how you are constrained in terms of capacity and how you optimize the capacity allocation? Okay, Johnny, would you like to go first?
I believe the capacity is for the design capacity, right?
Yes.
Okay. Like I mentioned before, our demand is much, much more than we can supply at this moment. Our priority remains the same. Serving the existing customer is the number one priority. I don't think we have a problem to support our existing customer. The new customer, when they come in, we have to do the beauty contest angle, I think it's straightforward. Either this customer has a future and business is sustainable, or they can immediately give us a certain profit. I think selecting customer is very painful, but it's straightforward. In the other way, we are increasing our capacity very aggressively. Like I mentioned before, by the end of this year, we'll have 500 people. We can support 20 projects in parallel by the end of this year, and also increase the design efficiency.
We have one platform that can reduce a lot of unnecessary resource. I think that's one of the areas we can say we're doing better than our competitor, design efficiency, and I think the detailed distribution is a long story. I'm not going to discuss at this stage. Right now capacity is growing. In the meantime, we have to select our customer wisely.
Eunice, in simple talk, I think your question will conclude into the engineering capacity, the engineering efficiency, and the engineering flexibility. For the capacity, we are doing aggressive expansion. As Johnny mentioned, we will hire new engineers, train them to do the design for us. For the efficiency, we are trying to build out the platform, and we are trying to separate the high-level design and the low-level block design, and to allocate the low level to the entry-level engineers and to allocate the more complicated project leading things to senior engineers to improve the efficiency. For the flexibility, also we are looking for and we already have some outsourcing partners. We can outsource or they can insource to our company to do the low-level block design for us to expand our total capacity. I think the time is limited.
This could be the last question. Guotai Junan Liu Zihao. You can ask the question now.
Hi, Daniel and Johnny. Can you hear me?
Yes.
Yes.
Okay. Thanks for taking my questions. My first question is regarding to your second-biggest company. I know you cannot mention the name, but I remember I asked about the revenue contribution last time in the last conference. I believe that, Daniel, you guided for the second-biggest company will contribute around $8 million in the fourth quarter. I'm wondering, are you still seeing this number, or are you seeing a little bit more than $8 million in the fourth quarter?
There will be upside, but depends on the capacity support by TSMC.
Okay. Got it. I just want to double-check that you mentioned that the revenue contribution will get equivalent to Phytium in next year?
I won't say equivalent, but the percentage we reach and the absolute contribution will pick up very quickly in the first half of next year.
Okay. Phytium has contributed about 40% of your total revenue. This one will reach around 40%, I mean, 20%, maybe?
Yeah, 20% we are pretty confident. Yeah.
Okay. Got it. Okay. Thank you. I also want to ask about Phytium's. Can you provide any updates on the Phytium three project? I know it has server, PC, and you mentioned about HPC last time. Can you provide any update on Phytium, especially in 7 nm in server and HPC? When will the NRE kick in, or when will the mass production kick in?
Sorry. Actually, recently we received great pressure from our customers. They don't like us to talk too much about their budgets and especially the schedule, the tape-out schedule, the shipment schedule. We have been trying very hard to let investors understand the future trend. For some specified projects and schedules, now it's difficult for us to go into detail. Please understand.
Okay. I see. My last question regarding to I'm just curious, who will be your third-biggest customer next year? Would it be a Chinese customer, or would it be a U.S. customer like Astera Labs, as you mentioned this year, or Amazon?
Yeah, hopefully it will be U.S. customer will be our number two or number three next year.
Okay. Is it going to be Astera Labs? I know the company sees the chance next year will be very great regarding the PCIe Gen 5, right?
Yeah. I think there are two particular customer you mentioned about that has a chance to contribute more to our company.
Yeah. We don't have the numbers of forecast for next year currently. We can only say that we put high expectation on several customers. First one is the second largest customer, China customer right now. We are pretty confident his contribution will pick up very quickly. For the other customers, of course, for those AI companies, like the U.S. AI customer and the European AI customer, we do believe that in the second half next year, the shipment contribution will be good. The expectation is high, we cannot confirm the numbers right now.
Okay. Got it. Understood. Very clear. Thank you.
Thank you.
Okay. Last question. Really the last question. Soon. Can you say your name and company first?
Esoon?
E-S-O-O-N.
Sorry. Hello both. Can you hear me?
Oh, yes.
Hi. Sorry, I forgot to unmute. It's Elizabeth Soon from PineBridge.
Yes, I'd like to ask very simple questions. I understand the founder's shareholding is actually quite low in the firm. Any options for either the engineers or the senior management, so there is less likelihood of them moving on to other competitors? Secondly, if I hear correctly, because I was late to dial in. I understand you had an investment, is it correct, in Virgin Australia, that you have made a loss? What is the future investment rationale in terms of usage of funds, and do we expect investment beyond their own businesses? Thank you.
Okay. We keep all the investment-related q uestion aside. I think, or Daniel can comment on later.
Yeah.
In terms of employee loyalty, I think that area is our company, we've been the most proud of in that area. Our company is 17 years old, we have engineer working with us for more than 15 years, I think total about 50 people. People working with us for more than 10 years is over 100. In terms of loyalty, I don't think that's the biggest concern. They all satisfy with our current performance. Also, all of them own piece of a company. I think this trend will continually grow. Yes, we do encounter some problem when we try to hire new college grad. I think, in China, relatively, we are the only company. We can do 5 nm and 7. I think we don't encounter much problem to attract the young guy.
I think employee loyalty and stability, I don't have too much concern.
Okay. For your question for the bond investment, actually, we have done the bond investment since year 2015. The portfolio in the beginning is about $ 20 million for the whole portfolio. We are a conservative company. We keep on shrinking our portfolio scale. Far, the total portfolio scale is about $10 million to $ 11 million. Yeah, most of the target companies bond are provided by the private bankers. Of course, unfortunately, the airline company, the original stock is Virgin Australia. It's a very famous company, and it's the second largest airline company in Australia. We think this company is relatively safe. Because of the virus, I guess no airline company is safe right now. Fortunately, the position is not high. Secondly, the Bain Capital has already taken over this company. We expect there will be future reverse amount of the write-off.
Okay. Do we assume that there'll be more investments from the $10 million-$ 1 million?
No.
I mean, in other stocks. Not in Virgin, but in other stocks.
Not stocks, bond. We only invest in straight bond, straight corporate bond.
Okay. Got it. Okay.
Yeah. We don't have plan to extend our portfolio in the future.
Okay, great. Thank you very much.
Thank you. Mm-hmm.
I guess that's the institutional investor meeting for today and for the second quarter. If you have any questions in the future, please, you are welcome to contact me. If I have time, I'll be more than glad to answer questions or have discourse with you. Thank you. Thank you for joining today's meeting.
Okay. Thank you very much. Thank you.