....at the end of the presentation. In order to do so, please use the raise hand function on your screen, or for those on in, it is star nine on your keypad. I have the pleasure handing over to Technoprobe CEO, Stefano Felici. Please go ahead, Stefano.
Good afternoon, everyone, and thank you for joining us for Technoprobe's results presentation for the first nine months of 2026. As usual, I am with Stefano Beretta, our CFO, presenting the nine-month 2026 results and the guidance for the last quarter of this year. We delivered strong growth during the period, supported by AI-related demand and our ability to translate that demand into operating results.
Our performance in the first nine months of 2026 reflects the combination of structural AI growth and strong company execution. Revenue reached EUR 787.4 million, up 68.8% compared with the same period last year. Gross profit increased by 116% to EUR 447.3 million, with a gross margin of 56.8%. EBITDA reached EUR 376.3 million, up 158%, with a margin of 47.8%. What stands out is that profitability grew significantly faster than revenue, demonstrating the operating leverage in our business as volumes increased.
On the demand side, growth in GPUs, CPUs, and ASICs continues to support our business. At the same time, increasing device complexity requires greater test intensity and more test insertions. Our opportunity, therefore, reflects both rising volumes and more demanding testing requirements. On the execution side, our focus will continue to capture this demand and delivering at scale while maintaining consistent operational performance.
We are also laying the foundations for further improvement. That means investing in capacity and strengthening the manufacturing processes needed to support a larger business. Being capacity the key driver also for the next year, in addition to a deep dive into nine-month and quarterly financial results, we will discuss the state of art of investments we are making to expand the manufacturing capacity. Let me now hand over to Stefano Beretta for further details on nine-month results.
Thank you, Stefano. Good afternoon, and thank you for joining us. Today, I will briefly review our third quarter and nine-month results before focusing on the most important message, how we are preparing Technoprobe for an extraordinary phase of capacity growth in 2027 and 2028. The third quarter marked another record for Technoprobe. Revenue reached EUR 323.3 million, up 130% compared to the same period last year, and 16.7% sequentially, reaching the high end of our outlook range.
Gross profit was EUR 197.7 million, representing a margin of 61.1%, while the EBITDA reached EUR 170.2 million, with a margin of 52.6%. For the first nine months, as already showed by Stefano, revenue was EUR 787 million, with an increase of 68.8% year-on-year, and gross profit reached EUR 447 million, with a margin of 56.8%. The EBITDA was EUR 376 million, up 158%, with a margin of 47.8%. This growth was organic and driven primarily by the artificial intelligence and data center applications. The consumer segment recorded moderate growth, while automotive and industrial continued their gradual recovery.
Foreign exchange remained a headwind compared to the same period of prior year, with a negative impact of approximately EUR 30 million on the nine-month revenue, with the nine-month average USD/EUR rate at 1.16 versus 1.12 for the same period of 2025. Profitability, both at gross margin and EBITDA level, benefited largely from the improvement in production efficiency, product mix, operating leverage, as well as the partial expansion of manufacturing capacity realized so far.
Stable overhead and R&D costs also supported the strong EBITDA performance, even including a negative forex impact for approximately EUR 22 million, which would have brought the year-to-date margin at 48.6%. Our financial position gives us the ability to keep investing ahead of growth. At the end of September, our positive net financials position stood at almost EUR 800 million, while operating activities generated approximately EUR 241 million of cash during the first nine months. These results matter not only for what we have delivered, but also for what they allow us to do next.
We can invest in facilities, equipment, people, and systems today, rather than wait until demand puts additional and unsustainable pressure on our ability to serve customers. Let me now turn to our capital expenditure plan, which is the central element of our preparation for the next phase. We confirm approximately EUR 350 million of investment through the first quarter of 2027, with around 75% expected in 2026 and the remaining 25% in the first quarter of next year.
The investment plan slide shows EUR 167 million finalized in the first nine months and additional EUR 158 million already committed at the end of October. Approximately EUR 270 million of the overall program is already allocated to Italy and EUR 80 million to the rest of the world, in particular to Taiwan and China. The program combines new facilities with the optimization of our existing manufacturing footprint. We are not simply adding floor space. We are increasing throughput, introducing the next generation laser cutting and assembly equipment, reconfiguring layouts, and preparing dedicated production lines.
At the Cernusco Lombardone , the completion of the new facility is now expected in December 2026, with a ramp-up targeted for February 2027. This brings forward the construction completion compared with the timing discussed at the half of year and adds 4,000 sq m for probe head assembly. Consistently, additional space in the other part of the facilities, especially in Agrate Brianza, will support further expansion in probes manufacturing. In parallel, we are repurposing existing production areas and improving manufacturing processes across our sites.
Together with the new facility, all these investments are expected to bring our annual manufacturing capacity run rate to approximately EUR 1.6 billion in the first quarter of 2027. We are also developing a further investment plan for a total amount between EUR 150 million -EUR 200 million to the end of 2027, and without increasing the footprint. That could progressively increase that capacity run rate to approximately EUR 2 billion by the end of 2027.
Just to avoid any misunderstanding, these figures describe manufacturing capacity, not revenue guidance, and the timing and utilization of that capacity will depend on the demand and the execution. Our ambition extends beyond the next quarter and beyond 2027, so we are preparing the ground for an extraordinary expansion capacity across 2027 and 2028, so that Technoprobe can support customers and to substantially grow the scale.
The priority is operational readiness. Buildings and equipment must be ready, but so must the people who operate them and the system that coordinates them. That is why the CapEx program cannot be viewed separately from our digital transformation and hiring program, through which, by the end of this year, we will have more than 5,100 employees, with an increase of approximately 1,700 people compared to the beginning of the year. Our preparation also extends to China, where Nexprobe combines our Shanghai office with the Suzhou facility, covering approximately 12,000 sq m.
The platform brings together local probe manufacturing and assembling, supported by our proprietary patents and technology transfer from Italy, with local engineering and customer support. The ramp-up is expected within the first quarter of 2027. Across the group, our manufacturing investment builds on a shared proprietary technology base spanning ceramics, lasers, and precision mechanics. Let me turn the floor to Stefano Felici, who can walk you through this page.
Yes. Thank you, Stefano. Let me close with the technology that underpins our growth and positions us for the opportunities ahead. At the heart of Technoprobe is a proprietary technology core spanning ceramics, laser technology, precision mechanics, PCBs and MLOs, electronics, and MEMS. Our strength comes from bringing these capabilities together to solve increasingly demanding test challenges. As semiconductor technology advances, this shared foundation allows us to deepen our presence in existing markets and expand into new applications.
In AI and high-performance computing, customers need greater test capacity and throughput while managing higher currents and more demanding thermal conditions. Our capabilities in parallel testing, high current handling, and thermal management directly address these needs. As silicon becomes more complex, test intensity per chip also increases. Fine-pitch MEMS, dense contacts, and precision integration enable us to meet tighter contact requirements, reinforcing our technical differentiation.
In HBM, the challenge is to combine test parallelism with speed and power handling. Our capabilities extend up to full wafer vertical probe cards, supporting our expansion in this application. Silicon photonics opens another avenue for growth. Here, our capabilities include autonomous electro-optical probe cards with embedded high accuracy and stable fiber array unit positioning. This brings optical and electrical testing capabilities together to address applications such as co-packaged optics.
Across industrial and AI power applications, we can apply our expertise in high current handling, thermal control, and stable contact across temperatures to markets, including ADAS, microcontrollers, and wide bandgap power devices. The key point is that the same proprietary technology base supports multiple growth opportunities. We can build on existing expertise, combine capabilities in new ways, and tailor our solutions to customers' evolving requirements. Our nine-month results demonstrate the financial performance this positioning can support.
The investments we have outlined today are designed to give us the manufacturing scale to pursue the opportunities ahead. Let me hand over to Stefano Beretta for the fourth quarter guidance and final remarks.
Thank you. Let me briefly address the near-term outlook. The forecast for Q4 largely reflects the impact of the digital transformation process launched on October 1, 2026, which led to the full adoption of a new ERP, a new MES, and new warehouse management systems, together with other integrated vertical systems. The impact on operation was, as expected, significant during the first months of adoption, and a full recovery in efficiency is expected by the end of 2026.
For Q4, we expect revenue of EUR 270 million, ±3%, a gross margin of 54.2%, ±200 basis points, and an EBITDA margin of 45.1%, ±200 basis points. As I set out in our prior press release, we almost confirm our full year 2026 range of revenue with a gross and EBITDA margins in the region of 56% and 47%, respectively, in the mid-range. The temporary operational effects of the systems transition and the cost of preparing our organization for the next ramp should be viewed alongside our long-term investment priorities.
Overall, we remain highly positive about the opportunity ahead while recognizing the uncertainties associated with geopolitics, trade policies, and the pace of recovery across end markets. Our response is to focus on what we can control, technology, capacity, people, and execution. Thanks for your attention. We can now move to your questions.
Thank you to the speakers today. We now have an opportunity for question. As a reminder, if you would like to ask a question, please use the raise hand function on your screen, or for those dialing in, it's star nine on your keypad. Once your name is announced, please unmute your line, state your company name before asking the question. Thank you. The first question comes from Giovanni Selvetti. Please, Giovanni, the floor to you.
Hello, guys. Can you hear me well?
Yes.
Hello. Hi. Thanks for taking my questions. I have a few. The first one is really on CapEx, if you can elaborate a bit more. Maybe, Stefano, you mentioned it before, but just to understand if I got it correctly. These new EUR 150 million-EUR 200 million of additional CapEx, is it going to be mainly, did you say machinery, or is like the purchase of another factory or the building of another factory?
What are the drivers of these additional CapEx, whether it is more demand for GPUs or if you are preparing for the ramp-up in production of HBM? If you can kind of give a sense on that, and maybe if we stay on HBM, whether if you have an update for us to share on the qualification process of the other two memory players. Thank you very much.
Thank you, Giovanni, for these questions. That brings me the opportunity to reiterate the message that it was mentioned during my prepared remarks. The new additional CapEx do not relate to new footprint. What we expect to invest in the second part of 2027, so starting from second quarter to the end of the year, relate only to equipment, machineries, and systems. That will allow us to increase the entire efficiency of our production, because we will be able to better allocate all the equipment altogether for certain department flows.
To answer to the second part of your questions, there are no dedicated departments by products, because our production is, let me say, transparent in terms of final output. We can manufacture needles for bed, board, and certification, and whatever is in our production flow. That can be almost easily translated into the final product, depending on the demand that we are facing from customers. There are no specific department allocated to specific products.
Okay. Maybe on HBM?
Yeah, I think you were asking about the qualification status. I think we can confirm that we have been qualified by one of the player and still in qualification for the other two. We can say that for next year, first of all, for this year, we expect a revenue of around EUR 10 million, and for next year, a multiple of that number that will be at least by three.
Okay, that is very clear. But with this, if it is possible, you are assuming the same, let us say, times three with the same customer, or times three including also, like, assuming the qualification of the others?
With the same customer, because we cannot, let us say, make any prediction until the qualification is done with the other two. The revenue is projected with the same customer right now.
Okay, thanks. I have more, but I'll jump back in the queue and maybe ask later. Thank you.
Okay.
Thank you, Giovanni. The next question comes from Alexander Duval. Please, Alexander, the floor to you. Please, Alexander, remember to unmute your line.
Oh, thank you so much. Alexander Duval from Goldman Sachs. Obviously, you've issued a new guidance of this EUR 2 billion capacity run rate by the end of 2027. Consensus on a full year 2027 basis is more like EUR 1.65 billion. Just wanted to understand what the run rate would mean in terms of that full year sales. How should we think about those overall revenues in 2027 that can be implied by that new run rate as you ramp, assuming the demand is there?
Secondly, for 2026, you did raise your revenue materially, but it seems like margins haven't increased to the extent one might have expected. Just wanted to understand that. Does it reflect conservatism, or is this perhaps a preparation for the expansion ahead? I've got a quick follow-up.
Okay. About the capacity run rate, let me reiterate clearly the message. This is what we are preparing to be able to manufacture by the end of 2027. At the end of Q1, we expect to be able to deliver EUR 1.6 billion on a yearly basis, starting from end of Q1.
Consensus reflects revenues rather than capacity. If you project at the year-end, the ramp between EUR 1.6 billion and EUR 2 billion will be gradual, based on the investment growth, together with the demand from customers and the utilization we will be able to put into our fab. This is the projection of a potential ramp that will be available only at the year-end. It doesn't mean at all we will be able to deliver EUR 2 billion right now by the end of 2027.
Understood. And perhaps on the margins?
For the margin, we expect the operating leverage will be more and more significant as far as we go from, how do you say, EUR 1 billion for 2026 to EUR 1.6 billion run rate to EUR 2 billion run rate. The EBITDA margin, considering especially the SG&A and the R&D impact on our accounts that are limited compared to the revenue growth, will be reflected in a growth of the EBITDA margin as well. Still to be quantified, because we need to see the ramp. By the way, if you look at the 2026 last quarter profitability, you will see a slight decrease in the profitability.
Most of that is related to volume, but there is also a component related to additional cost we are facing in Q4, especially related to labor. This is a message we want to deliver very clearly, because we have an hiring program for 2026 that consider approximately 1,700 hires. So, an exceptional effort entirely designed to support the rapid ramp-up we expect in 2027. A good portion of these hires do not produce revenue right now, but we are recruiting well ahead of the ramp because skilled manufacturing capacity cannot be created overnight.
The new colleagues need time to train, learn our processes, and become fully productive. We would rather have people in place and trained before we need them than arrive late. So, with the equipment installed, but insufficient trained staff to operate it effectively. This choice comes with some temporary inefficiency. We will incur costs before the full contribution of those hires is reflected in output. Our financial strength likely allows us to absorb that near-term impact, and we regard it as a deliberate investment in readiness and not a reason to delay the preparation of our customer we require.
Super clear. If I could just squeeze in one very quick follow-up. You have reiterated the EUR 10 million on HBM memory revenues, and you have now talked about tripling that. There has been some discussion in the industry about pushouts of stack heights of HBM. I just wondered on a multi-year basis to what degree that impacts how you are thinking about the potential on HBM, if we look out several years, if at all. Many thanks.
The potential for HBM is really related to our core technology, which is a vertical technology. We are leveraging from technology developed for logic devices like GPUs, CPUs, which are typically much more complex. The full potential, we expect to see the full potential of our technology with the next generation of HBM. I am talking about HBM5. Right now, we see already there is a lot of interest to qualify this new technology, the technology we are proposing, to be ready for the next generation. I think we will see the full potential in a couple of years from now.
Thanks very much.
Thank you, Alexander. The next question comes from Alberto Gegra. Please, Alberto, the floor to you.
Hi, good evening. Hope you can hear me. My first question is on capacity. You are giving this indication of capacity for 2027 to two billion run rate, but you stated that you are expecting to further increase in 2028. Just to have a sense, a rough idea of the CapEx that we should expect in 2028. The second on the guidance, the step-down in the last quarter seems a bit more pronounced than the initial expectation. I understand some internal inefficiency. Just wondering if there is anything to flag on the demand side.
The last one is a general question about visibility, because in August, you talked about good visibility from your customer, I mean, not in terms of order, but on project volume expectation, and so on. How this visibility has changed today? I mean, are you incrementally more or less confident about the growth of the next two, three years? Are you seeing stronger commitment from customer compared to the past? Thank you.
Thank you, Alberto. Let me take the first point about 2028 CapEx. It is a bit earlier today, again, to understand which will be the future of the further CapEx plan. We are now completing a very important run of investment, set to be completed in the next couple of months. Then we just mentioned another run of investment through the entire 2027, almost fully allocated to equipment and machineries and clean rooms, and what will enable us to increase internal efficiency without increasing the footprint.
For 2028, of course, there will be other options to be explored, but so far, they are still on the paper, and it is very, very early stage to show that. About the other point, we do not see any decrease on the demand from customer right now. The Q4 decline is almost entirely, if not entirely allocated to the inefficiency for the new ERP ramp-up. Same logic, as we mentioned for the hiring program, the same logic applies to this digital transformation. The new ERP went live on October 1st, and this is marking a fundamental step in building the infrastructure required for a larger and more complex Technoprobe.
The operational slowdown has been, as we initially expected, at the end quarters, the complete and necessary cutover, for example, required four days of total shutdown. October then involved amount of operating inefficiency as the team adapted to the new environment, and full operational recovery is expected by the end of November, beginning of December. We should be clear that a successful cutover does not mean that every process is immediately at full efficiency.
The priority now is to complete the stabilization and to ensure that through all our customer, translating that into new infrastructure for a day-to-day execution. The broader program includes a full adoption of the ERP, the MES, the warehouse systems. This is intended to give us a stronger visibility and control across planning, across production, and inventory, and help us to use our growing resources more efficiently.
This is why we would encourage you to look at the transition in the context of the next couple of years, rather than focusing on the last quarter, Q4 2026. We are putting in place people and infrastructure required for a completely new chapter of Technoprobe in the service of all our customers.
For the visibility, we are not seeing any slowing down, especially in the AI visibility. All the main players are still demanding us to increase the capacity, and they want this capacity as soon as possible. At this moment, we do not see anyone slowing down.
Thank you.
Thank you, Alberto. The next question comes from Bernd Klanten . Please, Bernd, the floor to you. Please remember to mute your line. Can you hear us?
Yes. Sorry, can you hear me now?
Yes, we can.
Sorry about that, and thank you for taking my question. Another follow-up on capacity, if I may. I appreciate that it's too early to talk about CapEx in 2028, but you've also shown that you can expand capacity quite a bit without spending new CapEx. Without any CapEx assumption, how do you think about that two billion number evolving over 2028? Then I just wanted to qualify on capacity as well, whether that excludes the opportunity from Nexprobe. Then I have a follow-up, if I may.
Let me say again, 2028, we will have still 2027 to be started. We are totally focused right now to deliver what we have planned already and secured with our customers. For 2028, of course, there are already some discussion in place, but that will be necessarily rediscussed with the main customers we currently serve to understand also the speed of their growth. For Technoprobe, it's pretty, let me say, easy to plan future growth as far as we have cash enough.
Again, to be self-funded is a really very clear base and important base for Technoprobe to have cash enough to fund our own investments, and also to be prepared with a strong supply chain. We are currently securing that will grant Technoprobe to have equipment enough in case we would need. A further development will be decided and disclosed in the next press releases. For now, we remain stick to 2027.
Got it. Thank you.
About Nexprobe?
My-
Yeah, the point of Nexprobe. Nexprobe is included in this plan for 2027. The ramp will start in January, and we expect in terms of volumes and turnover that they will be able to reach more or less EUR 100 million revenue in the next three years. So, by 2029, we expect the volume will reach a EUR 100 million contribution. Only for the Chinese market.
Thank you very much. Then one other question on custom ASICs in particular. Can you remind us of the status quo on the adoption of vertical MEMS there for custom ASICs and where you stand with the key players? Thank you.
No, at this moment, we are one of the main supplier for all the ASICs customers. We introduced new technologies, high-speed and RF technologies also for these devices. We are a point of reference for ASIC testing.
Thank you very much.
Thank you for your question. The next question comes from Martin Jungfleisch. Please, Martin, the floor to you.
Yeah, hi. Good evening. Thanks for taking my question. I also have a couple, please. Maybe if I can start on the CPU side. Can you disclose what the CPU testing growth was versus GPU and how you would that expect to develop into 2027? That's the first question, and then I have another one.
Sorry, Martin, can you repeat the beginning of your question? Because the line went busy for a couple of seconds.
Yeah.
About CPUs?
Yes. I am in Munich at the SEMICON. I was just wondering if you could disclose the CPU-related growth compared to GPU-related growth, and what you would expect how that growth would kind of trend into 2027, so the CPU versus GPU testing.
Actually, for 2027, we do not see so much difference compared to 2026 as far as the ratio between GPUs and CPUs. We expect some very large campaign of GPUs, new GPUs, and as well followed by CPUs. So, the biggest portion will be still on the GPU side, and the CPU will just follow.
Okay. That makes sense. Thank you. Then just two clarification questions, if I may. The first one is on the HBM side of things. When you expect this EUR 10 million this year and around EUR 30 million next year, that is just for the base die testing. Is that correct?
No, not only for the base die.
Okay. It is also the DRAMs. Just finally, also a bit more clarification on this Q4 sales guidance. Is it fair to assume that the impact on the Q4 would be roughly EUR 50 million on the top line, roundabout?
Yeah, on the top line, yes. Correct.
Okay. Great. Thanks a lot. That is all for me.
Thank you, Martin. The next question comes from Harry Blaiklock. Please, Harry, the floor to you. Please remember to unmute your line.
Hi. Good afternoon. Thanks for taking my questions. The first is just on a clarification on the comments on custom ASICs. You mentioned that you are one of the main suppliers for custom ASICs, but I thought last quarter you were saying that you were kind of just in the process of qualifying and gaining some momentum in custom ASICs. What has changed there? When you say main customer, is that for designs that are going into production next year, or are you already getting meaningful revenues coming through for custom ASICs?
No, no. We have already good revenue on the custom ASICs. So after qualification, we got very quick ramps on several new devices. That is why when I am saying that we are point of reference is already actual, it is not for the future. Of course, there are many new NPIs coming in the next months, but we have already very good technology for this type of devices already in production.
Okay. All right. On the capacity point, I mean, the guide for this year is super helpful for the two billion. Over the past year or so, as you have been adding capacity, you have shown that you can execute on this stuff relatively quickly and find more capacity when needed. Do you see any upside to that two billion number, or is that very much kind of capped by the existing space that you have?
Well, with existing spaces, I think we are comfortable for two billion capacity. Especially, we will be able to deliver all the growth, the equipment, and to receive from the supply chain all the equipment we will need. So we do not have particular constraint on space. What we have seen as a constraint was the hiring, and this is why we are trying to hire well in advance all the people that we need to run the fab in the next 12 months. This is the first point.
The second constraint, again, is on a supply chain base. We ordered already, we committed already a large part of that investment for 2027 to our suppliers in order to secure their capacity to serve Technoprobe. Most of the suppliers are, let me say, totally exposed to Technoprobe for their products and machineries.
This is very important for us. These are the only points we see as a constraint. Again, we do not have any constraint in terms of cash and in terms of decision. As you can see, when we invest EUR 200 million more, we can generate EUR 400 million more revenue. Even from a financial perspective, I am pretty comfortable as a CFO to approve an investment like that.
Yeah. Got it. Makes a lot of sense. One last question just on the competitive dynamic that you are seeing. I know historically you have kind of had one main competitor in the leading-edge logic space. But one of your Taiwanese competitors, who historically has had strong custom ASICs business, has been quite vocal about their ability to win share in the vertical MEMS space as they are adding capacity. I was wondering if that is something that you are seeing at all, or whether you still very much see this as kind of a duopoly at the leading edge between you and your U.S. peer.
Okay. First of all, I would like to say that we respect all the competitors. We watch them, what they are doing, and we very carefully check in the market the situation. But on my perspective, I can tell you that we think we are, I mean, the numbers are telling that we are the number one player for the logic in the logic space.
Probably the player that should worry is the number two, because the number three will, of course, will try to get some shares, but the situation will be worse for the number two supplier rather than number one. On our side, we think we still have the best technology, and we are already preparing for the next generations of chips.
Got it. Thank you. Very helpful.
Thank you, Harry. The next question comes from Craig McDowell. Please, Craig, the floor to you.
Hi, good evening. Thanks for taking my question. I had a question on capacity. Just when you put out these numbers, EUR 1.6 billion by the end of Q1, EUR 2 billion by the end of Q4 2027, can you maybe help us to understand what assumptions are in there in terms of price and mix? Presumably volume's fixed, but there must be a price and mix assumption in there. If you could give some color on that.
And linked to that, on pricing more specifically, I think we're hearing comments that on maybe on memory testing, there's been some pricing up on like-for-like basis in memory wafer test. Just maybe comment whether you're participating or is that something that you're able to do? Thank you.
Thank you, Craig. The capacity projection has been calculated with a cost and price mix impact. We do not project any particular increase in pricing or in the mix of the products. Of course, there will be some changes as far as the technology will go farther. There will be more complex product, probably with a higher price, but just because the components are more expensive.
But in terms of negotiation with customers, as usual, we do not expect that price increase will come directly from Technoprobe, even in a situation of extra demand compared to the offer. The current projection of capacity is, again, calculated with a comparable and constant price mix.
Thank you. Very clear. Just on gross margin as well, maybe if you just clarify if there's any, in the Q4 guide, if there's any one-off related to the ERP implementation that we should be aware of, just as we think about gross margin trajectory into 2027 as well. That'd be great. Thank you.
Thank you again. In terms of gross margin, there will be no particular one-off cost for the ERP introduction and all the digital transformation. The vast majority of this cost will be capitalized and amortized through the utilization of the softwares. They are not impacting directly the P&L as a one-off cost. The most impactful cost for Q4 with no particular revenue allocated related, as I mentioned before, to labor especially.
The entire hiring with data analyzation we have concluded in Q3 and Q4 will be impacting the Q4 P&L, but again, without associating particular revenues. This is the main explanation of the Q4 little squeeze of the profitability. That is temporary. It's not a one-off because the labor cost will be projected also over the other quarters and go on. But in particular for Q4, there will be some, let me say, not regarding, not correlated cost to revenue.
Very clear. Thank you very much.
Thank you, Craig. The next question comes from Fabio Pavan. Please, Fabio, the floor to you.
Yes. Hi, and thank you for taking my questions. Actually, most of them have been asked already. So maybe to sum up, I would love to ask you if you can share with us your expectation in terms of revenue line evolution for 2027, considering what you just shared with us. And also, Stefano, you were mentioning we should look to meet long-term prospects. So with that in mind, I was wondering if you are considering to share with the market an updated target model maybe at some point in 2027. Thank you.
Let me make the first point. So revenue 2027, we go back to the visibility. Visibility is still short by quarter, even if we have more conversation with customers when they want to be, let me say, more allocated with a longer term view for 2027. But still, let me say, to quantify on a right number, 2027 revenue expectation is still very early. What we want to focus again is the capacity. The revenue is a consequence. If you do not have capacity, you do not have revenue. So, our first point and our first focus in our mind is to secure the capacity.
Revenue will become more clear in the next few quarters. About the target model, again, is very consistent with the capacity run rate. So the target model in the short term, in little more than 12 months, is to arrive at EUR 2 billion. But if you think that only one year ago, 2025, we were still at less than EUR 700 million, it means like three times in less than two years. So it is a big jump. And for the moment, this is as much as we can do. Then for 2028, again, we are still discussing new plans, but that will come with new customer conversation and to see also the market, and how the market evolves.
Thank you.
Thank you, Fabio. The next question comes from Oliver Wong. Please, Oliver, the floor to you. Please, Oliver, remember to unmute your line. Can you hear us? I think you are having some technical problem. We will then just move on to the next question. The next question comes from George Brown. Please, George, the floor to you.
Yeah. Hi, guys. It is George Brown from Deutsche Bank. Thanks for taking my questions. I have three, if I may. I will take them one by one. Just firstly, I understand that you flagged this temporary impact in Q4 from this digital transformation project, the ERP system, et cetera. I think you said the top line impact was roughly EUR 50 million. Just wondering, what would be the underlying margin in Q4 without this sort of top line volume impact? Would the Q4 margin be similar to Q3, excluding this temporary impact, or would it be maybe slightly lower due to the labor cost? And I have a second and a third.
Yeah, that would have been slightly lower than Q3 profitability because of what I mentioned before. The new hiring, additional hiring completed at the end of Q3 and largely impacting also Q4, together with new depreciation and amortization. In terms of a normal run rate, we would have, let me say, similar revenue with similar capacity, but a little lower profitability.
Okay, perfect. Secondly, just on capacity, I know you've had a lot of questions, but I'm wondering what's the incremental capacity that the new facility in Milan brings, which is ramping in Q1 of next year? How much incremental sales capacity does that bring when it's fully ramped, and basically how many quarters does it take to ramp that to a full run rate?
The ramp rate will be pretty quick, because the new facility will include a complete assembly department, and that will be very, very efficient from a logistic point of view, internal logistic point of view. We will expect the increase of the ramp in the first quarter, let me say, to arrive at a rate of EUR 130 million revenue per month in the Q1. This is the run rate that we expect to reach very, very soon.
If you consider that, again, one year ago, we were at EUR 700 million per year, a little less, it means twice the capacity we had 12 months ago. This is exactly the multiple we have been able to reach, combining internal efficiency and new footprint.
Okay. Can I maybe just double-click on that? I mean, if I sort of annualize the Q3 run rate, you're sort of nearly at EUR 1.3 billion, roughly. I think, just correct me if I'm wrong, you're saying that the new facility would just bring you basically to EUR 1.4, so sort of an incremental sales of EUR 100 million. That feels a little bit low given that it's a fairly large facility that you've invested quite a significant amount of money in. Yeah, just maybe a clarification on that point.
Sorry, maybe I didn't follow exactly your calculation earlier, and I didn't write on my paper, so I couldn't follow exactly what you mentioned.
No, no. I was just wondering exactly how much incremental capacity that Milan facility brings, but I guess you answered it. I can go onto my third question, which is on the new products. I know you talked about ASICs and HBM. I'm just wondering on silicon photonics as well. Is there any updates on customer engagement, qualifications, or anything else that could be interesting on the silicon photonics side with Eclipse? Thanks, guys.
Yes, thank you for asking. Yes, silicon photonics, we are right now in qualification with three major players. We think to have a solution that is ready for HVM, so for high volume, rather than for a lab, okay? We expect to complete the qualification by mid of 2027, and then start with some production. Right now, we can't give you an estimate for the revenue because it's still early, but as soon as we're going to be close to complete the qualifications or already completed, we can give you an update for this.
Perfect. Thank you so much, guys.
Thank you, George. The next question comes from Oliver Wong. Please, Oliver, the floor to you.
Hi, everyone. It's Oliver Wong from Bank of America. Thank you for taking my questions. My first question is on revenues, specifically demand into next year. I think earlier this year, it was quite clear that you guys were capacity constrained. I understand that with the low visibilities, it's unclear if that will continue to be the dynamics for next year. I was just wondering for demand, I was wondering where sort of the uncertainty or sort of the variability stems from.
I understand a lot of your revenues pertain to GPUs. Is the uncertainty kind of stemming from sort of not being sure about GPU volumes for next year or perhaps customer inventories or perhaps something else? It'd be helpful to get more color there.
What we expect for 2027 is that GPUs and all the data center application will be largely impacting our P&L and our allocation. Currently, we exceeded largely more than 50% of our current production, fully engaged in data center, and within the data center, especially allocated to AI applications. We do expect for 2027 to reiterate this message, and we don't see any decline in that trend.
We have conversation with customers, and we have conversation also with new customers for new applications that we truly believe will become with large volumes in 2027, but what we didn't have in 2026 because of capacity constraints or other market share taken by our competitors. To answer to your question, the answer is yes. GPUs and ASICs and everything related to AI will be still the most important portion of our revenue.
Got it. Thank you. My other question is it fair to say that the ERP implementation, the sort of amortization costs on your margins will continue throughout 2027? Any color there?
Yes, but the impact of this depreciation will be almost negligible, because if you split them through, let me say, 10 years, the total cost of the system implementation would be not that relevant, impacting in the long term the figures, especially when you ramp revenues in that speed. We are not completely worried about that.
Okay, got it. Thank you very much.
Just to conclude, the depreciation of the prior ERP will be terminated. We will have, from next year, a saving from that side.
Got it.
Thank you, Oliver. The next question is from Alberto Gegra. Please, Alberto, the floor to you.
Thank you. Just a few follow-up. The first is on the number two supplier. Your competitor sounds more and more vocal on the possibility of gaining market share in 2027 in the GPU. Which kind of assumption are you making on the potential evolution of the market share? Then maybe a word on the consumer market, as we are seeing several outlook quite negative on 2027. Maybe if you can share your view. Lastly, on the industrial partner, you have this new shareholder that increased the stake. Maybe some comments on the strategic rationale behind it, their specific project. Anything will be helpful. Thank you.
Okay. About GPUs, I think was for us expected to have some competitors, of course. I mean, healthy market needs at least two suppliers. It is not a surprise, but we think, first of all, that the market will be big enough also for two suppliers, and we think to have the right technologies ready to address the next challenges for the GPUs. I would say that we are not worried about market share for GPUs. As far as consumer, it is still early to see what is going to happen in 2027. We see a slight recovery for the Q4 consumer market, but still early to speak about 2027 for consumer.
Yeah, about the industrial partnership with Vector Line, of course, this is a confirmation of the strategy of the group to take advantage of industrial partnerships like we did in the past, in the recent past with Teradyne and with Advantest. The strategy with Vector Line remained the same, having a more clear visibility on the ecosystem of the same space.
We are all involved in this, let's say, small market with very few players and connecting each other with concrete investments rather than commercial agreements, simple commercial agreement, will speed up the technological progress and customer relationships. This is what Vector Line will bring to us since the beginning. Additional commercial relationships and additional knowledge about the market or what we were missing until this partnership.
[Non-English content]
Thank you, Alberto. As there are no further questions queued, I will now hand back to the speakers for any final comments before bringing this presentation to a close. Please go ahead.
Thank you, everyone, for joining us today, and please stay tuned for the end-of-the-year results. Thank you.
This presentation will now come to a close. Thank you.