Good morning, everyone. Welcome to the conference call for the analyst and investor of Infineon's 2026 financial third quarter results. Today's call will be hosted by Alexander Foltin, Executive Vice President, Finance, Treasury, and Investor Relations at Infineon Technologies. As a reminder, this call is being recorded. This conference call contains forward-looking statements or assessment about the business financial condition, performance, and strategy of Infineon Group. These statements or assessment are based on assumption and management expectation reflecting upon currently available information and present estimates.
They are subject to a multiple and uncertainties of risk, many of which are particularly or entirely beyond Infineon's control. Infineon actual business development, financial condition, performance, and strategy may therefore differ materially from what it's discussed in the conference call. Beyond disclosure requirements stipulated by law, Infineon does not undertake any obligation to update forward-looking statement. At this time, I would like to turn the conference call over to Infineon. Please go ahead.
Many thanks, operator. Good morning, ladies and gentlemen. Thank you for joining our mid-summerly earnings call covering our fiscal third quarter 2026. On air today, you have our CEO, Jochen Hanebeck, our CFO, Sven Schneider, and our CMO, Andreas Urschitz. Following our well-established procedure, Jochen and Sven will provide an overview on the market situation and divisional performance, key financials, and our outlook.
After that, we will start our Q&A session. The accompanying slideshow for the call is available at infineon.com/slides, and we will provide a PDF with Jochen's and Sven's introductory remarks in the course of the call on our website, namely infineon.com/investor. This is also your go-to spot for a recording of this conference call, including the aforementioned slides, a copy of our earnings press release, as well as our investor presentation. Now, Jochen, over to you.
Thank you, Alexander. Good morning, everyone. At present, positive cyclical momentum and structural growth are converging into a bright picture. Infineon is generating value from it. The recovery continues to gain traction. The up cycle is fully on track. What initially started in selected segments is now becoming broad, supported by improving demand patterns, normalized inventory levels, and increasing order activity across many end markets.
The strongest dynamic continues to come from AI-related infrastructure. Investments in data centers continue to go up. Energy-efficient power delivery solutions are required to support ever-growing processing capabilities. Use cases for agentic and physical AI are emerging at a fast clip. Industrial markets are also showing improving dynamics, in particular related to power infrastructure. In automotive, we are seeing a clear pickup of customer order momentum.
In this environment, we are combining a strong operational focus on the current up cycle with targeted investments in our broad set of future growth opportunities. The prime example for the latter is the recent opening of our new smart power fab in Dresden. The clean room space available there enables us to ramp the world's largest fab for cutting-edge power semiconductors and analog mixed-signal technologies at just the right point in time.
Furthermore, we closed the acquisition of the sensor portfolio from ams OSRAM as planned within a very short timeframe. Let us now turn to our third quarter performance. The third quarter of our 2026 fiscal year was the first one with over EUR 4 billion of revenue for two and a half years. With EUR 4,172,000,000, we achieved an all-time high in quarterly revenues and came in a bit ahead of expectations, even considering a minor positive currency effect.
All our divisions contributed to 9.4% sequential growth on group level. Compared to the same quarter one year earlier, our reported revenue grew by close to 13%. The segment result for the June quarter amounted to EUR 797 million, corresponding to a segment margin of 19.1%, 200 basis points up from the quarter before, mainly driven by volume fall-through and positive mix effects, and clearly in the upper part of the predicted high teens range.
Our order backlog witnessed another material increase and stood close to EUR 30 billion at the end of June, a clear indication of recovery momentum getting even stronger. Now to our divisional review, beginning with automotive. The third quarter of our 2026 fiscal year, revenues increased by 6% quarter-over-quarter to EUR 1,932,000,000. Growth was driven mainly by microcontrollers and smart power components, as well as our Ethernet products.
All these are core building blocks of software-defined vehicles. The segment result increased by 8% sequentially to EUR 356 million, corresponding to a segment result margin of 18.4%. As a reminder, the refocusing of our business with high voltage components for electric powertrains that we explained in our last earnings call is expected to burden segment result margin of ATV in this fiscal year by a low to mid-single digit percentage reflected in our guidance.
We continue to see strong order intake in automotive against the backdrop of a muted car market. In its latest update, market researcher Mobility Global, carved out from S&P Global, is forecasting around 91 million light vehicles to be produced in 2026. This is slightly above its previous estimate and broadly in line with the approximately 2% year-over-year decline we had assumed since the beginning of our fiscal year.
The long-term trends driving automotive semiconductors demand remain firmly intact and continue to support content growth. Rising fuel costs are beginning to support xEV adoption in Europe, India, and Southeast Asia. While the shift towards software-defined vehicles continues to accelerate globally. Along side this structure growth drivers supply constraints in Chinese Automotive great semiconductor market market creates opportunity for us. Furthermore, ongoing inventory replenishment is contributing to near-term demand recovery. We have secured a major design win for zone controller architecture with a software company of a leading global car manufacturer.
The solution combines latest generation AURIX microcontrollers, power management ICs, and CoolFET smart power switches. This demonstrates again the breadth of our system offering across compute connectivity and smart power management. We are also seeing further design win momentum in China. For Xiaomi, we will support a cockpit and ADAS fusion unit incorporating three different microcontroller families.
In another win with a leading Chinese car manufacturer, our silicon carbide bare dies will be used in the traction inverter. Finally, we further simplify the evaluation of our automotive microcontrollers for our customers. Together with Amazon Web Services, we have launched a cloud-based platform for virtual MCU evaluation. By removing the dependency on physical hardware, the platform can shorten evaluation cycles from several weeks to minutes, lower evaluation costs significantly, and support hundreds of concurrent users globally.
The platform already includes our next-generation RISC-V architecture, enabling customers to gain hands-on experience with new microcontrollers much earlier in their development cycle, and further accelerates innovation for software-defined vehicles. Let's now take a look at Green Industrial Power. GIP's revenues grew by 11% quarter-over-quarter to EUR 447 million , making the June quarter the second one in a row with double-digit growth.
Reflecting the recovery on industrial markets, all application areas developed positively, in particular Power Infrastructure and HVAC. The higher revenue notwithstanding, the segment result of GIP contracted slightly to EUR 44 million , equivalent to a segment result margin of 9.8% after 11.7% in the quarter before. The decline was due to temporary operational and inventory-related effects, and hence not indicative of underlying profitability, as will be evidenced by the positive margin evolution in the running quarter.
Power Infrastructure is seeing strong structural momentum. Investments in grid expansion modernization continue to grow, driving demand for Energy Storage Systems, transmission and distribution gear, and high-voltage solid-state devices. AI data centers growth is fueling demand for uninterruptible power systems, general power supply, as well as cooling. Semiconductors are poised to replace electromechanical parts in various use cases.
For example, a semiconductor-based solid-state circuit breaker can protect electrical circuits from damage caused by short circuits or overloads by up to 1,000 times faster than conventional systems. This capability is essential for direct current grids and offers a significant increase in system availability in industrial manufacturing and AI data centers. As part of our partnership with Siemens, we will supply 1.2 kV silicon carbide power modules for use in circuit breakers to enhance the efficiency, power density, and reliability of Siemens' protection solution.
Now to Power and Sensor Systems. On the back of unabated AI power strength, PSS recorded revenues of EUR 1,442,000,000 in the June quarter, 14% up sequentially and a staggering 34% more compared to the same quarter one year ago. The margin evolution of PSS shows a bright picture as well. The segment result increased to EUR 359 million , corresponding to a segment result margin of 24.9%. The further increase of 4.5 percentage points compared to the previous quarter is evidence of profitable growth and value creation, strongly driven by our leadership position in AI power solutions.
This leadership position is being recognized by industry researchers. In a recent report covering AI data center power semiconductors, Gartner identified Infineon as the company to beat. Portfolio breadth and system-level expertise in conjunction with manufacturing capacity are the defining capabilities for data center operators seeking to scale AI. Infineon offers a unique portfolio spanning the entire power delivery chain and manufactures relevant technologies in-house.
By seamlessly integrating wide bandgap materials, specifically silicon carbide for high efficiency, high voltage grid-to-rack conversions and gallium nitride for ultra-dense high-frequency intermediate power stages alongside silicon at the processor level, energy losses are minimized at every single conversion step. Demand for our AI power solution continues to outstrip available supply. We are in allocation. Successful execution of capacity ramps and conversions from other areas will help us to achieve more than EUR 1.6 billion of dedicated AI power revenues in the current fiscal year, ahead of the so-far planned EUR 1.5 billion.
In addition, our business with non-AI data center power solutions is amounting to around EUR 500 million annually, making Infineon clearly the leading force in the overall space. We are represented in almost all platforms across all relevant players in the industry. Enabling further steep growth in the coming years will be a function of ramping and deploying additional manufacturing capacities. Our new smart power fab in Dresden, together with available clean room space at our other large front-end sites in Austria and Malaysia, puts us in a unique position.
To strategically secure access to critical power delivery solutions, several leading customers across the AI data center ecosystems have signed or are in negotiation on multi-year capacity reservation agreements with us. These agreements encompass a total cumulative sales volume of a high single-digit billion euro amount over multiple years. These agreements also feature certain prepayments, thus further strengthening our customer relationships and sharing investment risk. We will revise our projection of EUR 2.5 billion+ AI data center revenues for our 2027 fiscal year upwards as part of the annual guidance to be given in our November earnings call.
We expect such update to be material. The next waves of AI growth are already taking shape. Higher density power architectures for running the last frontier models, as well as the emergence of agentic and physical AI. The growing inference and task coordination requirements of agentic AI provide a massive tailwind for us. Our undisputed leadership in power solutions for CPUs, combined with a highly differentiated best fit product portfolio, will represent another significant growth driver as early as next fiscal year.
Accordingly, the aforementioned update of our revenue projection for 2027 will also include our power solutions for all different forms of data center configurations. Beyond the data center, our solutions bring physical AI to life, enabling humanoid robots, collaborative machines, and autonomous systems to perceive, think, and act safely and securely. With expertise from all our divisions spanning microcontrollers, power systems, sensing, connectivity, functional safety, and security, Infineon is the trusted partner across the full spectrum of physical AI platforms. This provides a good transition to complete the divisional review with Connected Secure Systems.
CSS recorded revenues of EUR 350 million in our fiscal third quarter, a sequential growth of 10%. All product areas contributed to this positive development, in particular authentication and identification solutions. The segment result of CSS increased to EUR 34 million, corresponding to a segment result margin of 9.7%. We are continuously seeing a broader adoption of AI at the Edge for industrial as well as consumer applications. To optimally address the expanding opportunity set, we are establishing the Edge Systems, or ES division, as part of the new organizational setup presented in our last earnings call.
The ES segment is formed from today's CSS and the sensor and RF, as well as the USB connectivity portfolio from PSS. The focus of ES will be on the interplay of sensors, microcontrollers, including software, connectivity, and security to enable integrated system-level solutions at the edge. In this context, I am happy to report that we have closed the acquisition of the non-optical analog mixed signal sensor portfolio from ams OSRAM at the beginning of July, just about five months after announcing it. The transaction is strengthening our position as a leader in sensors for automotive and industrial markets through a complementary portfolio and extending our product range in medical applications.
The acquired business has a current annual revenue run rate of around EUR 230 million and will support Infineon's profitable growth. The transaction is accretive to adjusted earnings per share right upon closing, with future synergies enabling substantial additional value creation. The acquired portfolio is a strong strategic fit to our ES division. Now to Sven for our key financial figures.
Thank you, Jochen, good morning, everyone. In line with the revenue increase, our June quarter saw a corresponding margin expansion. The reported gross margin went up by 210 basis points from 38.7%- 40.8%. The adjusted gross margin stepped up from 41%- 42.8%, mainly driven by higher volumes and positive mix effects. Also, pricing measures put in place from April onwards had a first positive effect. Annual merit increases kicking in from the beginning of April, as well as rising input costs for precious metals and logistics, were dampening margin progression to some extent. Idle costs were roughly in the same level as one quarter before.
Likewise, similar to the previous quarter, refocusing our high voltage automotive drivetrain business had a negative impact of about 1 percentage point on group level. Research and development expenses increased quarter-over-quarter from EUR 612 million- EUR 674 million. Selling general and administrative expenses went from EUR 379 million- EUR 433 million. Non-segment result charges for fiscal third quarter amounted to EUR 203 million after EUR 195 million before. The financial result amounted to EUR -63 million after EUR -68 million in the prior quarter.
Income tax expense for the June quarter was EUR 112 million, equivalent to an effective tax rate of 21%. Cash taxes amounted to EUR 220 million, adjusting for PPA effects, the quarterly cash tax rate stood at 19%. Our investments in property, plant, and equipment, other intangible assets, and capitalized development costs amounted to EUR 514 million in the quarter under report after EUR 541 million in the quarter before. Depreciation and amortization expenses, including acquisition-related non-segment result effects, amounted to EUR 466 million.
Free cash flow in the third quarter of our 2026 fiscal year improved significantly from EUR -63 million- EUR +599 million, to a large extent driven by the higher operating result. On the working capital side, inventory reach went down by 10 days from 175-165 days quarter-over-quarter, bringing us closer to our target for the end of the fiscal year of around 150 days. As stated before, slightly elevated inventory levels are helping us capture growth in the current upcycle and being prepared in case of geopolitically induced turmoil.
To our liquidity and leverage situation. During the June quarter, we made two scheduled debt repayments, redeeming a EUR 700 million Eurobond and $ 350 million private placement at maturity. The related cash outs were partially offset by the positive free cash flow. As a result, our gross cash position stood at around EUR 1.7 billion. Gross debt amounted to around EUR 6.8 billion as of June 30th. Therewith, our gross leverage of 1.8x is already back below our maximum target level of 2x .
The acquisition of the sensor portfolio from ams OSRAM, which closed on July 1st, has no impact on this figure as it had already been pre-funded. The net debt position of around EUR 5.2 billion at the end of June corresponded to a net leverage of 1.4x . Our after-tax reported return on capital employed was trending up to 7.9% for the third quarter of our 2026 fiscal year. Before handing back to Jochen, a brief comment. As you know, we are operating in a new three divisional organizational structure with automotive, power systems, and Edge Systems from July 1st onwards.
We will provide you with key financials for the known four-division setup also for the September quarter, allowing you to model the full 2026 fiscal year in a consistent way. In our November earnings call, we will give our outlook for 2027 in the new setup and at the same time provide adjusted historical data to you to enable relevant comparisons. Back to Jochen, who will elaborate on our outlook.
Thank you, Sven. The market environment for our outlook is turning increasingly favorable. The upcycle is gaining momentum. The recovery is broadening. Inventories are largely destocked across markets, Market is becoming a relevant factor for growth. Stronger customer bookings are leading to a further growing backlog and improving visibility. Business indicators like cancellation rates, escalation calls, or the preparedness to accept surcharges for expedite deliveries are speaking to underlying demand strength. Lead times are rising, According to third-party surveys, customers expect them to rise further over the next months, a typical early upcycle pattern as supply starts to tighten. Needless to say, the dynamic differs across applications.
At this point, AI, power, and industrial infrastructure are appearing the strongest, followed by automotive, whereas consumer is lagging. Generally speaking, though, our outlook is framed by high confidence, bearing further escalation of geopolitical conflicts. For the currently running September quarter, last of our 2026 fiscal year, we are in line with recent currency developments, adjusting the U.S. dollar-euro exchange rate back to $1.15. We expect revenues of around EUR 4.7 billion, corresponding to well above seasonal growth of almost 13% quarter-over-quarter and 19% year-over-year. By segment, for ATV, a moderate revenue growth is predicted, whereas for each of GIP, PSS, and CSS, revenue should grow significantly.
The sensor business acquired from ams OSRAM should contribute a mid-double-digit million euro figure in revenue in this quarter. We expect the final fiscal quarter segment result margin to come in around 23%, 4 percentage points up quarter-over-quarter. Besides the fall-through from higher volumes, we expect the positive evolution of prices in certain areas, notably AI and related product categories, to have a positive impact, offset partly by further rising input costs, which we'll address in customer negotiations.
For the full 2026 fiscal year, we now expect revenues to come in at around EUR 16.3 billion, equivalent to an annual growth of around 11%. From a segment perspective, PSS is poised to grow materially faster than this group average, driven by buoyant demand for AI power solutions. As the outlook for some industrial markets linked to AI continue to improve, GIP is now expected to grow just below group average. ATV should see slight revenue growth driven by its broad product portfolio and the broader adoption of software-defined vehicles, despite material headwinds from the high voltage drivetrain business.
For CSS, we expect revenues to remain stable year-over-year. Regarding profitability, we confirm our expectation for the full-year adjusted gross margin to reach a low to mid-40s level and for the segment result margin to land at a level around 20%. Idle costs are projected to amount to an annual level of around EUR 650 million. In our forecast, we have not included potential indirect effects from further escalating Middle East conflict or any other lingering geopolitical tension. Our investments in the fiscal year continue to be expected to come to around EUR 2.7 billion, and for depreciation and amortization, we anticipate an unchanged level of EUR 2 billion, including amortization of around EUR 400 million, resulting from purchase price allocations, which will be recognized in our non-segment result.
Based on our favorable business outlook, we are upgrading our projection for the adjusted free cash flow. The figure, which net of investments into major front-end buildings and M&A transaction, is now expected to come at around EUR 1.85 billion after EUR 1.65 billion before, corresponding to around 11% of group revenues. For the reported free cash flow, we are changing our guidance now from around EUR 1.25 billion to around EUR 900 million.
This is an implicit upgrade as well, given we are now for the first time considering the purchase price for the sensor portfolio acquired from ams OSRAM of about EUR 570 million, the impact of which is partly offset by the improved underlying cash flow. Ladies and gentlemen, before going into Q&A, let me summarize. The upcycle is clearly gathering steam and end market strength is broadening. Structural growth drivers are proving to be very strong. AI momentum is unabated and Infineon is at the forefront of powering AI from grid to core.
Our AI-related revenue will more than double this fiscal year and exceed EUR 1.6 billion in addition to the EUR 500 million of classic data center power. Our unrivaled portfolio, coupled with significant additional clean room space, will propel growth. Expect a material upgrade of our prediction for 2027 in our November call. A significant portion of our future data center revenue is going to be covered by customers signing CRAs with us. Automotive is further improving, driven by structural content growth, share gains, and inventory replenishment. Our third fiscal quarter came in fully in line with our guidance.
Based on a bright business outlook, we expect a much better than seasonal Q4, 13% sequential growth to EUR 4.7 billion, the segment result margin expanding 400 basis points quarter-over-quarter, a strong finish of our 2026 fiscal year, and a good point of departure for 2027. The opening of Dresden Four and the closing of the acquisition of ams OSRAM sensor portfolio are proof points of how we are actively positioning Infineon to capture profitable growth and create value. Operator, please start the Q&A session.
Thank you, sir. Our question- and- answer session will be conducted electronically. If you would like to ask a question, simply press the star key followed by one on your telephone. If you are joining us today using a speakerphone, please ensure that your mute function is turned off. Now we will take our first question coming from Sandeep Deshpande from JPMorgan. Please go ahead.
Yeah. Hi. Thanks for letting me on. I have two questions, if I may. Firstly, regarding your guidance on the margin in the fourth quarter. When you had guided in the prior quarter, the implicit margin was guided already for the fourth quarter. Has something changed from when you guided in the third quarter in terms of your cost structure or any other costs?
Secondly, can you quantify how the price increases are playing through on the margin into the fourth quarter? I have a quick follow-up on the AI-related LTAs that you have signed. In terms of the AI-related LTAs, is there a particular target that the company has in terms of how much of the capacity that you are going to outlay over the next three years that you've talked about in the past will be signed up in terms of these LTAs? Thank you.
Yeah. Good morning, Sandeep. Thank you for your question. I take the first one, and then I give it to Jochen on AI. On your question regarding the margin in Q4 implicit guidance, any change from Q3? The answer very easily is no. There is no material change. We had baked in a couple of positive effects from the price increases, but they are mainly contributing to next year, and we have also baked into our forecast already the cost increases, which we are seeing with regard to the geopolitical situation around the Middle East. No major change on that end.
Thank you.
The next question comes from Johannes Schaller from Deutsche Bank.
Sorry, I haven't answered the second part of Sandeep's question. My microphone was muted. Coming to your question, Sandeep, we would feel comfortable to target a high number. We evaluate the discussions ongoing with the customers. From our point of view, this is a win-win situation for us. It's de-risking, on the one hand, for the customer, it's security of supply, and very important, the prices for under the CRA are not fixed. It's a volume commitment, but not a price commitment. Prices will develop along market price.
Thank you.
We take now the question from Johannes Schaller from Deutsche Bank. Please go ahead.
Good morning. Thanks for taking my question. Just on the situation now with these CRAs and also specifically on AI, could you comment a little bit on what you see in stage two with your processor customers? There is a bit of a concern in the market that new players are coming in here. The space will become a bit more competitive. Maybe help us understand how confident you are on your market share outlook with the processor players and the revenue opportunity in euros also.
Does new competition even matter given that the industry is so capacity constrained? As a second question, Jochen, you mentioned some supply constraints in the China auto semi space that could create some opportunities for you. I think that's quite different to, let's say, the market view out there, which is more that this space is extremely competitive and there's a lot of supply. Maybe help us zoom in a little bit on that and give us some more color here. Thank you.
Yeah. Thank you, Johannes. If I may, I would like to ask which new processor companies do you have in mind? Do you talking about the ASICS ones, or which ones are you talking about?
No, sorry. New entrants on stage two power supply.
From a competition point of view.
Not new.
Competitors of ours.
Yes, exactly.
Got it. Yeah. Competitor of ours. Obviously, many companies are seeing this big opportunity, and are trying to enter. The hurdles are very high, right? It's about a proven track record in terms of power stages. It's having the combination always with a power controller. Right now, again, they are trying, but as we speak, we clearly see the typical incumbents in that market. In the power stage two, not really new competitors gaining significant business.
On the China auto side, yeah, I agree. It sounds at first counterintuitive, but what we do see right now, particularly in MOSFET and analog parts, is that Chinese competitors cannot deliver because they are, to a good extent, linked up to foundries, and these foundries in China are shifting their supply also partially to AI. There are also, again and again, quality missteps. Here we are seeing a good opportunity to pick up more business. Of course, we will not do it only for a short run, but we want to have then a midterm commitment from customers.
In terms of products, this is, maybe let's call it also including some legacy applications, if you want to call this that.
It's MOSFET, so not the IGBT side. The IGBT side is still difficult, as I explained last time. It's MOSFET and analog parts. These analog power parts, they go also into power distribution for software-defined vehicle. I wouldn't call it legacy. I would call it broad-based products that find multiple applications in the car. New ones and established applications. Very broad.
That's very clear. Thank you, Jochen. Very helpful.
The next question comes from Lee Simpson from Morgan Stanley. Please go ahead.
Great. Thanks for fitting me in and well done on the sales guide. Maybe if I just go back to question on margins, if I could. I'm just trying to understand the 23% guide that you've given us for Q3, where the moving parts are here, because we do have a sense that there's some Dresden startup costs t o be absorbed. There's probably underutilization.
Really, I guess we're trying to work out what the utilization rates could be, and how does this affect the run rates for that margin structure, really importantly, as we go into next year as well. Maybe my follow-up, if I could, just trying to understand the nature of the follow-on discussions around your LTAs, the new customers beyond those who have committed, maybe the size and scale there, and equally the size of prepayments that you're getting. Is this EUR 1 billion or EUR 2 billion, or is it a bigger number for the prepayments? Thanks.
Okay. Hi, Lee. I take your question. It's not totally surprising to ask the question about the margin in Q4. Let me help you here a bit. First of all, one thing you mentioned, Dresden module four ramp-up cost, no material inclusion in this year. That's next year. The situation is as follows. If you look at the last quarters, it's a very dynamic growth environment we are in. EUR 4.2 billion now going to EUR 4.7 billion. You also see it on the margin. We were in the high teens territory. Now we go to the mid-20s territory. Of course, in such a dynamic situation, you need to work with a set of assumptions. On the assumptions, I think you all know us pretty well.
We tend to be a bit more on the conservative side of things in order not to underdeliver. There are assumptions on growth. There are assumptions on the cost and price, as I just mentioned. There are, of course, assumptions on the balance. We have talked about it in the last calls, about managing the inventories and loading the fabs in the best possible way in a situation where more and more products are going into allocation. All that is included. If you ask me, why is it only 23%, I would probably say that the 23% are a tad conservative, and it could be 23%+ .
Let's look at the quarter. How is it really coming in? There is a better fall through than 50% included in the Q4. Let's see how much we will then really ship at the end of the quarter, also revenue-wise, and take it from there. One last sentence. You are asking the question not because you are so interested about Q4. You are asking the question, as you said, because you are very interested in hearing the first news on 2027. Maybe this is also part why we guide as we guide.
Good. Then I will take the opportunity on the CRAs to make some broader statements, because I guess others also have questions on that one. First of all, for the CRAs, the target customers are hyperscalers, AI processor makers, as well as hardware providers for data centers, covering therefore stage one and stage two products. We are currently in discussion or have finalized agreements with more than 10 customers. The essence of these agreements is that customer to commit to offtaking certain volumes over a period of several years.
Penalty clauses ensure that customers have skin in the game. Key benefits for Infineon, obviously, customer commitments underpin and de-risk of AI capacities. The agreements are based on existing and already planned capacities. As such, they are firming up Infineon's projection of AI power growth. The CRAs are obviously not in the backlog. Again, prices are not fixed. On the prepayments, I cannot comment at this moment in time, as we are still in negotiations with our customers. Hope that clarifies-
Right.
...a couple of questions around the CRAs.
That's very clear. Thank you very much.
The next question comes from Didier Scemama from Bank of America. Please go ahead.
Yes, thank you for taking my questions. Good morning, gentlemen. I've got an additional question maybe on the CRAs. First of all, I think it's really interesting that these developments are taking place a bit like we've seen in memory. I just wanted to understand a couple of things from you, Jochen, if possible. What's the embedded assumption on pricing?
I know that the pricing is sort of determined by the supply and demand in the future, but obviously in your high single digit billion, you've got an assumption on pricing. Should we assume flattish from here or down or up? That would be an interesting one for us to understand. Related to that, do you expect that high single digit billion number to be revised higher in the coming quarters as you finalize your negotiation with other customers? Thank you.
Okay, the first part, Andreas will take, and then I will answer the second.
First and foremost, related to pricing and Infineon pricing overall. Already in our last earnings call, I talked about supply constraints, in particular in our AI power business, and visible impacts on adjacent areas, leading to more favorable pricing environment. I also said we will adjust pricing to reflect market realities. Meanwhile, demand was outstripping and still continues to outstrip supply in several areas. No longer related to the entire AI power delivery chain only.
Thus, and walking the talk, we have informed our customers in these areas about price increases two times, and the last time happened in July. We are seeing good traction for these measures overall as customers value delivery capabilities in today's situation, and the value we create with our products in the marketplace, a very decisive role. Full visibility of this in our P&L and in our margin, as Sven was alluding to, however, will be there only from quarter one next fiscal year onwards. I give it back to Jochen-
Thanks, Andreas.
...specifically on this area.
Thanks, Andreas. Didier, on that question, let's say the momentum is on our side. With respect to your question on the high single-digit billion number, the momentum is on our side. I would not be surprised if that number is increasing, but we work now with those customers that have raised their hands and are interested, and then we will update you in the next quarter again.
The next question comes from Joshua Buchalter from TD Cowen. Please go ahead.
Hey, guys. Thank you for taking my question, congrats on the results. I'll also use German efficiency and ask two questions at once. I guess to start, can you speak to how we should think about the linearity and maybe magnitude of the Dresden capacity coming online over the next few years? Was that 50% fall-through number specifically related to Dresden, or should we think about all incremental revenue falling through at 50% moving forward? My second question: there's certainly been a lot of noise in both directions about 800 V architectures in data center. Could you maybe speak to how either a slower or faster 800 V adoption would change your AI content opportunities? Thank you.
Yeah. Hi, Josh. I take the first and the third, and fall-through, I leave it to Sven. Dresden, as we said at the beginning when we embarked on that project, filling the fab would take three to seven years. We are clearly now at the lower end of that range, so likely if the market demand is there, we can ramp at double the speed as before, and that would take us below three years to complete the build-out with equipment. On the 800 V, I think it's known in the industry that 800 V DC is nothing to play around with, so there are some safety concerns.
We do not see any material shift, but in case it comes, our bill of material, if I compare 800 V DC to three-phase AC is incremental on the 800 V DC side, but not tremendously. For us, it's not the biggest factor in our growth trajectory as we are very well positioned on the PSU side. I think the question was whether the fall-through only applies to the Dresden facility revenue or altogether, and I think you always commented on overall numbers not related to individual fabs.
Correct. Yeah. Thank you.
The next question comes from Jakob Bluestone from BNP Paribas. Please go ahead.
Hi. Good morning. Thanks for taking the question. I've got two questions as well. Firstly, just on the AI revenue guide, you've obviously put through a small hike for this year. I just want to understand, is that from the early impacts of the price hikes starting to feed through and therefore driving your guidance, or are you actually already starting to see supply coming through more quickly than expected?
Just secondly, on your order book, you obviously had a very strong backlog, EUR 30 billion at the end of the quarter. I think it was up EUR 5 billion Q- on- Q. Could you maybe just give us a little bit of color on how much of that relates to some of the inventory builds you mentioned? How much would you put more in the, I guess, more structural long-term camp? Thanks.
Yeah. The second question, I'm not quite sure whether I got it. The first one, the EUR 1.6 billion, again, we said above EUR 1.6 billion, it's now an operational task to squeeze out as much as possible in the quarter to serve the demand of our customers. It's a mix of supply and pricing, but I would say for this quarter, it's more on supply side. Again, let's see how the final number looks like, but we feel comfortable to overshoot or deliver more than the EUR 1.6 billion. Please always keep in mind, you need to add these EUR 500 million to it. The order backlog, I give it to Sven.
Yeah. Hi, Jakob. As far as I understand it, your question on the order backlog, you're asking how much of the order backlog increase is inventory build or more structural at customers. That's how I understood. Thank you for confirming. The backlog increase is driven mainly by, now again, I'm in the old divisional structure, GIP, PSS, and Automotive. If you go through these divisions, also following what Jochen has said in the intro, I think there's a really nice pickup in customer order entry on the GIP side. PSS, I don't think I need to comment. AI and most products in Allocation and Automotive, we said it in the intro, it's a combination of some market developments in China, some nice order momentum around the software-defined vehicles, and some order replenishment. It's a combination.
Very clear. Thank you.
The next question comes from Francois-Xavier Bouvignies from UBS. Please go ahead, Francois.
Thank you very much. I have two quick questions. The first one is on the seasonality in fiscal Q1. I think, Sven, last quarter, you said that you would expect fiscal Q1 to be, well, less pronounced in terms of seasonality, which is -5% to -6% quarter-on-quarter. That's what you said last quarter. I just want to check how you feel about this comment now, and if you would think that it could even grow quarter-on-quarter for the fiscal Q1, given the current dynamic.
Yeah, maybe I'd take that question, Francois-Xavier. We clearly see a very much different seasonal pattern for Q1. Let's see how it really comes out. Forget the seasonality we have seen in the past. Today, it looks rather as a very strong second half of the calendar year. Therefore, also please understand that our projections coming out of two, three quarters around high teens now coming into different territory have, of course, also some uncertainties, which we have to make certain calls on. I think we will see over the next two quarters where this new level of profitability will exactly land.
Makes sense. Thank you. Maybe my second question is on actually microcontrollers. Infineon did a very good job in terms of market share in the last three years. Now, if I look at the current dynamic, Renesas and NXP are growing their auto revenues by a mid to high teens percentage year-on-year. That's your main competitor in microcontrollers, which seems to be higher than what you are doing right now. I was wondering, in terms of microcontroller dynamic, is there any mechanism or reason maybe you would temporarily lose share because of some inventories or anything we should be aware on your market share on the microcontroller side that maybe would explain a lower growth from the automotive versus your MCU piece?
Yeah. Particularly on the MCUs, I always said it now for the last two years, that likely for the next two years, there will be still market share gains. What you observe is a different effect, but I can explain it in the easiest way. If you take the Automotive division and you take high voltage out and, to be fair, also the newly acquired Ethernet business.
That business without high voltage, without Marvell, would grow at constant currency exchange rate, 10%. I think we are rock solid in our core automotive business, having additional growth momentum now with Ethernet due to software-defined vehicles. We have a construction site called High Voltage, which we are resetting, refocusing on the profitable topics on innovations, while at the same time reallocate these capacities towards powering AI.
Great. Thank you for your answers.
The next question comes from Stéphane Houri from Oddo BHF. Please go ahead.
I have also two questions. The first one is about the Power AI revenue view for 2027. I think in another call you said you would be materially above the EUR 2.5 billion. My question is to know why you are not upgrading it today and to know what you will know better in November than now. Is it about the real level of demand, or is it about the ability to ramp the production? I have got a follow-up. Thank you.
Thanks for the question. I know there is high interest in that number, but first of all, of course, we are closer to the market. We have to give then a yearly guidance, and I would not like to go into one and a half yearly guidance. We are closer to the market. We are closer to customer development. We are closer to our operational insights, how much we can really deliver. Last but not least, we would like to give you a full set of financial numbers, CapEx, cash flow, all of that. Picking out one number, even though there is high interest, we would like to refrain from.
Okay, understand. The question is about the green industrial power issue that you had. You said it was a temporary operational and inventory-related effect. Are you saying that the next quarter, the margin will be back to normal already? Thank you.
Yeah. Stéphane, you can always say in this environment what is normal. I would say it will definitely go up materially compared to Q3. It will show a very positive trend, to answer your question.
Okay. Very clear. Thank you.
The opportunity for GIP or soon part of PSS is really great. If you think about this power infrastructure, right? Its ESS and SST and SSCB combined. It's today a market of a low to mid triple-digit million number. We expect that market already to be beginning of the next decade, a mid single-digit billion market. A great opportunity and really falling into our hands because high reliability requirements there, playing to the strengths of Infineon. The GIP business, particularly the power infrastructure part of it, will be a strong pillar of the PS division in the future.
All right.
The next question comes from Adithya Metuku from HSBC. Please go ahead.
Yeah. Good morning, guys. Thank you for squeezing me in. Two questions, please. Firstly, just, Sven, on the backlog. It's already at EUR 30 billion. Would it be fair to assume that roughly 2/3 of this is for delivery next year? Any color you can give around the proportion of the backlog for delivery next year? For Jochen, on GaN, there's been some recent talk about GaN adoption in stage two, closer to the XPU, potentially replacing silicon quicker than what people expected maybe three to six months ago. I just wanted to hear your thoughts on what you are seeing here. Are you seeing a quicker transition to GaN than maybe six months ago? Thank you.
Yeah. Thanks for the question. I take the second one first. We have seen the first emergence of GaN in PSUs, right? We talked about that a couple of times already. It's actual, it's fact, it's everyday business. The next big opportunity is the IBC level 48- 12, where there's a clear value proposition, and various customers are evaluating this, and we expect business to pick up in 2027. Now, you are talking about the third opportunity for GaN, and that's the low voltage GaN right in the power stage. Yes, that's technically clearly our goal, but the technical hurdles are also significant.
We are very well prepared in the sense of manufacturing footprint for this, as it would be also part of our 300 mm footprint in GaN. Here probably being the only one, being able to deliver really volumes. The technical hurdles to overcome in the power stage are high, I would like to bridge to an earlier question, how are newcomers doing in the stage two? Here, the same applies, that changes in the stage two, you really need to know what you're doing, and therefore, it will take some learning cycles, probably also first starting with some low volume business in none of the main platforms. Power stages is a piece of art in terms of power electronics.
Adithya, I take your other question on the backlog. The 2/3 assumption as of today seems to be reasonable to me.
Got it. Excellent. Just quickly, Jochen, essentially, would you say that development on GaN has accelerated closer to the XPU in the last six months, or there's no big change there?
We are definitely accelerating. Again, it will take several years before you see it in the market given the technical challenges here.
Got it. Thank you.
The next question comes from Tammy Qiu from Berenberg. Please go ahead.
Hi. Thank you for taking my question. The first question is on your LTAs. Are you eventually targeting to have 100% of your [AR] revenue on the LTA? Based on the terms, it wasn't that clear that if it's cancelable or uncancellable because pricing is not fixed, it's just basically about volume. Secondly, I have a question on your microcontroller position in China. We do hear that, for example, given the pricing pressure, there has been more consideration of taking local supply. Do you see that becoming a more discussed point among your customers, or you are very confident in your market share? Thank you.
Okay. Let me talk about first the CRAs again. These agreements are not rigid take-or-pay contracts, or they are not NCNR. They encompass the different product groups and include some reasonable flexibilities. Again, future prices are not fixed in advance. We have not defined an exact target in terms of coverage of the capacity, but I would feel comfortable with a high coverage given the dynamics. Again, it's also at the advantage of the customers to get security of supply. With respect to automotive micros in China, we are still clearly the market leader.
Of course, there are also local competitors very bluntly trying to copy our products, so far not successful. Our major competitors are still the well-known global ones, but we have a very strong franchise. We have a very strong, high portfolio density in China, which customers value because it doesn't make much sense to offer a customer a single microcontroller. You need to show a full product family.
That we have seen also in various design wins recently where non-Chinese competitors try to offer then individual microcontrollers, but customers need a portfolio. I think portfolio density is, besides the individual technical lead, a very strong argument in this place. Of course, we are staying paranoid and we are very carefully evaluating what competition is coming up in China as well as elsewhere.
The next question comes from Veysel Taze from Metzler. Please go ahead.
Yes, good morning. Thank you for taking my question. Related to your AI power business and specifically really on the part with server level or board level opportunity. On the technology side, are you fully betting on GaN, or would you see other opportunities or other technologies as well? Related to this as well, what is the difference or your positioning regarding GPU vendors and hyperscalers internal XPU or ASIC programs? Is there a difference in your exposure to these both camps?
This is Andreas. Thank you very much for the question. Regarding AI power stage one and two, are we fully betting on GaN or others? Well, as a matter of fact, over the course of the last years, we put ourselves in the position of having a very large portfolio of silicon-based solutions, silicon carbide-based, and also gallium nitride-based solutions for both the areas, power stage one and two, including the PSU, which is not part of your question, but just for completion, I tell that. We see ourself being in a unique position in terms of having then also reliability and experience for having the products robust, which our customers love a lot nowadays, since everything is about uptime in these server farms and AI and hyperscaler environment.
Another element for sure is then also the broad variety of interconnect, so to say, technologies that allows very great combinations in between controllers, drivers, and also FETs, be it in gallium nitride, be it in silicon carbide. The beauty of this positioning that we do have is that we let our customers choose. This we believe very much, and that's also the feedback of the market, that together with our unprecedented capacity, so to say, offering, which also plays a decisive role, makes customers go for Infineon as the primary choice along the entire power flow from what we call the grid, even towards the core, which is the stage two.
Got it. The second part of the question, is there a difference in your exposure to the GPU vendors or merchant processor makers versus the hyperscalers internal programs? Is there a difference in your positioning?
Again, Andreas speaking, I take this question. Look, the way how this industry is working is that us as primary suppliers of any kind of semiconductor solution for grid-to-core power for powering AI data centers is pretty much based on working with all the value chain players, starting from the processor maker. These are companies that are mostly, but not only residing in the West, in the U.S. We are then collaborating in parallel with what we call OEMs or data center operators. Those are companies that entertain large hyperscaler farms or AI machine learning data centers as such. They are pretty much decisive for then overall power flow architectures.
Thirdly, we're working together with what we call hardware makers or ODMs, who typically reside in the East and make sub-components such as intermediate bus converters or PSUs on behalf of the so-called OEMs or data center operators as such. By bringing all these elements together, so talking to the processor maker, the data center operator, i.e., the architect of this environment, plus then the sub-component makers, this gives us an ideal position in order to provide tailored power flow solutions. We call it power flow or from grid to core, was the word I was saying before, which is unique in terms of, so to say then, at the very end, cost per compute power, which AI and also hyperscale data center is all about.
Thank you, Andreas.
We finally take Didier Scemama for a rounding off question. Please go ahead.
Thank you for taking my follow-up. I just had another question on the CRA, perhaps. I wanted to understand a little bit the sort of T&Cs on the CRAs. First of all, is it the hyperscalers or the system builders or the processor, I mean, your customers that are asking to sign those CRAs, or is it you trying to enforce it? Related to that, I would assume that part of the sort of CRA commitments you're making, you've got certain capacity addition to make. Up against that, are you seeing visibility through, say, I don't know, 2028 from your customers against those capacity commitments? It would be great if we could get a bit of color on this.
Sure, Didier. Happy to take a second round from you. Honestly, customers are approaching us because they sense that there is a shortage looming, and they want to secure supply. As I said before, target customers are hyperscalers, are AI processor makers, as well as hardware providers for data centers. The whole coverage, stage one and stage two. In terms of what do we offer, of course, includes our planned capacity build-out, which we alluded to you in Dresden, in the other sites, Kulim and Villach. Yes, these agreements reach out multiple years, which also covers the end of this decade, so the late 2020s.
Brilliant. Thanks very much.
We have there a spectrum, right? Not all are covering that long. Some customers only want to engage in shorter agreements. Others are till the end of the decade. A broad spectrum, and please understand that I cannot go into individual customer agreements.
No, of course. Actually, just had a quick follow-up maybe on the backlog and the commentary that Andreas made earlier on a positive reaction from the client based on the price hikes you've announced in July. How much do you think this is driving sort of a pull-in in orders ahead of the price hike that you may put through again either in Q4 or into next year? Just wanted to understand that a bit.
Look, the pricing which Andreas alluded to is, of course, going into the AI direction, also with distributors. You know that a big majority, or not a big majority, but a big chunk of our business is under VPA. I think we will get a lot more clarity from the price increases, how they come in exactly, now negotiating with our customers over the next five, six months. The vast majority of the VPAs kick in January. That's the way I would position it, and that's why, of course, you would not expect now a pull in the backlog because the VPA negotiations just will start in early autumn.
Okay. To be clear there, the CRAs are only with AI customers, right? Could it be that given the looming shortages, Tier Ones or even some industrial customers feel the need to either secure CRAs or are effectively forced to pay up for capacity?
Not quite sure. Say again, please. There was an interrupt.
What I meant is the CRA is only with hyperscaler customers, so I get the concern if you're an automotive customer or industrial customers, that your capacity is going to go primarily towards those guys. Does that motivate your non-AI customers to sign either CRAs or to actually pay up to get capacity?
First of all, the CRAs are not only with hyperscalers, as I said. Processor makers as well as other hardware providers are in the queue for signing CRAs. Of course, there is a certain recognition in the market that the power market is impacted by AI. I explained that already in the past, right? Before AI, the total MOSFET market below 100 V was EUR 6 billion. Now we are talking with AI, of course, about very different numbers.
The order income, and let's jump now to automotive, as I said, is strong in microcontrollers, which has no capacity linkage whatsoever with powering AI. Is strong in analog. I alluded to the opportunities we see in China. Then, yes, in automotive MOSFETs, you could think of such an effect, but the order entry in auto is much broader than this halo effect from powering AI.
Okay, brilliant. Thank you very much.
Okay, time to wrap up. I think we've been generous with our time. Thanks for all the questions to the callers and for the answers to our Board members. We are herewith concluding our fiscal third quarter conference call. For further questions, please reach out to the IR team. We wish you an enjoyable August break. Of course, only after writing your reports on IFX. Take care and have a good day.