Hi everyone, and welcome to Soitec Fiscal Year 2026 Results Conference. I'm Alexandre [Petovari?]

, Head of Investor Relations, and on behalf of the team, thank you very much for joining us today. Before we begin, let me remind you that today's presentation contains forward-looking statements, please read the disclaimer that's available in the presentation materials.

Today's session will begin with our CEO, Laurent Rémont. You'll also hear from the executive team: Albin Jacquemont on finance, Steve Babureck on strategy, Christophe Maleville on innovation, and Cyril Menon on operations.

After these prepared statements, we'll open the Q&A, and Laurent will end the session today with some remarks. Thank you again for being with us, and with that, Laurent, over to you.

Thank you, Alex. Hi everyone.

It's a real pleasure to be here with you for the first time. Since I took over this position on April 1st, I have spent my time getting to know the company from the inside, meeting all the teams and the key customer.

I look forward to sharing with you my observation so far, the immediate priorities for Soitec as I see it today, and my reading of our position within the broader industry in which I have spent the last 30 years. To beginning with, I want to take a step back and share with you what I think are the three main takeaway from our fiscal year 2026 results.

It has been a difficult year for Soitec, with inventory correction impacting volume and low visibility in some of our end market. However, like the rest of the industry, we can also see AI accelerating further and starting to drive a radical shift in semiconductor demand and requirements.

My first message for today is our fiscal year 2026 performance reflect that. Revenue is down.

Our fab loading is below normal, and deliberately so, and it is impacting our margin. On the positive side, our balance sheet is robust, liquidity is strong, and free cash flow has turned positive.

Albin will take you through the number later on. Second, we continue to innovate, optimize, and strengthen our portfolio to benefit from the strong megatrend, notably AI: so photonic in data center, FD-SOI at the edge, power over insulator in advanced connectivity.

We are strengthening what differentiates us: strong ecosystem intimacy, unique R&D capability, and an agile and scalable manufacturing footprint. Third, we have started 2026, and we will continue in 2027, to position the company for a return to profitable sustainable growth.

We have done that by reducing customer inventory, sustaining discipline on cost, and moderating working capital and CapEx to improve cash conversion. Regarding fiscal year 2026 numbers, let me give you the headline before Albin talks about it.

Revenue was down year on year at €592 million, largely due again to ongoing inventory correction in RF-SOI. This was partially offset by strong momentum in AI.

Gross margin stood at 16%. This is a temporary low, linked to the choice to reduce our fab loading and prioritize inventory normalization and working capital discipline.

Our efforts have started to pay off, the free cash flow has turned positive at €63 million, thanks to the focus on working capital and the discipline on CapEx extension. Free cash flow restoration is the first step of our trajectory ahead of revenue growth and operating leverage, as Albin will explain later.

Let's now take a closer look at each of our end market, starting with mobile communication. We see very different dynamics at play in our mobile RF business.

First, piezo on insulator, POI. POI is gaining market share over SAW filter technology.

The adoption remains strong. We have now 13 customer in production and 9 in qualification.

A proof point of this is a long-term agreement signed with Skyworks in March. At the same time, fiscal year 2026 POI revenue saw a temporary softness in Asia after a very strong fiscal year 2025.

Looking forward, we expect volume to increase, partially compensated by ASP erosion while the POI market matures and reaches scale. We will manage this transition at scale with discipline and execution and investment.

Then RF-SOI, the second pillar. Customer inventory correction is ongoing.

It is moving in the right direction, but there is still a lot to correct. Customers are still holding around 2 million RF-SOI wafers.

The correction in Q4 was limited, in line with seasonality and impacted by memory shortage in the mobile market. Looking ahead, under-shipment will continue as customer reduce their inventory, and the pace may also be influenced by other market dynamics.

Finally, FD-SOI, our third technology used in mobile market, especially for 5G millimeter-wave transceiver. Here we secured key foundry design wins with leading flagship smartphone, including iPhone 17.

We are also diversifying into satellite communication and ultra-wideband. Now moving on our second segment on edge and cloud AI.

Here performance was strong and is accelerating. The first is FD-SOI at the edge.

FD-SOI is uniquely positioned when it comes to RF, as we have seen in our mobile segment, but also for ultra-low-power consumption. This is what matters for edge AI application and personal AI assistant.

The ecosystem continues to build up. Adoption by major foundries and IDMs are progressing.

We are seeing new design wins for low-power microcontroller in smart glass and wearables. We are also diversifying beyond microcontrollers into image signal processor and cybersecurity application.

On top of this, we see further potential moving forward with physical AI and battery-powered AI assistant. The second story on edge and cloud AI is photonic SOI in data center.

I am sure you will be all ears for this one. As we have been a leading supplier in photonic SOI for optical transceiver for more than 10 years.

Here lately, we have witnessed a very sharp acceleration of the demand, especially since March, fueled by scale-out and scale-up architecture evolution in data center. This applies to both pluggable optics, but also to the preparation of the ramp-up of co-packaged optics.

Shipments are progressing with more than 10 customer who are either already in full production or in qualification with their own product. Longer-term agreements are being put in place both for 200 that will remain strong, and 300 that is getting more widely adopted.

We expect strong momentum to continue, but we remain aware that there will be challenge and uncertainty along the way. Customer product qualification cycles, data center architecture choices, reliable interoperability between all the building blocks, and timing of deployment, all of that can create volatility.

Our approach here is to be ambitious, but also to carefully check progress of all building blocks through the supply chain at the same time. Photonics is a clear focus for the company.

Responsiveness and execution excellence are key to leverage our unique, flexible manufacturing at scale, and to continue to implement differentiated roadmap and features like LNOI for the next generation of modulator. Overall, edge and cloud AI is a strong growth area for Soitec.

Let's now move to automotive and industrial segment, where market dynamics are again quite different. Automotive demand remains weak, and there is still excess inventory at major analog players.

Power-SOI reflects the end market more generally. A long-term agreement with a key customer gives us good visibility, but the volume remains low, mainly because of the overall market condition.

We continue to prepare the transition to 300, especially for battery management system, where we continue to see a solid long-term potential. FD-SOI is a different picture.

Customer appetite here again remains strong. We have good visibility on the ramp trajectory, supported by design wins in automotive microcontroller, in radar ADAS, in wireless communication and ultra-wideband, or global navigation satellite system.

That said, we know that automotive qualification cycles are long, and ramps do not always move in a straight line. On SmartSiC, the technology value for high-performance segment has been demonstrated, but the ramp-up is not taking the shape we initially expected because of radical market evolution in EV automotive market and very intense price competition from mono-SiC Chinese players.

As a result, we have recognized a €41 million impairment. We remain focused on identifying the right pass forward for SmartSiC, but we will do so with a more selective approach.

Now, based on my experience in this industry and the two months in the job, let me spend a few minutes on what I believe makes Soitec unique. I see three main assets to leverage.

Firstly, the depth and the intimacy with our ecosystem built over decades: so research institutes, foundry, IDM, fabless players, design house, technology providers, and end customers. It helps us to drive our roadmap and gives us a track record in shaping new markets and transforming innovation into commercial success.

The second asset is our innovation engine. This is built on more than 30 years of experience and know-how in engineered substrates, reinforced by strong partnership with leading research ecosystem around the world.

The third is our agile industrial footprint. We have state-of-the-art manufacturing capability in large-scale high-volume production, high quality, high yields.

This gives us the speed today in photonics. We can reliably scale by reallocating fungible SOI capacity already installed without years of capacity ramp and qualification.

Underlying all of this, the talent and the engagement of our people, without them none of this is possible. These three differentiating pillar matter most when applied to where the industry need them.

Right now, this is AI. AI is not just accelerating growth in our industry; it is redefining it, both on the data center side and at the edge.

Both matter for Soitec. First in data center: so hyperscaler investment is expected to quadruple in just two years with the rise of inference and agentic AI, requiring not only more GPUs, but also to have these GPUs more closely connected together—so-called scale-up for those who heard about it.

Optical connectivity is replacing copper connection to increase transfer rate and reduce power consumption. Pluggable optics are growing fast now, and the next generation, co-packaged optics, CPO, is lined up.

Our photonic SOI platform brings differentiated performance for optical interconnects, with waveguide loss specification difficult to match at scale, and also features enabling next steps for system integration. Longer term, there might be even a third wave after pluggable and CPO, with optical I/O and silicon photonics interposer.

The second dimension is what is happening at the edge, with new battery-powered AI assistant and physical AI. Many of the 5 billion AI devices by 2030 will be always-on and extremely sensitive to active and passive power consumption.

They will require extremely good capability to integrate analog mixed-signal, RF, and digital together. FD-SOI excels in these two aspects.

Always-on AI coprocessor, automotive MCU, AI assistant, IoT AI sensors, all are convergings on these requirements: more intelligence with less power and higher level of integration. More broadly speaking, we expect AI to drive new performance requirements that will play along most of our portfolio.

First, as we said already, on the computing side, this is photonic SOI in data center and FD-SOI at the edge. Also, we expect edge AI to drive new needs for connectivity, with more data volume to be exchanged, lower latency, higher band, lower power consumption, and higher integration.

This is where POI, FD-SOI, and RF-SOI come into play. Third is energy efficiency, which is of utmost importance in data center, where we could seek for engineered material opportunities in power SOI or in wide bandgap.

What does that mean in practice? It means three short-term priorities for the company.

First is to accelerate the strong momentum we have on AI. The opportunity is definitely here, but we need to properly execute.

Success will also rely on some external factors: architecture choices, customer qualification timeline, competitive landscape at all the stage of the value chain. We are positioned to capture the opportunity with ambition and focus.

The second priority is to sharpen focus and resource allocation. A full review of portfolio priorities and capital allocation is underway.

The technology differentiation potential remains; what change is execution and concentration of resources. We will share findings publicly when conclusions are reached.

The third priority is to rightsize the organization. The cost reduction program initiated in fiscal year 2026 will continue, including headcount reduction, with formal consultation of employee representatives in progress.

The objective is to structurally reduce the fixed cost base and improve operating leverage through the cycle. At the same time, R&D intensity and strategic capability are being protected.

We are not cutting our way out of innovation, nor in our strategic areas. We do this for very good reason: progressive return to sustainable and profitable growth, this is underway.

That's my reading of where Soitec is and where I'm taking it. First, we have a portfolio built for the right megatrends, with a unique position in AI value chain.

Second, our talented team has developed structural advantages, and we will continue to do so. Lastly, we have a disciplined approach to converting all of this progressively into financial performance.

There is real work ahead, but I'm confident we are doing the right thing, in the right order, with the right team. Finally, with these priorities in mind, let's turn to our outlook for fiscal year 2027.

Uncertainty remains high across several of our end markets, we believe it is right to remain prudent on the guidance horizon. For Q1 2027, we expect revenue to be up 15% year on year at constant exchange rate and scope, factorizing in our effort to reduce seasonality.

As we navigate in a challenging environment, we will continue to face margin headwinds. Some will progressively fade away as our fab loading returns to normal.

Albin will give you more detail on the P&L and CapEx, but one element I want you to remember: we are leveraging our asset fungibility and the previous investment we have made to moderate the cash outflow. We make sure that we secure our capability to capture gross opportunity.

To conclude, having restored a positive free cash flow in fiscal year 2026, our next priority is to get back to a gross trajectory. Thank you.

This is in a nutshell what I wanted to share with you today. I will handover to Albin for more detail on the 2026 number, and I will come back at the end of the Q&A for some concluding remarks.

Thank you.

Thank you very much, Laurent. Good afternoon, ladies and gentlemen, and thank you for joining us today.

Fiscal 2026 was a year of decisive execution for Soitec. We took bold, proactive actions to reset our operating base, protect our balance sheet, and position the company for the next phase of profitable growth.

Despite a challenging demand environment, we delivered on all the commitments we set out at the start of the year. We took decisive action to proactively manage RF-SOI inventories and correct supply.

This was the right call. This discipline was essential to streamline our working capital and support a healthy rebalancing of the channel across the RF-SOI ecosystem.

Yes, this weighed on gross margin in the second half due to temporary under-absorption of fixed costs, but it was a necessary step to restore structural balance. Second, we successfully restored positive free cash flow.

This was achieved through disciplined working capital management and a more moderate CapEx profile, thereby strengthening our financial flexibility and enhancing our capacity to deploy capital into attractive, high-return growth opportunities going forward. Third, we are executing a three-stage roadmap to sustainable and profitable growth.

Stage 1 was the restoration of positive free cash flow, which we have now achieved. Stage 2 focuses on cash generation while progressing toward revenue growth as we continue to work through excess RF-SOI inventories across the channel.

Finally, Stage 3 will be the scaling phase, where we will leverage our structurally efficient operating model to drive meaningful margin expansion and cash conversion as end-market demand improves and recovers. Taken together, these three priorities reflect a clear sequencing of discipline, stabilization, and value creation, positioning the company for a balanced and resilient growth trajectory.

Let's now look into key numbers, some of which Laurent has already touched upon. Revenue came in at €592 million, down 30% year on year on an organic basis, reflecting sharply divergent end-market dynamics.

EBITDA margin was 25.4%, an 810 basis points decline year on year, driven primarily by lower volumes and our deliberate fab deloading. Despite this, operating cash flow remained resilient at €202 million, broadly in line with last year, underscoring the strength of our cash generation model.

As a result, we generated €63 million of free cash flow, a significant improvement from the €23 million outflow in fiscal year 2025. Consequently, our net debt position improved to €56 million, resulting in a strong and comfortable leverage ratio of 0.4 times EBITDA.

In fiscal year 2026, our top line declined by 30% year on year on an organic basis, reflecting sharply contrasted end-market dynamics across our portfolio. Mobile communications declined 41% year on year, continuing to be impacted by RF-SOI inventory corrections at direct customers.

That said, we are seeing encouraging adoption momentum in POI, highlighted by a first multi-year LTA signed with Skyworks. Automotive and industrial remained tepid, down 44% year on year, as the ongoing global automotive downturn and softer demand in both power SOI and FD-SOI persists.

Edge and cloud AI was once again the standout performer, up 8% year on year and up a strong 19% excluding imager SOI. Photonics SOI is now a $100 million-plus platform, growing in excess of 30% year on year.

This is consistent with the continued expansion of our addressable market as optical interconnect demand accelerates across AI data centers. As Laurent has already provided an overview of the operational context across our core product line, I will now turn directly to gross margin performance.

Gross margin came in at 16.3% of revenue, down from 32.1% last year. Beyond lower volumes, three factors drove the decline: lower utilization rate at circa 50% versus circa 70% last year, representing a circa 800 bps headwind; unfavorable mix price, driving a circa 300 bps contraction; operational efficiencies and subsidies provided circa 180 bps of relief.

Looking ahead, as utilization recovers and volumes ramp back up, we would expect gross margin to naturally expand through operating leverage. Turning now to our operating performance, we reported a current operating loss of €8 million, compared to a current operating profit of €136 million last year, driven by gross margin compression.

This was partially offset by disciplined cost management, which we achieved while maintaining our commitment to R&D investment. Net R&D expenses decreased to €45 million in the year, reflecting Dolphin Design divestiture, higher recognized subsidies, and lower material consumption, particularly within SmartSiC development activities.

Stripping out these items, gross R&D remained on par with last year. SG&A declined by 10%, driven by lower share-based compensation expense, the rest of variable profit-sharing accruals to zero, and disciplined cost savings across the operating base.

On the next slide, we set out the breakdown of our non-recurring items. Net loss amounted to €220 million, significantly impacted by non-recurring items, most of which result from decisions taken many years ago.

More specifically, we recorded €123 million of other operating expenses, primarily comprising a €41 million impairment charge related to SmartSiC assets, a €29 million impairment on the currently unequipped Pasir Ris extension, an impairment charge on advanced payments made in 2022 under long-term raw material supply agreements, and a single-digit earnout loss associated with the disposal of Dolphin Design. These non-cash charges reflect disciplined portfolio management and a clear-eyed reassessment of market dynamics, particularly in light of a rapid expansion of lower-cost Chinese monocrystalline silicon carbide capacity and the resulting pricing environment impacting SmartSiC-related assets.

Moving below the line, net financial expense amounted to €31 million, reflecting higher interest costs associated with recent financing activity, and a €17 million non-cash forex loss recorded in April and May 2025. The group has since implemented an enhanced hedging framework to mitigate residual currency risk going forward.

Adjusting our reported net income of €220 million for the exceptional, non-recurring non-cash items previously outlined results in a current net loss of €14 million and current EPS of -€0.48. We fully delivered on our commitment to restore positive free cash flow generation, which reached a positive €63 million in fiscal year 2026, marking a strong improvement compared to the negative €23 million recorded last year.

This performance clearly demonstrates the strength of our execution, rather than reflecting any support from market conditions. I would like to sincerely thank our teams for this achievement.

This marks a structural inflection point. Soitec is now back to generating consistent free cash flow over the cycle, with a more selective and returns-driven capital allocation framework.

This improvement is underpinned by two key drivers. First, we maintained strict discipline on working capital, resulting in a positive €49 million contribution to cash year on year.

This was mainly driven by a €145 million reduction in trade receivables, reflecting lower activity levels and tighter collection management, as well as a €24 million reduction in inventories, driven by deliberate actions to reduce fab loading and operate below underlying end demand, particularly in the second half of the year. This reflects a clear prioritization of cash generation and balance sheet strength over short-term margin optimization.

To give you more context, this slide offers a closer look at these working capital dynamics from a balance sheet perspective. As you can see here, our optimization efforts paid off, particularly in the second half of the year.

Turning now to the second driver of our free cash flow improvement: capital discipline. We further reinforced capital discipline, reducing capital expenditures by 40% year on year to €145 million.

Our capital allocation framework has become decisively more selective, with resources increasingly directed towards some of the most attractive long-term value creation opportunities, in particular, photonics SOI and POI. In parallel with proactive inventory optimization, this sharper capital allocation has enabled Soitec to return to structurally positive free cash flow generation.

Leveraging significant past investment in capacity expansion allows us to secure growth opportunities at the same time. This is a key inflection point.

It strengthens our financial profile, significantly enhances earnings quality, and increases our strategic flexibility as we enter the next phase of development. Beyond performance, we are further strengthening the quality and consistency of our financial reporting framework.

As part of this ongoing effort, we have updated free cash flow definition to reflect prevailing market conventions. The revised definition consistently includes capital expenditures irrespective of funding structure, as well as the net impact of interest paid and received.

This enhancement materially improves comparability across reporting periods and provides a clearer view of the group's underlying cash generation capacity. It reflects our continued focus on strengthening the credibility, transparency, and investor relevance of our financial disclosure framework.

Turning to the balance sheet, we ended the year with a strong and further reinforced financial position. Net debt decreased by €38 million to €56 million, driven by continued strong cash generation and disciplined financial execution.

This further strengthens our balance sheet and enhances our financial flexibility going forward. Liquidity remains very strong, with €562 million of gross cash and €270 million of fully undrawn committed facilities, providing substantial financial flexibility to support both operations and future growth initiatives.

A major achievement this year was the successful repayment of our October 2025 OCEANE maturity, partially refinanced through a €222 million Schuldschein issuance with an average maturity of approximately four years. This refinancing significantly enhances the quality of our balance sheet by extending debt maturities, diversifying funding sources, and substantially reducing near-term refinancing risk.

To conclude on our fiscal year 2026 performance, we continue to maintain a solid balance sheet, underpinned by strong liquidity and a moderate net debt position. In parallel, we are further progressing on our commitment to greater transparency, consistency, and higher quality financial reporting.

Bringing all this together and moving to the outlook: we expect Q1 revenue to be up around 15% at constant currency and scope year on year. Mobile communications is likely to remain challenged, reflecting the ongoing RF-SOI inventory correction, as well as softer smartphone demand amid current memory supply constraints.

In stark contrast, edge and cloud AI should continue to deliver robust growth momentum, underpinned by the accelerating ramp-up of photonics SOI solutions for optical interconnects in AI data centers. Automotive and industrial demand is likely to remain muted in the near term, pending a more meaningful recovery in demand trends.

In parallel, we remain firmly committed to our disciplined capital allocation framework, with fiscal year 2027 cash CapEx expected to remain contained at around €100 million for the full year. Fiscal year 2027 will be a transition year, but the path to margin recovery is clear and tied to fab utilization improvement and the scaling of our high-growth platforms.

Fiscal year 2027 will be reflecting fab utilization rates which, while improving, will remain materially below optimum levels, as well as a level of subsidies expected to be significantly lower. Indeed, discussions are ongoing, as calendar year 2026 marks the conclusion of the IPCEI 2 program ahead of the anticipated launch of a potential IPCEI 3 framework.

Finally, the strengthening of euro against the US dollar is expecting to create an additional headwind for our financial performance under the current operating conditions. Our net forex exposure for fiscal year 2027 is now 95% hedged at 1.19 euro per US dollar.

In closing, the company is well-positioned for the future, supported by the significant diversification of its portfolio achieved over the past several years. Execution of the plan articulated back in November remains firmly on track.

As a reminder, the plan is structured around three distinct phases: Phase 1: restoring cash generation through a reduction in working capital and the return to positive free cash flow. Initial results are already tangible, and we expect further progress ahead, with a meaningful portion of working capital expecting to be monetized over the coming years.

Phase 2: returning to growth. For the first quarter, we are guiding to approximately 15% organic revenue growth, reflecting the improving momentum across the business and some efforts to reduce seasonality.

Phase 3: restoring a structurally higher level of profitability, with operating leverage and cash conversion as growth resumes. This phase is expected to follow the sustained recovery in top-line growth.

At this point, let me pass you on to Steve to take you to our strategic priorities. Steve.

Thank you very much, Albin. Hello, everyone.

Great pleasure to catch up with you today. Of course, in this brief section, I will talk about AI, and I will double-click on a few messages that Laurent conveyed earlier today.

First, AI deployment continues to accelerate through overlapping waves, from gen AI that was introduced a few years ago. We are now moving towards agentic AI, and eventually physical AI towards the end of the decade.

Second, each AI wave requires system-level optimization across compute, memory, connectivity, interconnect, and power efficiency. Engineered substrates are enabling this transition with more connected and more efficient semiconductor systems.

Third, AI deployment impacts our end markets with different timing, different intensity, progressively expanding Soitec's opportunity and footprint across data center infrastructure, edge AI, and connectivity ecosystems. Let's start with the applications.

AI is moving fast, both for consumer and enterprise use cases. The first important phase started with deep learning last decade, systems enabled with perception, classification, and optimization.

Then gen AI introduced content creation across text, images, code, and video. Today, we are entering the agentic AI phase, where systems plan, decide, and execute multi-step tasks autonomously.

Looking forward, physical AI will progressively embed AI into robots, vehicles, and intelligent systems, interacting directly and dynamically within the physical world. These AI waves overlap, reinforce each other, continuously expanding semiconductor and infrastructure requirements.

AI deployment is globally driving a massive infrastructure scaling. We're shifting from traditional hyperscale data centers towards multi-gigawatt AI campuses.

For each of these gigantic campuses, which could cost roughly $50 billion, power consumption is expected to reach around 1 GW, which is literally what is needed to power a major metropolitan area. At this scale, the main challenge becomes system orchestration, energy, networking, and integration.

Scaling token generation exponentially requires a fundamental shift in the architecture. To handle high bandwidth, networking speeds must rise by an order of magnitude compared to current data centers.

We need to shift away from copper and move towards optical interconnect, which enables both ultra-high speed and much better energy efficiency. What does it mean for us?

What does it mean for engineered substrates? In the cloud, first: traditional compute system architectures reach physical, thermal, and somehow economic limits.

The transition towards optical interconnect creates significant opportunities for photonics SOI and LNOI platforms, supporting the next-generation optical architectures. When it comes to power management, AI infrastructure also requires much higher conversion efficiency and thermal performance, which is a great support to the adoption of silicon carbide and gallium nitride technologies.

Moving to the edge: connectivity evolves from 5G towards 5G-Advanced and eventually 6G, increasing RF complexity and integration requirements. FD-SOI technology stands out as the premier architecture capable of delivering the ultra-low-power efficiency which is required by edge AI for various applications, such as wearables or smart glasses.

In conclusion, for both infrastructure and edge applications, we are targeting every layer of the AI stack with our expanding portfolio of engineered substrates. Working side-by-side with our fabless and foundries customers, as well as the entire AI ecosystem, we believe Soitec is extremely well-positioned to play an increasing and strategic role in the build-up of the global AI backbone.

In order to power that backbone, we need to push constantly the boundaries of physics and materials science. This is my transition to introduce our Chief Technology Officer, Christophe Maleville, our next speaker.

Christophe.

Thank you, Steve, and hello, everyone. Steve just walked you through where we play and why.

My objective today is to show you why what we do is hard to replicate and getting harder. I'm really happy to cover this year again a more and more exciting story about our technology and our innovation.

I have three messages for this section. First, we keep expanding our moat and winning at scale: 4,800 patents, an extensive Smart Cut toolbox, 30-plus years of RTO partnerships, all actively renewed.

This is a strong combination of experience and dynamism. Second, we are a unique innovation powerhouse.

We keep strengthening the Smart Cut moat at the core. We built a complementary technology toolbox around it, and we leverage a strong position across the entire ecosystem.

Third, we keep expanding where value is created. Performance starts with the hardware, five platforms delivering value at device level.

Of course, we are extending that logic to AI optical infrastructure with photonics SOI, LNOI, and new materials in development. We apply the same discipline across all three pillars.

We invest where value is created, and we work with the leaders of each segment. Let me show you how.

I want to start with the foundation: four pillars built over 30-plus years, reinforced every year. First, our innovation is different by design.

We design pilot products for manufacturing that address specific customer needs. We filed 470 new patents this year alone, all anchored in industrial relevance.

Second, we go from lab to fab faster than anyone because we leverage models we keep updating for each new material and each new structure. Know-how compounds: patents becomes public after 20 years; know-how does not.

More than 30 years of Smart Cut development sit behind every new product we launch. Third, we protect our freedom to operate at each step of the product life cycle.

15 R&D partners worldwide keep us co-developing next-gen architectures ahead of public roadmaps. Finally, we have learned how to balance our innovation effort.

30-plus years of cycles taught us when to push and when to consolidate. Our 70/30 split between incremental and disruptive R&D is what sustains the model.

Our RTO ecosystem now stretches from lab to fab to application lab. This foundation is what makes everything else possible.

Let me now show you the toolbox we have built on top of it. This toolbox is the operational translation of the foundation I just described.

You have to understand that Smart Cut is not only bonding and splitting. We have extended it into a full toolbox: interface engineering, crystal engineering, advanced processing, refresh and repolish.

Each brick is a competence we have built, qualified, and industrialized. On top of this, we have developed a deep expertise on new materials.

Each new material we develop leverages the know-how built across three decades. This is what makes our roadmap incremental by design, reducing time to market and capital intensity for every new product.

This is the technical core, but a toolbox is only as strong as the ecosystem and the experience around it. Let me detail this second layer.

Around this technical core, we have built an ecosystem, and that ecosystem is itself a moat. First, we work hand-in-hand with the leaders of our supply chain.

We have built an ecosystem, and our privileged relationships with material suppliers and toolmakers are another differentiating factor. We co-design with them from day one.

That is how we secure both performance and industrial readiness on every new material. Second, our reach extends all the way downstream to the users of our substrates: foundries, fabless players, design house, IDMs, OEMs are part of the same ecosystem.

Being embedded with them is how we anticipate their needs and shape the next generation of products together. Third, our partnerships stretch across the entire value chain, from CEA-Leti on the upstream pilot line to RTOs and research partners.

This is what gives us our lab-to-fab speed and what raises the entry barriers around us. Combined with the toolbox, this ecosystem is what makes Smart Cut hard to replicate.

Let me show you what it delivers with one example, one platform that captures it all: photonics. Photonics is where the structure of our engineered substrates matters the most.

Let me show you what makes our photonics SOI unique. First, the substrate quality is what sets us apart, and we cover the full roadmap.

Such engineered substrate brings atomic layer level control of the top layer uniformity, very low roughness at surface and interface, extreme low defectivity, and all of this over engineered silicon-based substrates. This is what enables low optical losses, better modulator performance, and the mechanical stability needed for reliable high-volume manufacturing at our customers.

We do not bet on a single architecture. With photonics SOI and LNOI today and new materials in development tomorrow, we have a holistic product roadmap that addresses the different architecture choices our customer may take across the value chain.

Second, this is not only a lab result. Our product is mature, and our infrastructure is ready to scale.

We have built it on more than a decade of expertise, industrialization, and ecosystem partnerships. We have accelerated customer prototype deliveries by 11x over the past three years, the clearest signal on the ramp ahead.

Today's prototypes are tomorrow's volumes. Third, we keep extending the platform to secure our customers' roadmaps.

We now work with more than 10 customers on this platform. We have seen a strong demand acceleration, especially over the past couple of months.

Again, a key message: customers come to us for this prototypes, and they move from prototype to qualification faster and faster. This is where engineered substrates become the cornerstone of AI infrastructure.

With that, let me hand over to Cyril, who will take you into how we operate this innovation engine at scale. Thank you.

Thank you, Christophe, and hi, everyone. You've just seen what make us unique.

My job and the job of operations is to turn that uniqueness into volume, on time, and at the right cost. A large part of that is our ability to scale rapidly, which is critical right now with photonics.

This year, more than any other, it is key to strike the right balance between optimizing capital allocation and securing growth opportunities. What do we do?

First, we manage a cycle. Demand in some part of the business is still soft, we deliberately run our fabs below capacity, and we optimize where we can to absorb this cycle.

Because we have already built our footprint, we can capitalize on past investment to moderate our CapEx while making sure we are able to deliver on a growing demand for specific products. Second, at the very same time, we address a growth that is already in front of us.

Photonics SOI and POI need capacity now, and we are putting it in place. The high degree of fungibility of our assets allow us to accommodate more demand than anticipated on a given product and to scale rapidly.

Third, we do all this on a foundation that make us more resilient. Our ESG discipline is not a side topic here; it lowers our cost and protects our supply chain.

How do we hold that balance in practice? It runs on two sides at once.

On one side, we optimize how we allocate capital, and that is twofold: first, we deliberately run our fabs below capacity to correct our inventories. This contribute to reducing our working cap and to restore positive free cash flow this year.

Second, because the footprint is already built, we lean on past investment to moderate our CapEx while still being able to deliver on-demand white growth. On the other side, we invest where it counts to capture growth in a targeted way.

Where a segment accelerates, like photonics SOI and POI, we can marginally invest and scale fast. We keep strengthening our core SOI technology across the board, we stay ahead where they are already growing, like FD-SOI, and ready where the market still has to turn.

The two sides connect through fab fungibility. It lets us moderate CapEx by investing selectively only in the process steps that matter, and by optimizing our workflow to increase throughput.

The same agility absorbs the cycle on one product and redeploy capacity toward another that is taking off. Let me show how that works on the ground.

You've heard us talk about asset utilization for years now. Agility is not something we improvise in a downturn; we engineer it.

Our SOI fabs are built to be fungible, which means a line serving one product can serve another with very little new CapEx and very little time. POI is a great illustration: we can turn idle SOI 200 capacity into POI capacity for a very limited incremental investment.

The same agility is now being tested by photonics, and this is where it matters most. Demand has been accelerating this year, and our ambition is to deliver on all of it.

We can accommodate the demand by activating four levers at our disposal: First, we can leverage existing capacity in Bernin by reallocating underutilized SOI lines to photonics SOI, optimizing both 200 and 300 footprints. Second, we are actively qualifying Pasir Ris facilities in Singapore and have already shifted prototypes production to mitigate the need for future qualification.

Third, we are tooling unused capacity in Bernin 4, specifically for 300 photonics SOI. Finally, if demand surge even further, remember that we have built the extension in Singapore.

With this multilayer approach, we have room to follow demand right across the group. This is what set us apart: we can scale from day one.

We built the footprint in several years of significant investment. Today, that base is what let us scale on demand, not on new spend.

One last important point for us: ESG is operational discipline. Our efforts make the business more resilient, our operations more sustainable, and our cost structure leaner.

Let me give you three concrete proofs: First, we are running ahead of our own roadmap. On carbon, we will reach our 2026 emission target two years early, which gave us the confidence to commit to a tougher one.

We lower our 2030 absolute emissions ceiling by more than 20%. On water, the same discipline shows: we have already cut our withdrawal intensity by 30% since fiscal year '21, and tripled our reuse rate to nearly half of all the water we use, putting us well on track for our fiscal year '30 target.

Second, we anticipate risk rather than absorb it. We hedge energy and lock in low-carbon sourcing, and we double source the critical materials and utilities we depend on, from helium to process gases.

We did this ahead of the market, which protect us from price spike, from shortages, and ultimately keeps our fab running through any disruptions. Third, this discipline is recognized externally.

Our MSCI rating was upgraded to AAA this year, our fourth rating upgrade in five years. The way we run operation is consistent, end-to-end, disciplined in the cycle, ready for the growth, and resilient by design.

With that, I will hand back to Alex for the Q&A.

Thank you, Cyril. We will now open the Q&A session.

Gentlemen, if you want to sit on stage. We have microphones in the room, if you'd like to ask a question, please raise your hand.

For those of you online, you can ask and type your question via the platform. Please try to limit yourself to two questions, but I guess you won't.

Emmanuel, yes.

Hello. Yes, it works.

Emmanuel Matot from ODDO BHF. Two questions.

First, you have different scenarios for photonics this year, if I understand well. Can you detail those scenarios?

Maybe you can focus on the most optimistic and the one that is the most cautious. Are those scenarios mostly related to your capacity to answer demand, because if I'm right, you are facing production bottleneck for those solutions?

Second, how do you see the RF business over the next few years? Did you reach the bottom last year, according to you?

Is it premature to be so optimistic given the negative trend in the smartphone market, which could lead to an acceleration of destocking by your customers, meaning foundries? Thank you very much.

Thank you. Maybe I start.

Regarding your question on silicon photonics: so as we said, we see definitely an acceleration on silicon photonics demand, especially past couple of months. This obviously put our organization at stretch, because this demand is coming rapidly, and we are putting everything together in the organization to respond to this demand.

As Cyril was explaining, we have the capacity. The question is more the speed to respond to this surge of demand, the lead time.

That's what we are focusing on right now, optimizing our production flow, enabling Pasir Ris qualification, all of this, okay? That's the first part.

Now, we remain aware that that's a full ecosystem that is moving regarding silicon photonics. It's not only us.

It's a complex ecosystem in the data center moving from copper to silicon photonics. This means packaging technology, fiber attached, data center architecture choices.

There are multiple factors influencing the end demand. We see clearly, compared to what we said in the past, we said 20% to 30% CAGR.

We see something above 30%. Now, giving a number exactly, we have to be humble in this industry.

We track that carefully. We ask as well our customer to take commitment on the long term, but I will not give you a firm number definitely regarding that.

That's the first part. The second part on RF-SOI: I have done products on RF-SOI for the mobile market, I know this market quite well.

That's a market that is definitely in a different cycle compared to photonics. That's a market that is mature, and that's a market where we have still inventory correction to happen, and that's a market as well having currently some memory shortage impacting some of the mobile phone makers.

Definitely not the same dynamic. This will take time to correct.

We plan this market still to grow moderately. That's a mature market, but that's still an important market for us.

Hello, Sébastien from Kepler Cheuvreux. I've got one follow-up question on photonics.

How do you see the competition in that market? Do you see more competitors coming to this big market opportunity?

With LNOI technology ramping up later towards the end of the decade, do you see other players coming to this photonics market? The second question is on the margin side.

You are currently, for Albin, on a breakeven margin. You had in the past an ambition to have mid-20s operating margin.

What are the building blocks to move from today's breakeven margins to mid-20s, and what could be a timing to reach this kind of margin level? Thank you.

Maybe I start with the first part, and then Albin, you reply to the second one. Regarding competition: I have been in the semiconductor industry for 30 years.

I know that everything is very competitive in the semiconductor. My approach to that is to consider there is always a strong competitor in the market.

I do not expect photonics to be different. If this turns out to be different, very well, but my assumption will be there will be competition.

The way out of that is to be excellent in what we do. That's what we are focusing on with all the team here.

In innovation and being sure what we learn over the past 10 years, we use that, we continue to develop the roadmap, we continue to develop differentiated features like LNOI or others in operation, being sure that also what we learned ramping up with customer—and every customer here is a different solution, that's not one size fits all—so we are sure that all this learning, we leverage that. My assumption is there will be competition.

Maybe I hand over to you?

Yes. Thanks for the great question, especially in light of a significant decline of our gross margin this year.

I think it is a bit premature for us to provide specific guidance into fiscal year 2027. That being said, the main levers for margin improvement are expected to be the reduction of under-absorption costs—I'll come into the detail of that in a minute—as utilization recovers, and also a normalization of a price mix environment.

Like we alluded to, offsetting these factors, however, we'll see lower subsidy levels and that will weigh on next year and before recovery in 2028. I think your question leads us to how to estimate operating leverage, and that's the way I will answer your question.

I will not give you the year where we will reach 20% EBITDA, although the executive committee has an idea on this. Operating leverage: so we benefit from operating leverage at two distinct levels within our cost structure.

First, it's gross margin, and second, EBIT. First, at manufacturing level, a meaningful portion of process cost is fixed, creating absorption effect as volume increase.

Second, at the corporate level, both R&D and SG&A expenses will remain largely fixed in the short term, allowing incremental revenue growth to translate into a stronger margin profile over time. Specifically with respect to process cost and to help you support the modeling assumptions, you should consider that approximately 70% of process costs are variable and cash in nature, while the remaining 30% remaining largely fixed and non-cash.

In the very short term, incremental revenue carries a contribution margin of approximately 60%. As revenue growth becomes more meaningful and capacity utilization progressively normalizes, a more appropriate assumption, according to us, would be for a drop-through rate closer to 50%.

To close on this very relevant question, you can consider that an additional €100 million of revenue would be expected to generate approximately €50 million of incremental gross profit and EBIT, you have all the ingredients to do the math.

Alexander?

Yes, hi. Alexander Peterc from Bernstein.

I just have two questions. One is again on photonics.

Sorry about that, but it's a bit of a hot topic at the moment. I just like to understand, you're spending a bit more on CapEx this year.

It's not a big number, but it's a little bit more than you thought a few months ago, and that is primarily attributable to photonics. Could you maybe quantify for us by how much, by what factor you're increasing?

Are you basically doubling your photonics SOI capacity? I understand you're now qualifying Pasir Ris, once that's qualified, will you have basically double the capacity?

That would be number one. The second one is very simple, it's for Albin on the tax dispute with French authorities.

Could you tell us if you heard back from the taxman? Thanks a lot.

Cyril, do you want to take the first one?

Yeah. Obviously, speed matters, and how we deploy the capacity matters a lot, and we adapt our capacity for the photonics ramp-up is one of the cornerstone of this discussion.

As you have seen, we don't have only one avenue to improve our capacity. Qualification of Pasir Ris comes for free, in a way, we don't have to invest.

We have already the assets over there. It's really important as well to understand that we have already initiated this qualification for all our customers in Singapore in the last month, and this does accelerate right now in order to support the growing demand for photonics.

This is one item which makes sense in terms of capacity availability, in terms of return on capital employed, because obviously the assets are already there. We have to ensure that the consistency of the product that we deliver from Singapore is 100% similar to the one that we deliver from Bernin, and we focus a lot on this topic, which is quality-related, and we are able to demonstrate that this is the case.

Second item, once again, which is not requiring any CapEx, is to start from scratch from Singapore with new customers, and we have new customers starting prototype from scratch from Singapore. We have more and more in the last month, and this does accelerate for sure as the growth of the photonics ramp-up.

Third, when it is necessary, for instance in Bernin, we deploy CapEx. We have seen that our CapEx this year is going to be €100 million.

A part of it, less than half of it, but still a part of it, is to adapt our assets to the need, to adjust some assets, to add some assets in order to increase capacity in Bernin. On that one, we're quite lucky in a way to have or it has been anticipated to have Bernin 4, which is able to run SmartSiC and 300 in the same factory.

This is about agility once again. That this plant has been qualified with refresh in the past.

The time to cash is very short because we use the same utilities, we don't need to requalify at the customer level, and this is a good way to scale up easily without adding additional clean room and additional fixed costs in our model.

Albin, on the tax?

Yes, on the tax reassessment. For reference, you all know that Soitec received a tax reassessment notice in January 2025 relating to first the alleged abuse of law characterization associated with the defeasance of the solar activities, as well as a challenge regarding the valuation of the Singapore activity shares in the context of a restructuring of a transaction which took place before 2020.

In March 2026, we received the tax authorities' response to our formal observations, and the French tax authorities accepted part of our claims. In summary, the proposed reassessment could have a twofold impact.

First, in the case we don't agree on a settlement with the tax authorities—maximum impact—first, it could reduce our tax loss carryforward by €384 million out of a total of €765 million tax losses carried forward. Second, and in addition to what I said, it could result in a maximum cash exposure of €205 million.

Discussion with the French tax authorities are scheduled to commence next week, and Emmanuel and I have a meeting scheduled in Romainville next Friday, and they will continue thereafter in July. Importantly, the advisors assisting the company continue to view Soitec's position as robust on the merits.

To be very comprehensive, should an amicable resolution not be achieved, which is not our scenario, the company would pursue litigation proceedings, which are expected to be lengthy, and would require the provision of a payment guarantee in favor of the French tax authorities in the amount of €128 million. That's as much as I can say, and I would hate to comment on what I don't know.

I think Nigel has been holding the mic for a while.

Thank you. Hi.

Nigel van Putten, Morgan Stanley. I'll take a breather on photonics for a bit.

Two questions on the mobile communications business. First off, on RF-SOI, do you still target a reduction of about a million wafers as per the end of the year, as was previously commented on?

Then on POI, I think, Albin, you've mentioned POI headwinds, but why is that? Can you provide a little bit more color?

I think it's typical as part of the LTA, but should there also be an offset in your ability to plan and thus gross profit would actually be beneficiary? That's it from me.

Thanks.

I'll leave the second question to Albin, but before, I will give a little bit of color. I wasn't really sure what your question was about RF-SOI, but I think it's about inventory digestion.

I think we come from a place about a year ago where we had a very excessive amount of inventory. As Laurent already shared in the beginning, we have been burning that down, and the plan is to further bring this down to a level which is, and that's a bit a loose term, but acceptable.

That level of acceptance is not determined by us; this is determined by our customers where they feel comfortable. The expectation is that moving forward, we will be further burning down the inventory.

That should restore. Also already indicated by Laurent, there are other dynamics playing.

Right now, we see that the smartphone market is under pressure. Memory shortage is one of those modulators.

It is a little bit more difficult to really pinpoint when this is going to happen, but in that horizon, I do believe things will normalize. I think that was the part on the RF-SOI.

There is another part which has, I think, two elements. I have a color a little bit more about what's happening in the POI space.

I think it was also a more technical question. Albin can probably answer that as well.

We already talked about what's happening in POI. There's a lot of adoption, there is growth, there is a clear need for POI because it's simply a better product for the RF demands in the future.

Now, what we have seen is two things: we see things outside of China, and we see things inside China. Inside China, there was a lot of momentum, let's say, 1 1/2, 2 years ago, a lot of inventory was built up, there we see a little bit of softness.

What we also see, and I think that's exciting because we also signed a LTA, and probably you all have seen that, that is outside of China. I think that's very exciting.

The moment you start talking about LTAs, this is doing two things. It gives you a lot of certainty as in what kind of volumes you're talking about, but as a consequence, there is a possibility that your pricing is something you need to give in.

That is specific for POI. Albin, anything you want to add?

No,

that's exactly the point. We are very happy, nothing wrong with the product, all the contrary, and Laurent explained pretty well the good prospects of our product, 6G millimeter-waves, and so on.

We are very excited. We refer to POI pricing in the context exactly like you're explaining, LTA, which gives visibility and volume, and it is normal that as we ramp up the volumes and get more visibility, pricing is going down to a reasonable extent.

That was the only message we wanted to carry today.

Oliver?

Thank you, and welcome to the team, Laurent. I guess a couple of questions on photonics.

First question, I just wanted to understand a bit more about the conversations that you're having with your customers and your visibility. Would you say that the lead times are fairly lengthy right now?

Would you say you have pretty good visibility on let's say photonics growth for this fiscal year? How should we kind of understand that?

The second question is on CPOs, perhaps ramping in calendar '27 and onwards. My understanding is that that is incremental to the growth that we're seeing in pluggables right now.

Would it be fair to characterize it as potentially accelerating your photonics revenue growth as CPO ramps? Thank you.

Thank you. Works, yeah.

Regarding visibility on photonics: so again, that's a very dynamic environment, visibility we had two months ago is not the visibility we have today, just to be clear. Yes, we have visibility even on the long term from the customer.

Now what is important is to have some skin in the game, I would say, from customer as well on this topic on the long term, because right now, this is a market where everybody want to grab the market share. We are cautious as well that there might be some double-counting overall.

We are trying to assess as well what is the end demand and being sure that somehow our customer have some skin in the game as well on the long term. That's what we are trying to do right now.

That was the first part of your question, and I'm sorry, I forgot the second part. CPO incremental.

I can answer or if you like, go ahead.

It's my favorite topic, so.

Go ahead then. Yeah.

This question, I get a lot. I think it's important to understand where does pluggables play, because that's the current business, and where does the CPO play.

What we see is that CPO will play a key role in the scale-up in the data center. It will not, let's say, jeopardize or even cannibalize the silicon photonics that we see today, which is all about the datacom.

It's about connecting the data centers, the AI clusters, which is happening with the pluggables. For the foreseeable future, that market is still growing, which is the interconnect, the datacom.

CPO is really on top of that. It has a slightly different application.

Maybe that's the shortest answer: it's not cannibalizing, it's really on top of what we currently see as the main application for silicon photonics, which is the pluggable transceivers.

I'll be a bit inclusive and ask a question online for those who couldn't come with us today. Question from Craig McDowell from J.P.

Morgan on the Q1 guidance and fiscal '27 phasing: "The press release refers to less pronounced seasonality. Has there been broad-based changes to supply contracts or customer ordering practices to make this possible?"

Q1, we are still trying to reduce the seasonality and as well to reduce inventory that we have at our customer. That has to be factored in, yes, in our guidance for Q1.

Which mean you cannot extrapolate from Q1 basically the full year behavior, because you should expect less seasonality within the year.

Thanks, Laurent. Maybe a second one from the platform from Aditya Metuku from HSBC: "GlobalWafers is talking about ramping photonics SOI wafers using their own technology and ongoing customer qualifications.

How do you see competition from GlobalWafers within photonics SOI?"

Do you want to take that, or shall I?

You can start.

We are working closely with GlobalWafers, and so for sure on this topic, like all the other customer as well. As I said on competition, I do not underestimate competition at all, I'm paranoid on this topic, and I expect there will be competition.

Maybe a few words to add to that. Tying back to what Christophe has shared before, for Soitec to get at a level where we are, and photonics is extremely sensitive to details, it takes a lot of know-how.

It will take any competitor that is starting and wants to enter right now a lot of time to get to the level which is required for the most advanced products we're talking about. CPO is a very, very demanding; photonics in general is very demanding.

Yes, competition is stepping up, and I can only copy what Laurent is saying, that's a good thing. I think it will take them a lot more time also to really develop their products to the level which is demanded right now by the leading customers.

Christophe, do you want to spend a bit of time on talking about the differentiation we bring?

On photonics, you mean? I described it a little bit today, but in photonics, everything counts.

More than any other products, all the levels of roughness are so important. The thickness uniformity is pushed to the extreme.

To give you a number, it's 0.2%. There's no many industry where you go to 0.2% uniformity control.

We have also to combine it with some additional levels of crystal engineering. It's the results today of working 10 years on the topic.

We started, I think it's public, we started 2015 with Intel on the photonics, and we learned a lot. These learnings we can apply today to the platforms that we are providing to our customers.

Can we move back to the room, maybe? Alexandre, go ahead.

You have the mic.

Hello. Good afternoon.

Alexandre Tran from CIC. Quick question, very simply.

If we were about to see Soitec achieving $500 million in photonics, I don't know what would be the time by then you would be able to reach that, how many wafers would that be to be honest?

You noticed we are not guiding on the full year. I'm not going to give you a guidance for multiple year.

What I can tell you, as Cyril was mentioning, we have capacity to grow on photonics, we have the chance to have capacity available. We can as well work on our product mix, at some point, some of you mentioned that the company was built or was targeting a pass to €2 billion revenue.

That does not mean we have capacity to achieve €2 billion revenue installed at this stage, but if needed, we will adjust. That's not a topic for today.

Sébastien?

Yes, Sébastien from Kepler, coming back on RF-SOI inventory adjustment. The previous management was talking about some inventory depletion of 200,000 wafers to 300,000 wafers per quarter, maybe 4, 5 quarters to go to normal level of inventories.

Q4, we have not seen any change in the inventories at your big foundry customers. How do you see the inventory depletion from the coming quarters?

Are you still on this kind of cadence, or?

You have the right number. Indeed, 200 to 300 kilowafers per quarter, this is what happened the previous two quarters.

Indeed, this quarter, this was not the case because of seasonality, but as well probably the mobile market dynamics overall with memory shortage. That's the difficult part to assess right now.

Yes, I expect this inventory correction to restart at this stage, but this is dependent as well on topics like memory shortage that are not so easy to forecast. This will take several quarters, for sure.

One follow-up on photonics SOI. Could you comment a little bit on the price dynamic in this very hot market?

How do you see prices trending, and how do you see the profitability of this business? I suppose it is highly profitable, but can you comment on that?

Thank you.

Do you want to take that one, or shall I? Go ahead.

I can start, and you will correct. No.

Obviously, we're not going to break down our profitability by product line, but what we can safely say is that profitability on the photonics line is good, and every time we shift the mix toward photonics, we improve the bottom line and the quality of the business. That's very very clear.

I'm still relatively new or very new in the industry, but it strikes me that how fast things can change. We had a detrimental pricing impact in fiscal year 2026.

It was not always easy to negotiate prices with clients, even in absence of LTAs. Here, the landscape is completely different, and the priority for the client is to get the products in time, secure the capacity.

We can even discuss measures to improve the cash profile of the company. Yes, Soitec has a firmer hand on the pricing in photonics.

I guess here what is important is we focus on delivering to our customer. It starts always with that, being sure we do what we say.

Jakob? Oh, we need the mic to Jakob.

Oh,

François from BNP Paribas. Just a question on fungibility: how much time does it take to switch a fab to a 300 SOI photonics?

Cyril, I'm sure you will be happy to answer this one.

Actually, we have some asset already there, and we can use that asset, which is, for instance, we talking a lot about 300, but we have a demand as well on 200, and we know that we had some idle asset, idle cost, and I'm more than happy to start overnight this asset to supply our customers. This is the same on 300 in Singapore, basically, it's just a matter of qualifying our product with our customers, and we have more and more ongoing right now.

Last item for sure is how we can deploy capacity. Deploy capacity typically in our industry, when we have a new asset to introduce, this takes 9 to 12 months.

We are used to prepare that ahead of time by having a kind of what we call a letter of intention in order to lock the slot and to reduce this amount of time, minimizing the risk at the same time without any commitment from the company. Sometimes, we have to find smart ways to minimize even further this type of timing.

For instance, case-by-case, in the last two weeks, we have decided to transfer specific critical assets from Singapore to Bernin, because Bernin is qualified, Pasir Ris is not fully qualified yet, even if all the customers are qualifying Singapore, in order to benefit of the asset now—every day is important—now, and to buy a new asset for Singapore for the future. We are even able to cope with this type of, let's say, de-hooking a tool in several days, transferring, re-hooking in France.

We don't do that like study tools, but when we have to do it on critical tools, we do it because what matters is to deliver the wafer to the customer.

Jakob, please.

Thank you. I had two questions, please.

Firstly, on the mobile side, you mentioned it a couple of times in your presentation with the satellite and LEO. Could you maybe just help us understand how material an opportunity do you think this is, particularly in terms of, I guess, working through some of these inventory issues on the mobile side?

Then secondly, just to sort of reconcile your comments on the phasing, kind of basically slowing post-Q1, but at the same time also your guidance that photonics is accelerating, can you maybe just help us understand how these two sort of square? Is it just that the photonics acceleration takes maybe a little bit longer, we don't see the full effect this year?

Is it too small? Just to sort of help us understand why is it one bit is accelerating very sharply, but at the same time, you're guiding for what sounds like a deceleration of growth through the year?

Regarding LEO, this is mostly linked to the beamformer in FD-SOI. The number of players on LEO is not very huge, you can draw your own conclusion on that.

That's the first part.

The second part was... It was on revenue growth, Q1 guidance, 15%.

Oh, yeah. How photonics is evolving.

As I was saying earlier, that will be progressive. That's something, yes, we see acceleration of the demand.

This acceleration is pretty recent. It was growing steadily, now the shape of the curve has changed.

This means we will adjust progressively to this new shape. You should expect this.

Maybe adding up a little comment on the guide itself and the comment on reducing seasonality: over the past couple of years, in a high-inventory environment, we've been navigating a little bit less visibility. That probably led to a bit more seasonality, Q-on-Q, and thinking about the last two years, maybe.

We've started to normalize that as well, and seeing RF-SOI inventory correction going the right direction, although we still have work to do, helps a lot in that regard.

These two things will compensate.

Maybe switching to the online questions if I may: "Do you expect AI data center CapEx spending to persist at such high level as seen today?" Steve, if you want to double-click on the change in scale or in the industry, you alluded to?

Yeah. On the overall global CapEx, I think we're reaching about $700 billion this year for the top five US hyperscalers alone.

That's a huge increase from previous years. You need to add also all the neocloud service providers in Europe, you need to add all the CapEx in China.

We don't have yet an outlook to share on this global CapEx, but clearly the demand should continue to be very strong given the different waves of applications ahead of us. More importantly for us, or equally importantly for us, is what's happening within those data centers with technology transitions that we have talked multiple times today and that benefit photonics SOI, LNOI, and later on, of course, the impact on the edge AI ecosystem.

No crystal ball, but as far as we can see, the strong momentum in this overall CapEx spending should continue.

Okay.

Robert Sanders, Deutsche Bank. Two questions.

On the photonics side, can you just discuss a bit the X-FAB, Lucentix partnership on TFLN, how does this play into LNOI business, and is it complementary, and how does it affect your foundry customers? The second question would just be, if CPO has a huge takeoff, would that mean then that your customer set would transition quite rapidly from the large two foundries, specialty foundries, to TSMC?

How would that affect your bargaining power? Thanks.

I'll try to answer your first question, even though I'm not sure I know exactly what you're mentioning, but TFLN is another name of LNOI. The people from the photonics call it thin film lithium niobate when we call it more from a material standpoint lithium niobate on insulator, but well, that's the same thing.

The point is until now, what we deliver, the modulation of the signal is done on the top silicon layer of the silicon photonics wafer we deliver. This works up to about 1.6 TB per second, and the modulation has to be faster, and the silicon is running out of steam.

The people are doing it on lithium niobate. This TFLN chips are done on thin lithium niobate, and then reported on top of the silicon photonics platform.

Whenever we play on LNOI, I'm not saying what you mentioned, but our play on LNOI is complementary. It's complementing what we're doing in silicon photonics, it's not replacing, it's on top, and then we can gain share into the modulation part, okay?

On the CPO,

I will get back actually to questions that was related. Actually, pluggable and CPO are not replacing each other.

That's in complement, the scale-out is more the pluggable, the scale-up is more the CPO. The actors might be different, but it's not replacing the other.

Maybe one last for Alex?

It's going to be two last, sorry. The...

That's four.

Just one is very quick. Is SmartSiC definitely dead, or is there a second life for this product line?

We've seen quite a lot of renewal of interest in silicon carbide these days with some a little bit, I'd say, less price-sensitive customers in data centers that are interested in this product for power. Do you think there is a possibility that SmartSiC comes back?

The second just also very quick: when you say CPO comes on top of the pluggables, I say pluggables, are they going to stay at this whatever, 25%, 30% growth, and then once you get the CPO into the mix, that will accelerate that growth on top, is that what you're implying? Thanks.

I can start with the SmartSiC, and then maybe you take the second one, René. SmartSiC, first, the value proposition for high-performance silicon carbide for SmartSiC is demonstrated.

Yes, there is a benefit for high performance. There are some applications requiring high performance.

The difficulty in silicon carbide is the market landscape changed radically the past two years, three years, with vertically integrated players, and as well with some Chinese suppliers, mono-SiC Chinese suppliers, very aggressive in price. The question really is to reassess, based on this new context, what is the best option and best play for Soitec in this area.

I was mentioning before that we are reviewing our full portfolio overall, not only for SmartSiC, but overall, to be sure we allocate our resources, our energy, on the most promising topics. That's something underway, and we will share a conclusion when we reach them.

Let me take the other part, which is about whether or not CPO is cannibalizing let's say pluggables. The short answer is no, but let me elaborate a little bit more.

What is the trend with CPO is that we are replacing, displacing copper. There is more photonics.

In that sense, it's always incremental. Now, the use case for CPO is mostly scale-up in a rack.

The use case for pluggables is mostly to connect racks with each other and to connect data centers. The short answer to your question is no, it's not cannibalizing, because the amount of data centers is growing, it means that we need even more pluggable transceivers to do that datacom connectivity.

On top of that, we see more and more photonics in the rack itself in the scale-up, so that's an additional, it's incremental.

Okay. I think that marks the end of our Q&A session.

Laurent, if you want to have a few remarks before we... No,

first of all, thanks a lot to all of you for supporting us and being present today. I guess you got our main message.

Fiscal year 2026 has been a difficult year where we made, however, significant progress thanks to our discipline. We made progress regarding free cash flow, we made progress on working capital, this gives us the strong foundation to be able to prepare a progressive growth.

Obviously, this brings me to the second message: yes, we see definitely AI as a growth engine for the company. We are all focused on that to be able to deliver.

That's the key: we have to execute, we have to deliver, we have to stay disciplined. All the team here is focused on that.

That's the second key message. Obviously, beyond AI, we continue to develop our core SOI and POI technology.

More on a personal note now, I can tell you I'm enthusiastic to join this team. The onboarding has been quite efficient, and so thanks to all the guys and ladies of the team here.

I'm confident. Yes, we have challenges, we have opportunities as well, but I'm confident we can address that.

Thanks a lot.

Thank you very much, and a quick date important to note: July 29th, we'll be hosting our AGM and releasing the results of Q1 '27. Thank you very much.