Tower Semiconductor Ltd Q3 FY2025 Earnings Call

Good afternoon, ladies and gentlemen, and thank you for standing by. Welcome to today's Tower Semiconductor Third Quarter 2025 Earnings Conference Call. I must advise you that this conference is being recorded today. I would now like to hand the conference over to your speaker today, Ms. Noit Levy, Senior Vice President of Investor Relations and Human Resources. Please go ahead, madam.

Thank you. Good day, and thank you, everyone, for joining us. Welcome to Tower Semiconductor's Third Quarter 2025 Financial Results Conference Call. With us today are Mr. Russell Ellwanger, our CEO; Dr. Marco Racanelli, our President; and Mr. Oren Shirazi, our CFO. Before we begin, please note that certain statements made during this call may be forward-looking and are subject to risks and uncertainties that could cause actual results to differ materially. These risks are detailed in our SEC filings, Form 20-F and 6-K, as well as filings with the Israeli Securities Authority, all available on our website. Tower assumes no obligation to update forward-looking statements. Our third quarter 2025 results are prepared in accordance with U.S. GAAP. Some data presented may include non-GAAP financial measures as defined under SEC Regulation G. Reconciliations to GAAP figures and full explanations are provided in today's press release and financial tables. For your reference, the supporting slide deck is available on our website and integrated into this webcast. With that, I'd like to turn the call over to our CEO, Mr. Russell Ellwanger. Russell?

Hello, everyone. Thank you for joining this earnings call. We are in the best position, growing our core technologies, Power Management, CMOS Image Sensors, 65-nanometer RF Mobile, each of which demonstrates year-over-year revenue growth, providing an excellent foundation on top of which the extreme AI-driven data center demand for our silicon photonics and silicon germanium RF platforms is driving unprecedented company growth. We ended our third quarter with revenue at $396 million, resulting in a net profit of $54 million. We guide our fourth quarter to be a revenue record of $440 million, plus/minus 5%, fulfilling our beginning-of-year target of quarter-over-quarter growth throughout the year with strong acceleration in the second half. This underscores the increasing demand momentum we see in our served markets and is also the result of further manufacturing capabilities, namely the very first step of a large ramp, having repurposed with added capacity for factories towards new and/or stronger silicon photonics and silicon germanium capabilities. The fourth quarter guidance indeed demonstrates the burgeoning trajectory we are on. In the following minutes, we will present the successes that we share with our customers, driving top and bottom line growth over the years to come. Now to review our third quarter of 2025 revenue breakdown and discuss the key trends, please see Slide 4 as reference. Our RF infrastructure business continues to deliver exceptional growth, increasing its contribution to corporate revenue from $67 million or 18% of corporate revenue in the third quarter of last year to $107 million or 27% for the third quarter of this year. For the full year, we expect this business to grow by 75% with silicon photonics more than doubling from the 2024 $105 million. This significant expansion reflects the strong customer adoption of our advanced technologies and validates our strategic investments in these markets. The momentum we are seeing positions RF infrastructure, silicon photonics, and silicon germanium platforms as a key pillar for our long-term growth, fortified and propelled by deep partnerships-based innovations with the foremost industry titans. Our silicon photonics business grew in the third quarter to $52 million, approximately 70% growth compared to the third quarter of 2024. Market demand for silicon photonics continues to surge, driven by a stronger-than-anticipated ramp in 1.6T products, on top of the robust 400G and 800G demand. We have expanded capacity with our advanced SiPho platforms in Fab 9, San Antonio, having shipped revenue wafers in the third quarter and expecting multiple thousands to be shipped in the fourth quarter of this year. We are in advanced stages of qualifications in Fab 2, Israel, expecting our first production shipments in the first quarter of 2026. In 300-millimeter, we have started wafer production for the innovative receipt products we announced last quarter and anticipate revenue contribution from 300-millimeter silicon photonics to start in the fourth quarter of this year. Our capacity growth is fully aligned with and spoken to by our customer demand outlook. Silicon photonics continues to increase market share over EML solutions, given its significant cost advantage. SiPho typically requires half the number of lasers as an equivalent EML product with performance benefits, especially seen at 1.6T. As such, we anticipate this market share shift to be permanent. Hence, we are, at this point, going to add additional CapEx to address an even increased surging demand. Looking at next-generation 3.2T data rates, which will require a doubling of speed for each lane from the current 200 gigabits per second to 400 gigabits per second, we have multiple programs with industry leaders to both extend silicon capability, but we are also pursuing integrating indium phosphide modulators for our previous announcements with OpenLight and are investigating other material systems to ensure that our customer partnerships are not just ready, but leading the transition to next-generation requirements for 3.2T and 6.4T. In the past quarter, in partnership with Xscape Photonics and NVIDIA-backed start-up, we delivered the industry's first optically pumped on-chip multi-wavelength laser platform for AI data center fabrics. This innovation further expands our participation in the laser market, particularly for co-packaged optics applications, a significant adjacent opportunity, leveraging our high-volume SiPho platforms. We showcased these advancements along with others at a highly successful Tower Technical Symposium event in China with approximately 300 attendees, where eOptoLink delivered the keynote addressing Tower's role in supporting phenomenal market growth. Later this month, we anticipate another great TGS event in Santa Clara with Broadcom's President, Charlie Kawwas, delivering the invited customer keynote. Looking at silicon germanium, silicon germanium transimpedance amplifiers and laser drivers are essential components for optical transceivers. The growth in SiGe demand is a function of data center build-outs, be it SiPho or EML-based solutions with an additional accelerator, the adoption of linear pluggable optics. Due to the elimination of the DSP in the LPO module, LPO requires both the driver for transmit and the TIA for receive to have an added function of continuous time linear equalizer, which significantly adds to the silicon area of each of the TIA and drivers, hence, nicely increasing the amount of silicon needed per unit. Multiple SiGe customers have begun material LPO production volumes, indicating a clear upward trend in this market. We started silicon germanium production in Fab 2 of our most advanced SiGe platforms and are seeing eager adoption by these customers, driving meaningful additional contributions to SiGe capacity and shipments projected to ramp throughout 2026 and delivering high volumes thereafter. In addition to infrastructure, we have secured a new silicon germanium wall noise amplifier designed for a Tier 1 handset customer with an initial ramp in Q4 '25 and then proceeding through 2026 and beyond, adding a significant new growth opportunity to our existing leading market share in optical transceivers as discussed, itself being a high-growth market. Moving to RF Mobile, it represented about 26% of our Q3 '25 corporate revenue. RFSOI has shown steady quarter-over-quarter demand increases with our more advanced 65-nanometer 300-millimeter platform at higher than 20% increase in the second half of 2025 over the second half of 2024. We released this quarter an updated RFSOI technology that not only provides double-digit better Ron-Coff relative to our competitors as measured by multiple Tier 1 customers, but also reduces layer count by 15%, improving our customers and our margins, hence, enhancing our market share. This technology also allows customers the freedom to make a trade-off between the better Ron-Coff to enable a smaller die size or to have higher power handling. As such, we are seeing strong customer traction, providing much confidence in multiyear growth. This quarter, we made important advancements in our sensor and displays technologies, which represented 14% of our third quarter corporate revenue and expected to show mid-teens full year-over-year growth. We received our first production PO for Q1 '26 shipments for OLED display backplane silicon and continue to enhance this offering, having added high-speed logic and high-speed SRAM capabilities, enabling support for 120-hertz refresh rates, critical for next generations of VR and MR applications. Medical and photography sensor revenue remained stable, while the majority of our growth and strongest position is in machine vision, where we supply the second largest player in this market in addition to other key customers. Power management represented 17% of our third quarter corporate revenue. Our power business has performed well, targeting a year-over-year growth of 15% with disproportionately higher growth for advanced 300-millimeter platforms, one driver of which being the strong ramp of the handset envelope tracker volume expected to continue through the next multiyears. Targeting the growing market of data center power, we have recently demonstrated 16-volt operating voltage devices with more than 40% lower Rdson than prior technology and introduced new elements to our 1.2, 3.3-volt 65-nanometer BCD flow, improving power conversion efficiency aligned to our key customer needs. The wireless charger IC market is growing rapidly and demanding higher voltage LDMOS. To that end, working closely with lead customers, we have provided a 40-volt extension to our 300-millimeter BCD process. Specific to automotive and battery management applications, we have multiple engagements for a differentiated 140-volt reserve flow allowing higher voltages without the need for the added high expense of SOI substrates. We've continued to add to the competitiveness of this platform, greatly reducing the cell size through added features and optimized architecture. Moving to utilization. At the beginning of the year, we announced a repurposing of several of our factories, predominantly towards higher capacities for RF infrastructure, namely silicon germanium and silicon photonics. To update on the progress, qualification and initial ramps are going well with hundreds of silicon germanium wafers shipped in Q2 and thousands in Q3 from Fab 9 in San Antonio. We met our first internal milestone of customer SiPho shipments in Q3 from San Antonio with several thousands planned to ship in Q4. Fab 2, Migdal Haemek, Israel is on track to ship our first and meaningful number of silicon germanium wafers in Q4 '25 and silicon photonic wafers in Q1 '26. Additionally, we expect to see our first production revenue for SiPho at 300-millimeter in Fab 7 in this present quarter, Q4. Therefore, while we are in the final qualifications and initial ramp of the repurposed fabs and added capacity, our third quarter utilization levels, Fab 2 in Israel operated at about 65% utilization. Fab 3 maintained our model full utilization of 85% with growing activities for SiPho capacity. Fab 5 was at 75% utilization. Fab 7, 300-millimeter was fully utilized, well above our 85% utilization model. Fab 9 was at 60% utilization. In summary, we are in a strong position with all our core technologies demonstrating year-over-year revenue growth, the right technologies growing with the right customers. On top of this, our long-term silicon germanium leadership for optical transceivers, coupled with correct market insights, namely believing in the benefits of silicon photonics, having begun 8 years ago with the right partner and adding those who have become the most significant adopters of this technology has enabled us to be in the lead position for silicon photonics manufacturing and development. This has proven timely to meet the soaring demands of data center technology road map and build-out. A present and future pathway for unprecedented growth for Tower. In close collaboration with our customers, we have advanced both capacity increases and technology road map deliverables real-time, addressing the soaring requirements for AI infrastructure. Specific to demand-driven capacity expansions, we target 2025 silicon photonics revenue to be above $220 million, up from $105 million in 2024, and very importantly, at a Q4 '25 annualized revenue run rate exceeding $320 million. The $320 million SiPho run rate is enabled by the very first steps in qualification and ramp of the previously announced $350 million investment. We have begun an additional investment of $300 million for further substantial SiPho capacity expansion and next-generation capabilities in Fab 3, Fab 9, Fab 2, and Fab 7, this investment targeted to achieve full volume in wafer starts in the second half of 2026. The total capacity is fully aligned to and directly requested by our customers. The resulting capacity should increase our SiPho shipments by over 3x against our targeted fourth quarter '25 qualified utilized capacity. With that, I'd like to turn the call to our CFO, Mr. Oren Shirazi. Oren, please.

Hello, everyone. Earlier today, we released our quarterly financial results and balance sheet. For the third quarter of 2025, we reported revenue of $396 million, reflecting a year-over-year revenue increase of 7% and a quarter-over-quarter revenue increase of 6%. Gross profit for the first quarter was $93 million, 16% higher compared to $80 million in the second quarter, and operating profit was $51 million, 27% higher sequentially compared to $40 million in the second quarter. Net profit for the quarter was $54 million, 15% higher compared to net profit of $47 million in the second quarter and earnings per share were $0.48 basic and $0.47 diluted as compared to $0.42 basic and $0.41 diluted earnings per share reported for the second quarter. As mentioned in today's press release, to address the continuous and growing SiPho and SiGe demand and given the full utilization of our Newport Beach fab, we are extending the Newport Beach fab lease by up to an additional 3.5 years beyond its previous 2027 term. An upfront lease payment of $105 million will be recorded as cash used for operating activities in our Q4 '25 statement of cash flows with a corresponding impact on our balance sheet cash amount, while the resulting P&L impact would be, as announced earlier today, $6 million per quarter to be recorded over a 5-year period as required by GAAP in the COGS line. I would like now to describe our currency hedging activities. In relation to the Japanese yen, since the majority of TPSCo's revenues are denominated in yen and the vast majority of TPSCo's costs are in yen, we have a natural hedge over most of our Japanese business and operations. To mitigate part of the remaining yen exposure, we are executing zero-cost cylinder transactions to hedge the currency fluctuations. Hence, while the yen rate against the dollar may fluctuate, there is limited impact on our margin. In relation to the Israeli shekel currency, while we have no revenues in this currency, since a portion of our cost in Israel is denominated in shekel currency, we also hedge a large portion of such currency risk by engaging in zero-cost cylinder transactions to mitigate this exposure. Hence, while the shekel rate against the dollar may fluctuate, the impact on our margins is limited. Moving to our balance sheet and future cash and cash CapEx and cash plans. As I noted earlier, our balance sheet remains very strong as evidenced by the following indicators and financial ratios. As of the end of September 2025, our assets totaled over $3 billion, primarily comprised of $1.4 billion in fixed assets net, mainly comprised of fab machinery, and $1.8 billion of current assets. The current assets ratio is very strong at about 7x, while shareholders' equity reached a record of $2.8 billion at the end of September 2025. Our strong financial position allows us to invest in strategic opportunities that support our corporate vision as follows: for our high-margin SiPho and SiGe business, we previously announced plans to invest $350 million to expand our capacity in our 8-inch fabs in Israel and Texas and in our 12-inch Uozu fab in Japan. This CapEx includes a large portion of capability CapEx for advanced development and high-end RF technology-related projects. 50% of this amount has been paid to date, while the remaining 50% is expected to be paid in the coming year. In addition, as we announced earlier today, we have decided to allocate an additional $300 million investment for capacity, growth, and next-generation capabilities, mainly for machinery for additional SiPho and SiGe capacity growth for our 8-inch fab and for our 12-inch fab. This would put total SiPho and SiGe capacity and capabilities related CapEx plan at an aggregate of $650 million. All of these investments are fully reflected in our previously presented strategic and financial model. Under this model, we are targeting $2.7 billion in annual revenues at full loading of our existing fab and qualified capacity, including the previously stated capacity expansion plan, $560 million in annual operating profit, and $500 million in annual net profit. That concludes my prepared remarks. Now I'd like to turn the call back to the operator so we can take your questions.

And we're going to take the first question, and it comes from the line of Cody Acree from Investment Bank.

Congrats on the progress. Oren, if I can just get a quick clarification. You said that the incremental $300 million was already considered in your $2.77 billion total revenue expectations long term. Is that right?

Yes, yes. It may mean that we will achieve this target earlier than somebody previously expected. But yes, it is included.

Okay. So if no incremental upside, then what's the accelerated pace do you expect? I guess, what's the give and take of that extra CapEx?

Acceleration of achievement towards the $500 million net profit run rate, which, as you know, we are still not there. So we will accelerate the achievement. And of course, the accelerated achievement will enable higher profit sooner.

Okay. Great. And then maybe, Russell, can you just talk about some of the applications that you see driving the aggressive growth you're seeing in RF infrastructure?

Yes. The biggest and strongest is just really the need for build-out, specifically, I think, AI-driven, but it continues for high volumes of 400 gigabits per second, a very high volume of 800 gigabits per second in multiple formulations of it, both DR8 and 2xFR4s. And then as stated, a very high volume right now going into 1.6 gigabits, where we're seeing somewhere about 30% of all of our starts being dedicated to that platform presently. So it's really just for the continual build-out of data center and a big movement right now going into the 1.6G.

Now we'll go and take our next question, and it comes from the line of Tavy Rosner from Barclays.

Congratulations on the strong results. I wanted to ask 2 quick ones on the silicon photonics, please. You mentioned the leading position of Tower. So who do you see as your main competitors these days? And given the supply-demand imbalance at the moment, are you able or considering raising prices?

Abel is probably the answer to that would be yes. But considering no, we're very close with our customers. We understand what their needs are. We have long-term road maps, and we're not opportunistic, if you would look at it that way, because demand is very, very tight, to gouge somebody for an extra couple of wafers. I think that's the surest way to lose goodwill and partnership. What we are seeing, however, is extremely strong demand. And having stated that, that's the reason for after having invested $350 million, which we're seeing the first signs of that ramp right now to increase another $300 million of investment. I had stated that we're targeting a Q4 SiPho revenue shipment run rate of over $320 million. And against the qualified capacity that we're shipping in Q4, we're increasing that by over 3x in start capability within the next 4 quarters, expecting to have that entire 3x plus available for starts in the second half of 2026. So that's quite a bit, if you were to say 3x of a $320 million run rate, that's quite a substantial growth in SiPho that we're projecting for ourselves. So having come from, what, plus/minus $28 million in '23 to $105 million in '24. So this year, we're targeting over $220 million for the full year, with a $320 million plus run rate in Q4, but bringing that to well above $900 million by target. And that target is really spoken to by customers. That's not field of dreams, if you build it, they will come. That's customers saying, please build it, we need the capacity. And so I think we sit very strong with our relationships with SiPho. Now also I stated that in the script that we're doing quite a bit on capability, not just for the present 1.6T generation, but to make sure that our customers are in a leading position for the capabilities needed for 3.2T or for 6.4T, specifically the needs for a faster capable modulator of 400G. And those activities are very real-time. So we're investing real-time in increasing capacity, which is really being demanded by the industry. And fortunately, as stated, we are by far in the leading position on manufacturing, but believe also to be in the leading position as far as developing next-generation platforms with the leaders so that we're both prepared well before the demand actually arises.

Yes. Just to clarify the first part, do you see any changes in competitive dynamics? Anyone else deploying capital in that field to try and take some share away from you guys?

I believe that most people would like to take share from us. It's a good growth market with good strong customers. The point is really to the question that you asked, to be opportunistic on pricing would be probably a good invitation for our customers to say, hey, we don't want to be a long-term partner, go look for someone else that you can leverage us with pricing on, and they don't have to do that. We're working very closely with our customers to be reasonable and to have win-wins on both sides. So, but yes, I'm sure that there are others that are trying to eat into where we're at. It's very difficult, though, for somebody to break into our position right now.

Yes. I think on the pricing discussion, as you say, we're not taking advantage of the situation just because capacity is tight in the industry today. But we are adding value in our advanced platforms. And so in that regard, we do price higher technologies that deliver more value to customers. So in that aspect, over time, we do anticipate some price improvement as customers migrate to these more advanced technologies. And I will say as well that SiPho is already very accretive in margin. So we do get paid for the value that we add, and that's how things should be based. We stated very strongly that SiPho has a very strong benefit for our customers against EML. So for our customers to be using our platforms now, and if you look at a halving of lasers, it's a big deal. You have as well at this point then for the 1.6T, the use of a silicon modulator that's inside of the PIC itself versus needing an indium phosphide modulator that costs much more money and quite a bit of 3.5 area. So there's value in the platforms. And obviously, we both share in whatever value is created.

And it comes from the line of Richard Shannon from Craig-Hallum Capital Group.

Congrats on some very nice numbers here. Let me start off with a few questions here on silicon photonics. Russell, you made an interesting comment that I probably didn't transcribe in my own notes here very well, but you mentioned something about the shift to silicon photonics is permanent. Can you explain what you mean by that, please?

What we stated in the script, it's very cost conducive against EML. It takes half the lasers. So if you look at twofold, the cost of 3.5 and also the capacity constraint of 3.5, that's both a big benefit to be able to use silicon photonics at the 400, 800, 1.6T, and potentially, depending on development, even at 3.2T, going to SiPho, you have a silicon modulator, which is also very cost conducive against needing to have a 3.5 modulator. So yes, by stating that there's a cost benefit that certainly drives long-term stickiness. But in addition to the cost benefit, there's performance benefit as well. And when you combine cost and performance, that's really an absolute winning combination for market share stickiness, right?

Okay. My second question is looking at 1.6 here. Obviously, we're seeing module makers talk about some nice ramps you're starting next year. What kind of mix do you expect to have of 1.6 versus slower speeds in your business, I don't know, next year or in a particular point? Just trying to get a sense of how fast this is scaling versus the older slower technologies.

Right now, the 1.6 is close to 1/3 of our starts. So that gives you a feel for where we're at right now, and that's up from almost nothing at the beginning of the year, but a very quick ramp movement to the 1.6T. I would expect that it will go to over 50% within the first few quarters of the next year.

Okay. That is helpful. My last question on silicon photonics is about your comments regarding reaching 3x capacity from the current fourth quarter run rate. What time frame do you anticipate for achieving full utilization? It seems like it could be in 2026, but I would like to know your expectations on this.

The demand is there or will be there by customer forecast. By planning, we should be fully installed within the first half of '26 and having all the tools up and running, being able to hit the full start capability within the second half. So the shipment level really depends on what point in the second half we do the wafer starts. You're typically looking at somewhere of a 3- to 4-month cycle between the starts and the shipment, given whatever the PO size is. So we certainly foresee seeing a portion of this increased capacity coming into revenue within the second half of '26. But take into account that the first ramp that we're doing has not been realized yet either. So the previous $350 million investment is in the first stages of ramping. So within the first half and predominantly in the second quarter, we should see a very big pickup in our silicon photonics shipments revenue, and that should continue in the third and fourth quarter.

Okay. Perfect. One last question for me, I'll jump out of line. Oren, can you talk about kind of gross margin fall-through in the next few quarters, obviously, knowing that silicon photonics is a margin-accretive business for you, and it sounds like that will be your major driver here. How do we think about this going forward here? And I want to get a sense of also when additional depreciation builds in here to think about that going forward?

Yes. So I think currently, the gross profit in Q3 was 24%, $93 million over $396 million, that's the actual number. And it should be better. As you see our long-term financial model, which has higher percentages. Usually, we speak about incremental margin of 50%. And of course, because of the cycle, it will be higher nicely, and it will be offset by 2 elements. One is the Newport Beach lease amount that we said we will pay a total of $6 million. And the second is what you mentioned here correctly, the depreciation from the additional CapEx. The total additional CapEx is $600 million over 15 years. So it's about $10 million a quarter. But some of that already started. So it's a gradual ramp towards the $10 million.

Now we're going to take our next question, and it comes from the line of Mehdi Hosseini from SIG.

A couple of follow-ups, and I apologize, I joined the call late. I apologize if I'm repeating some of the questions already covered. Russell, when you look at the opportunities associated with transceiver, are you also baking in increased content? In other words, I'm under the assumption that most of the opportunity is currently on the transmit side and whether the receive side of the transmitter would also give you the opportunity to increase content.

To increase volumes, we have a very promising platform that will be shipping 300-millimeter receive SiPho in the fourth quarter. We plan to explore several formulations beyond what we're currently doing in 2026. The receive market is an additional opportunity for us, as currently, everything we're focusing on, aside from these first 300-millimeter activities, is related to transmit. That's a great question.

Is there any way we could kind of think about how this SiPho capacity increase by 3x would be split into increased content versus units of transceiver shipped to the end market?

I'm not exactly sure what you mean by increased content, but it relates to the units being used in transceivers. Please continue.

As you increase the mix of receive and transmit within the same transmitter, how much of that is incorporated into the capacity increase of three times?

Right now, the increase of 3x is including very little receive. It's mostly all transmit.

So would you need to increase capacity again as your customers migrate to receive SiGe with SiPho?

Yes, and we'd be really thrilled to do so. We have platforms that are doing it. I mean, it's not that we're not proactive against it. It's that right now, the transmit demand is so high. But yes, receive, but receive really comes into very strong capabilities with DEMuX, and that's a big area that we're focused on.

Okay. Just a quick follow-up. In the longer term, where are we with packaging? I think a couple of quarters ago, you highlighted doing R&D so you would extend your addressable market to include some packaging. Is there an update you can share with us?

We have an activity with a leading packaging house focused on formulating a CPO, a full CPO. We certainly have other activities around NPO, big activities with through silicon via that's required for NPO and potentially for CPO as well. So we're making progress there. It's not that there's a lot of CPO being used right now. It's not, but we're making good progress. We also have other activities driving additional really strong capabilities. We're talking multiple generations down the road, but a really big program focused on 51.2T to have a very, very advanced modulation that would possibly be required within a PIC structure for whatever CPO would be used.

Okay. If these opportunities turn into revenue later this decade, it would take a couple of years for commercialization.

I think it's a couple of years before CPO is strongly commercialized period, independent of Tower.

And the question comes from the line of Lisa Thompson from Zacks Investment Research.

I was wondering if you could talk a little bit about 3.2T. It seems like the technology may not be able to go from 1.6 to 3.2. Can you tell us kind of where they are specifically with trying to solve that problem? And is there still a risk that an entire new technology might be needed?

I don't think an entire new technology will be needed. There are certain issues even around the DSP that's being worked on. But no, I don't think a new technology. From our standpoint, the thing that's really required is the modulation, as mentioned at 400G. And as stated in the script, we're working on 3 different pathways depending on customers' needs and desires to do the 400G modulator. So from the pure modulation standpoint, I think we're addressing it very strongly with very good progress from the entire transceiver, the build-out or CPO, I mean those are other issues, but I don't think that they're not attackable or solvable. So I don't think that 3.2T will be held up.

Okay. And let me just clarify, is the revenues from SiPho totally gated by capacity? I mean is the demand there if you can build it?

Yes.

Okay. And then one small question. When are you going to start reporting Agrate utilization? Is that way far off?

It needn't be. We could start reporting it any time. We just haven't. We're in the midst of the first ramp there. So, but we could start reporting it. There's no reason that we wouldn't.

And now we're going to take our last question for today. And it comes from the line of Richard Shannon from Craig-Hallum Capital Group.

Great. Let me ask one more question here. Russell, kind of big picture, looking across your business with the obvious angle also inclusive of silicon photonics. How do you think about 300-millimeter? You obviously have some through a couple of joint ventures or whatever the right term is there with Intel and ST and then have some capacity in Japan here. But how do you think about acquiring more of that? It seems very important for you to have here and obviously very important for silicon photonics. It's clearly a big growth driver here. Is this something where you can find more partnerships like you have with the 2 partners you've already announced in the past? Or do you see greenfield? Or how do you intend to address that?

Richard, it's an excellent question, and it's one that we're focused on. No, I don't think that we'll be addressing it through partnerships such as what we have with ST or what we have with Intel. We'd be addressing it somewhat organically should we go forward there. But I think it's an excellent preface to be excited for our Q4 release, and we'll be talking more to it at that point.

That does conclude our Q&A session. I would now like to turn the call to Mr. Russell Ellwanger for any closing remarks.

Well, truly, thank you for continued interest in Tower. Thank you for the support of Tower. Certainly appreciate the questions that were asked during the call. Really, we're very excited to update you over the coming quarters as this more than 3x capacity that we talk about for the SiPho, SiGe growth as that comes online and to update you about the other progress and activities. Again, very good questions about how this CapEx investment impacts the financial model, where Oren mentioned about it bringing in the timeline of reaching certain revenue levels, and it's accelerating the timeline due to the SiPho being mentioned as accretive margins, doing it at a somewhat different financial model. So we're very excited to update on all of these things. And as we get ready and prepare and give the end of year comments, which is always a summary of the year and an outlook of what we're doing in the coming years, I think there are many exciting things that we'll be talking about with you about the direction of the company and how all of these accretive actions really turn out to be a very strong benefit and a strong ROI. If we consider our initiatives with SiPho concerning capital expenditures, an impressive aspect is that once we begin shipping wafers, the return on investment is realized within half a year from the time we start using the tools, not from when we order the capital expenditures. The ROIs are very rapid after we qualify and start shipping wafers. As Oren mentioned, we are undertaking a significant expansion driven by customer demand for additional capacity, and the swift turnaround in this market enhances our revenue timing considerably. We'll provide more detailed insights in our end-of-year and forward-looking statements during the Q4 release. I'm very eager to discuss this, as it's an exciting period for the company, our customers, and our partnerships, where clients rely on us for their solutions and business growth. We're also thrilled to host our 2025 Technical Global Symposium in Santa Clara next week on November 18, which is open to all customers. We are pleased that many of you will attend. Dr. Charlie Kawwas from Broadcom will deliver an invited talk, which promises to be engaging. I've heard him speak before, and he is an excellent speaker with extensive knowledge. Dr. Racanelli will provide a comprehensive overview of our technologies, and I'll introduce the session by discussing Tower's culture, our direction, and the unique partnerships we cultivate with customers. Additionally, we'll participate in the 23rd Barclays Annual Global Technology Conference in San Francisco on December 10 and the 28th Annual Needham Growth Conference in New York on January 13 and 14. Any investors or analysts wishing for further discussions with our company should contact Noit or anyone else, and we will gladly accommodate your requests. Thank you, and I'm looking forward to this exciting time ahead.

This concludes today's conference call. Thank you for participating. You may now all disconnect. Have a nice day.