Tower Semiconductor Ltd. (TSEM) Earnings Call Transcript
November 10, 2025
Earnings Call Speaker Segments
Thanks. Hello, and I appreciate all of you being here. This is Tower Semiconductor's conference call covering financial results for the third quarter of 2025, and you are welcome to it. Joining us on this call are our CEO, Mr. Russell Ellwanger; our President, Dr. Marco Racanelli; and our CFO, Mr. Oren Shirazi. Prior to starting, we want to point out that a number of remarks on this call might be forward-looking, and they carry risks and uncertainties that may lead actual outcomes to differ materially. Those risks get spelled out in our SEC filings, Form 20-F and 6-K, along with documents filed with the Israeli Securities Authority, and every one of them can be found on our website. Tower undertakes no duty to revise forward-looking statements. The results we are reporting for the third quarter of 2025 were put together following U.S. GAAP. A portion of the data shown might contain non-GAAP financial measures, which SEC Regulation G defines. Today's press release and financial tables contain reconciliations to GAAP figures along with complete explanations; for your convenience, a supporting slide deck sits on our website and is built into this webcast. Having said that, I would now like to hand the call over to our CEO, Mr. Russell Ellwanger. Russell.
Good day, everyone, and thanks for being here on this earnings call. We find ourselves in the strongest spot for expanding our foundational technologies — power management, CMOS image sensors, and 65-nanometer RF mobile — and every one of them is showing revenue growth versus a year ago, which gives us a superb base. On top of that base, the extraordinary demand coming from AI-driven data centers for our Silicon Photonics and Silicon Germanium RF platforms is propelling company growth unlike anything we've seen. Our third quarter closed with revenue of $396 million, which produced net profit of $54 million. Looking at Q4, our guidance points to an all-time high in revenue of $440 million, give or take 5%. That meets the goal we laid out at the start of the year: growth each quarter versus the one before it throughout the entire year, along with a notable pickup during the back half. What this reflects is strengthening demand momentum in the markets we serve, and it also comes from extra production capacity — more precisely, the first phase of a major ramp, after we converted and expanded factory space for new and/or enhanced Silicon Photonics and Silicon Germanium offerings. This Q4 guidance really captures the upward path we're on. In the coming minutes, we'll cover the wins we've achieved alongside our customers that are set to fuel both revenue and profit expansion in future years. To review the numbers for our third quarter of 2025 and discuss the main trends, please refer to slide four. Our RF infrastructure segment continues to post impressive gains, increasing its share of total company revenue from $67 million—equal to 18% of corporate revenue in the same quarter a year ago—up to $107 million, which represents 27% for this year's third quarter. Looking at the entire year, we anticipate 75% growth for this business, while Silicon Photonics is expected to more than double compared with the $105 million it reached in 2024. The scale of this growth demonstrates that customers are embracing our cutting-edge technologies in a big way, and it validates the bets we've placed in these markets. What we're witnessing in terms of momentum cements RF infrastructure, Silicon Photonics, and Silicon Germanium platforms as a core driver of our long-term expansion, reinforced and propelled by innovations built on deep partnerships with the leading giants of the industry. In the third quarter, our Silicon Photonics business expanded to $52 million, which represents roughly 70% growth versus Q3 of 2024. Demand from the market for Silicon Photonics continues to climb sharply, fueled by a 1.6T ramp that came in hotter than we expected, on top of solid demand for 400G and 800G. Our advanced SiPho platforms at Fab 9 in San Antonio have given us more capacity — we shipped revenue wafers there during the third quarter, and we expect multiple thousands to go out in this year's fourth quarter. Over at Fab 2 in Israel, qualifications are at an advanced stage, and we look for our first production shipments there in the first quarter of 2026. Turning to 300mm, wafer production has begun for the innovative receipt products we told you about last quarter, and we anticipate revenue contribution from 300mm Silicon Photonics beginning in the fourth quarter of this year. This capacity growth lines up completely with what our customers' demand outlook tells us. When it comes to Silicon Photonics, it keeps gaining market share against EML solutions because of a substantial cost advantage. An equivalent EML product usually needs twice as many lasers compared to SiPho, and the performance gains are particularly noticeable at 1.6T. Because of that, we expect this shift in market share to stick around permanently. So at this stage, we're going to invest more CapEx in order to meet demand that's surging even higher. If you think about what 3.2T next-gen speeds require, the rate per lane has to go up twofold — from 200 gigabits per second, which is where we are now, to 400 gigabits per second. We've got multiple efforts in flight with industry leaders aimed at both advancing silicon performance and moving toward indium phosphide modulator integration, building on what we announced earlier with OpenLite, and we're also looking into other material systems. The goal is that our customer partnerships aren't just ready — they're out in front, driving the shift toward what 3.2T and 6.4T will demand. Last quarter, together with Escape Photonics, a startup that NVIDIA backs, we delivered the first optically pumped on-chip multi-wavelength laser platform in the industry for AI data center fabrics. This advance extends our reach within the laser sector even further, particularly when it comes to co-package optics use cases — a significant nearby possibility that builds on our high-volume SiPho platforms. Along with other items, we showcased these achievements at a highly successful Tower Technical Symposium gathering in China, which attracted about 300 people and featured a keynote from Eoptolink about the role Tower played in driving extraordinary market expansion. Before this month ends, we anticipate another outstanding TGS event in Santa Clara, where the invited customer keynote will come from Broadcom's President, Charlie Kawwas. Shifting to Silicon Germanium, transimpedance amplifiers and laser drivers made from Silicon Germanium are essential elements for optical transceivers. CIG demand growth comes from data center build outs, whether the solutions are SiPho or EML-based, plus one more accelerator, which is the adoption of linear pluggable optics. Since the DSP is removed from the LPO module, LPO needs the driver on the transmit side and the TIA on the receive side to each carry an added continuous time linear equalizer function, and that adds considerably to the silicon area of every TIA and driver, which conveniently raises how much silicon is required per unit. A number of CIG clients have reached significant LPO production levels, which points to a distinct upward movement in that market. In Fab 2, we kicked off Silicon Germanium manufacturing for our most sophisticated CIG platforms, and lead customers are embracing it with real enthusiasm. That is generating considerable extra contributions to CIG capacity and shipments, with a ramp anticipated throughout 2026 followed by large volumes thereafter. Outside of infrastructure, we secured a new Silicon Germanium low-noise amplifier designed for a Tier one handset customer, beginning its initial ramp in Q4 2025 and carrying on through 2026 and beyond. This brings a sizable fresh growth opportunity in addition to our current leading market share in optical transceivers, which, as previously covered, is itself a rapidly expanding market. Turning to RF mobile, that represented about 26% of corporate revenue in Q3 2025. Demand for RF SLI has grown consistently from quarter to quarter, and our more advanced 300mm platforms at 65 nanometers climbed over 20% in the second half of 2025 versus the same period in 2024. We introduced an upgraded RFSOI technology during the quarter. As multiple Tier one customers have confirmed, it provides double-digit improvements in RF CMOS relative to competitors, and it reduces layer count by 15%, which helps both our customers and our margins, and thus reinforces our market share. With this technology, customers can also choose between superior RF CMOS for a smaller die size or for greater power handling, so we're experiencing robust customer interest, which gives us considerable confidence in growth over multiple years. In sensor and display technologies, we achieved significant progress this quarter; they accounted for 14% of corporate revenue in the third quarter and are anticipated to deliver mid-teens growth for the full year on a year-over-year basis. For OLED display backplane silicon, we booked our initial production PO for shipments in Q1 2026, and we keep improving the offering, having introduced high-speed logic along with high-speed SRAM capabilities that make 120 Hz refresh rates possible, which matters greatly for upcoming generations of VR and MR applications. Revenue from medical and photography sensors stayed level, whereas the bulk of our expansion and our strongest standing come from machine vision, where we serve the market's second-largest participant as well as other important customers. Power management accounted for 17% of corporate revenue in the third quarter.
In our power segment, results have been solid, and we're aiming for a 15% increase versus last year, with even sharper gains coming from advanced 300mm platforms. One factor behind that is the strong ramp of handset envelope tracker ETC volume, which we expect to carry on over the coming multi-years. As we target the expanding data center power market, we recently showed 16V operating voltage devices featuring RDSON more than 40% below that of earlier technology, and we also brought in new elements for our 1.2 to 3.3V 65-nanometer BCD flow, boosting power conversion efficiency in line with what our key customers need. The wireless charger IC market is expanding quickly and calling for higher voltage LDMOS. So, collaborating closely with lead customers, we delivered a 40V extension for our 300mm BCD process. For automotive and battery management uses specifically, we have several engagements around a differentiated 140V reserve flow, which supports higher voltages while avoiding the extra high expense of SLI substrates. We keep enhancing this platform's competitiveness, substantially shrinking cell size via added features and an optimized architecture. Turning to utilization, early in the year we announced that several of our factories would be repurposed, mainly toward greater capacity for RF infrastructure, specifically Silicon Germanium and Silicon Photonics. To give you a sense of where things stand, qualification and the early ramp stages are moving along well. Out of Fab 9 in San Antonio, we shipped hundreds of Silicon Germanium wafers in Q2, and that figure reached the thousands in Q3. During Q3, we also reached our initial internal milestone for customer SiPho shipments from San Antonio, and we expect several thousand to go out in Q4. Fab 2 in Migdal HaEmek, Israel is still on schedule to deliver our first significant volume of Silicon Germanium wafers in Q4 2025, followed by silicon photonic wafers in Q1 2026. Looking further ahead, we expect to see our first production revenue from SiPho at 300mm in Fab 7 in the current quarter, Q4. During this period, we're completing final qualifications and beginning the initial ramp on fabs we've repurposed along with capacity we've added. Here's how utilization looked in Q3: our Israeli Fab 2 came in around 65%; Fab 3 maintained our 85% model for full utilization, with expanding activity around SiPho capacity; Fab 5 reached 75%; Fab 7, the 300mm line, ran at full utilization, considerably higher than the 85% model we use; and Fab 9 registered 60%. To sum up, that's quite a place to be — every one of our core technologies posted year-over-year revenue growth, and the right technologies are expanding alongside the right customers. Beyond that, we've held a leading position in Silicon Germanium for optical transceivers over the long haul, and that's been paired with getting the market read right — specifically, we bought into what Silicon Photonics could deliver, we started down that road eight years back alongside the right partner, and we brought in the companies that have since turned into the biggest adopters of this technology. Together, that's put us clearly ahead when it comes to both manufacturing and developing Silicon Photonics. The timing has worked out well, letting us serve the skyrocketing requirements of the data center technology roadmap and its buildout. That gives Tower a path — today and going forward — toward growth on a scale we haven't seen before. By collaborating closely with our customers, we've advanced capacity expansion and technology roadmap milestones at the same time, in real time, to handle the booming needs of AI infrastructure. When it comes to expanding capacity based on demand, our goal is for 2025 Silicon Photonics revenue to exceed $220 million, rising from $105 million in 2024, and crucially, with a Q4 2025 annualized revenue run rate surpassing $320 million. That $320 million SiPho run rate is made possible by the initial qualification and ramp phases of the $350 million investment we announced earlier. We've kicked off another $300 million investment aimed at significantly expanding SiPho capacity further and adding next-generation capabilities across Fab 3, Fab 9, Fab 2, and Fab 7; this investment is aimed at reaching full volume in wafer starts during the second half of 2026. All of this capacity is completely in line with what our customers have directly asked for. The capacity that results ought to boost our SiPho shipments by more than 3X compared to our targeted Q4 2025 qualified utilized capacity. Having said that, I'd now like to hand the call over to our CFO, Mr. Oren Shirazi. Oren, the floor is yours.
Hi everyone. Earlier today, we put out our quarterly financial results along with our balance sheet. In Q3 of 2025, revenue came in at $396 million, which represents a 7% rise compared with the same period a year ago and a 6% gain versus the prior quarter. Gross profit during the first quarter reached $93 million, up 16% from $80 million in Q2, while operating profit stood at $51 million, a sequential improvement of 27% over the $40 million recorded in Q2. Net profit for the quarter totaled $54 million, an increase of 15% from Q2's net profit of $47 million, and earnings per share came to $0.48 basic and $0.47 diluted, versus the $0.42 basic and $0.41 diluted EPS we reported in Q2. Newport Beach Fab Lease Extension. As we noted in today's press release, in order to meet ongoing and rising demand for SiPho and CIG, and because our Newport Beach Fab is fully utilized, we're extending that fab's lease by as much as an extra 3.5 years past the prior 2027 term. A $105 million upfront lease payment will show up as cash used for operating activities in our Q4 2025 statement of cash flows, with a matching effect on the cash figure on our balance sheet, and the resulting P&L effect will be, as we announced earlier today, $6 million each quarter, recorded across a five-year period as GAAP requires, in the COGS line. Hedging. I'd now like to walk through our currency hedging activities. When it comes to the Japanese yen, because the bulk of TPS revenues is denominated in yen and the overwhelming majority of TPS costs are also in yen, we enjoy a natural hedge across most of our Japanese business and operations. To address part of the leftover yen exposure, we're carrying out zero-cost cylinder transactions as a way to hedge currency fluctuations. So even though the yen-to-dollar rate can move around, our margins see only a limited effect; as for the Israeli shekel, we generate no revenues in that currency, but because some of our costs in Israel are denominated in shekels, we likewise hedge a substantial share of that currency risk through zero-cost cylinder transactions to reduce this exposure. So even though the shekel-to-dollar rate can move around, the effect on our margins is limited. Turning to our balance sheet and future cash and cash plans, CapEx and cash plans, as I mentioned earlier, our balance sheet is still very strong, which the following indicators and financial ratios demonstrate. At the close of September 2025, our assets came to more than $3 billion, mostly made up of $1.4 billion in fixed assets net, chiefly Fab machinery, plus $1.8 billion of current assets. The current assets ratio is very strong, at roughly 7X, and shareholders' equity hit a record $2.8 billion at the end of September 2025. Our strong financial position lets us invest in strategic opportunities that support our corporate vision, as follows. When it comes to our SiPho and CIG operations, which carry high margins, we had said before that we'd put $350 million into boosting capacity at our 8-inch Fabs located in Israel and Texas, plus our 12-inch UMC Fab over in Japan. A significant chunk of that CapEx is capability CapEx tied to advanced development work and projects related to high-end RF technology. As of now, 50% of that sum has already been disbursed, and the other 50% should go out over the next year. On top of that, earlier today we made an announcement that we've chosen to set aside an extra $300 million investment aimed at expanding capacity and building next-generation capabilities, primarily for equipment to grow SiPho and CIG capacity further at our 8-inch Fabs and at our 12-inch Fabs. That brings the combined CapEx plan tied to SiPho and CIG capacity and capabilities to $650 million in total. Every one of these investments is already baked into the strategic and financial model we laid out before. According to that model, we're aiming for $2.7 billion in yearly revenues once our current Fab and qualified capacity are fully loaded, which factors in the capacity expansion plan we mentioned previously, along with $560 million in yearly operating profits and $500 million in yearly net profit. That wraps up what I had prepared to say. At this point, I'd like to hand the call back over to the operator so we can start taking your questions.
Hi, Cody.
Yeah, I appreciate you taking my question. Morning, everyone, and thanks for fielding my questions. Nice work on the progress. Oren, could I just get a fast clarification — you mentioned that the extra $300 million had already been factored into your long-term total revenue expectation of $2.77 billion. Is that correct?
Yes. Yes. That could mean we hit this target sooner than someone had anticipated before, but yes, it's in there.
All right. So if there's no additional upside, what faster pace are you anticipating? I suppose what are the trade-offs tied to that added CapEx?
It moves us more quickly toward hitting the $500 million net profit goal, and as you're aware, we haven't reached that point yet. We'll get there faster. Naturally, reaching it sooner means greater profit earlier.
All right, that's great. Thank you, everyone. Russell, perhaps you could describe a few of the use cases that you believe are behind the strong expansion we're witnessing in RF infrastructure?
Sure. The largest and most powerful driver is simply the demand for buildout, and I'd say it's chiefly AI-related, though it persists for large quantities of 400 gb per second, plus extremely high volumes of 800 gb per second across several configurations of that, including both DR8 and 2xFR4s. As we mentioned, there's a very substantial volume at present moving toward 1.6 gb, and roughly 30% of everything we start is now committed to that platform. This stems mainly from the ongoing construction of data centers, along with a significant shift happening right now toward 1.6G.
Excellent. Thank you all. I'm grateful for the opportunity to speak with you.
Hello. I appreciate you fielding my question, and well done on the impressive numbers.
Thanks so much.
Thanks. I'd like to ask a couple of quick questions about Silicon Photonics, if I may. You brought up Tower's leading position. Who would you say your primary rivals are at present? With the current mismatch between supply and demand, do you have the ability, or are you thinking about, increasing prices?
The ability part — the response there would likely be affirmative, though on the thinking side, no. We maintain very close ties with our customers; we grasp what they require; we have roadmaps extending long term, and we don't act opportunistically — if you were to frame it that way — since demand is incredibly, incredibly tight, in order to squeeze someone for a few additional wafers. In my view, that's the most certain route to destroying goodwill and partnership. What we do observe, on the other hand, is demand that is extraordinarily strong. Having said that, that explains why, after putting in $350 million — and we're now seeing the earliest indications of that ramp — we're adding a further $300 million of investment. I had said we aim for a Q4 SiPho revenue shipment run rate above $320 million. Relative to the qualified capacity we ship the Q4 against, we're lifting that by more than 3x in start capability over the coming four quarters, and we expect that full 3x plus to be ready for starts during the second half of 2026. That's a considerable amount. If you took 3x of a $320 million run rate, that represents a very substantial expansion in SiPho that we're forecasting for ourselves. Having gone from roughly $28 million, give or take, in 2023 to $105 million in 2024, to this year, we're aiming at more than $220 million for the entire year, with a run rate of $320 million-plus in Q4, and then lifting that to well beyond $900 million by target; that target is genuinely voiced by customers; it isn't a field of dreams. If you build it, they will come. It's customers telling us, "Please build it; we need the capacity." I believe we're in a very strong position regarding our SiPho relationships. I also mentioned in the script that we're doing a great deal on capability, not only for the current 1.6T generation, but to ensure our customers hold a leading position for the capabilities required for 3.2T or for 6.4T, particularly the requirements for a faster capable modulator of a 400G. Those efforts are happening in very real time. We're putting money in real time into expanding capacity, which the industry is genuinely demanding. Fortunately, as I said, we are by far the leader in manufacturing, but I believe we're also the leader when it comes to creating next-generation platforms alongside the leaders, so that both sides are ready well before demand actually appears. I hope that addressed your question?
Yeah. To wrap up on the opening topic, are you noticing any shifts in the competitive landscape? Is anybody else putting money into that area and attempting to grab some share from you folks?
My guess is that plenty of companies would love to win share away from us. It's a solid growth sector, with great, powerful customers. The real crux is what you just raised. Being opportunistic when it comes to pricing would likely be a fine way to invite our customers to respond, "Hey, we're not interested in being your long-term partner. Find somebody else you can use pricing to take advantage of." That's not something they're required to do. We collaborate very closely with our customers so things stay fair and both sides come out ahead. Sure, I'm certain there are others attempting to cut into our territory. Even so, it's extremely hard for anyone to crack our position at this moment. Luckily, and really for the purposes of this call, Marco Racanelli is here with us, and the RF activities report directly under him. Marco, I'm not sure if there's any color you'd like to add to that.
Sure. Regarding the pricing conversation, like you mentioned, we aren't exploiting the current tight capacity across the industry, but our advanced platforms do bring added value. In that sense, we charge more for higher-end technologies that give customers greater benefits. From that angle, we expect some pricing gains over time as customers shift toward these more sophisticated technologies. Also worth noting, SiPho is already quite margin-accretive. We are compensated for the value we provide, and that's the right foundation. We've made it very clear that SiPho offers a substantial advantage to our customers compared to EML. For customers adopting our platforms today, when you consider a halving of lasers, that's significant. You also have, at this stage, for 1.6T, the use of a silicon modulator within the PIC itself rather than requiring an indium phosphide modulator, which is far more expensive and involves considerable three, five area; so the platforms carry value, and clearly, we both participate in whatever value gets generated.
Excellent. That's much appreciated. Please return to the line. Thanks.
Thank you, Russell and Oren, for fielding my questions, and congratulations on those impressive figures. I'd like to begin with a couple of inquiries regarding Silicon Photonics. Russell, you offered a fascinating remark that I likely didn't capture properly in my notes, but you said something regarding the move toward Silicon Photonics being permanent. Could you clarify what that means, please?
Yes. As we said in the script, it's extremely favorable on cost versus EML. It requires only half the lasers. Considering double the expense of 3, 5 and likewise the supply limitation of three, five, it's a significant advantage on both fronts to be able to employ Silicon Photonics at 400, 800, 1.6T and perhaps, contingent on development, even at 3.2T. With SiPho, you get a silicon modulator, which is also highly favorable on cost compared to needing a 3, 5 modulator. Saying there's a cost advantage certainly creates long-term stickiness; alongside that cost advantage, there's also a performance advantage. Once you put cost and performance together, that's truly an unbeatable combination for market share stickiness, correct?
Alright, thanks for that. For my second question, when we look at 1.6, it's clear that module manufacturers are describing some solid ramp-ups beginning next year. What sort of blend should we anticipate between 1.6 and the slower speeds in your operations — maybe next year, or at some specific moment? I'm just attempting to gauge how quickly this is scaling relative to the older, slower technologies.
At present, 1.6 accounts for nearly one-third of our starts, which shows you where things stand today, and that's risen from practically nothing at the start of the year. It's a really rapid ramp toward 1.6T. My expectation is that we'll exceed 50% during the first several quarters of next year.
All right, that's useful. For my final question about Silicon Photonics, I want to build on what you said and also on one of the questions here regarding hitting triple, I believe that was a capacity remark for Silicon Photonics relative to the run rate in this present fourth quarter. Over what sort of period do you anticipate being able to reach your maximum utilization level there? It sounds like that might happen this year or, pardon me, in 2026, but I just want your view on what you expect.
Customer forecasts say the demand exists or will exist. According to our planning, we ought to have everything installed during the first half of 2026, with all tools operational and able to achieve full start capability during the second half. When exactly in the second half we begin wafer starts really determines the shipment level. Typically you'd be looking at a cycle of three to four months from starts to shipment, depending on how large the POs happen to be. We would definitely expect part of this added capacity to show up in revenue during the second half of 2026, but bear in mind that the initial ramp we're carrying out hasn't been realized yet either. That earlier $350 million investment is presently in its earliest ramp stages. During the first half, and mainly in the second quarter, we ought to see a very large increase in our Silicon Photonics shipments revenue, and that ought to carry on through the third and fourth quarters. Will we reach the complete 3x or +3x shipment capacity in 2026? Honestly, I don't know. The demand exists. That's the question, and I do think we'll begin the full amount during the second half of the year. Shipment might extend into +-Q1, but I believe we'll see, we ought to see, a substantial portion of it. I mean, that's the reason we're making the investment, and certainly that's the reason we're speeding it up.
All right. That's great. I've got one final item, and then I'll leave the queue. Oren, could you address how gross margin will flow through across the coming quarters? Clearly, we recognize Silicon Photonics is a business that adds to margin for you, and it appears that will serve as your primary engine here. How should we view this moving ahead? I'd also like to understand when extra depreciation comes into play so we can factor that in. Thanks.
Yes. At present, I believe gross profit for Q3 came in at 24%, which is $93 million on $396; that's the real figure. It ought to be stronger. As you're aware, our long-term financial model carries larger percentages. Typically, we discuss incremental margin at 50%. Naturally, thanks to SiPho, that figure will be considerably higher, and two things will counterbalance it. The first is the Newport Beach lease payment we indicated we'd make, amounting to $6 million in total. The second is exactly what you pointed out, the depreciation tied to the extra CapEx. That added CapEx comes to $600 million spread across 15 years; that works out to roughly $10 million each quarter, though a portion has already begun. It ramps up gradually toward that $10 million.
All right, that's great. I'll break from the queue here, everybody. Thanks.
Thanks, Richard.
Yes, I appreciate you taking my question. There are two or three more things I'd like to ask. My apologies — I've come in alongside a colleague, so forgive me if some of what I ask has already been asked. Russell, regarding the opportunities tied to transceiver, are you factoring in greater content as well? To put it differently, if we assume the bulk of the opportunity today sits on the transmit side, would the receive portion of the transmitter also create a chance for you to add content?
Absolutely, that would boost volumes. I noted that a quite attractive near-term upside came from a fairly sophisticated platform shipping 300 mm receive SiPho during the fourth quarter. During 2026, we'll move into multiple formulations past what we're delivering in the fourth quarter. Yes, the receive side represents an additional served market, and at present the majority — in fact, all of our efforts apart from those initial 300 millimeter activities — target transmit. Excellent question. Thank you.
Thanks. Could we somehow break down that 3X expansion of SiPho capacity into how much comes from more content per unit versus how many transceiver units actually get shipped to the end market?
I'm not entirely clear on what you're getting at with more content, but these are units going into transceivers. At the 1.6, sorry. Please continue.
Okay, okay, okay. When you raise the proportion of receive and transmit inside that same transmitter, what portion of that is already accounted for in the three times, or 3x, capacity gain?
At this point, that 3x gain includes very little receive. It's almost entirely transmit.
As your customers move over to the receive side using SiPho, would you have to add capacity once more?
Absolutely. That would make us extremely happy. We already have platforms handling this. It isn't as though we're sitting back and not pushing ahead on it. The thing is, at this moment, demand on the transmit side is incredibly strong. Definitely, receive becomes truly powerful once DMUX is involved, and that's a major focus area for us.
All right. Thank you. I appreciate the detail. Just one quick follow-up. Over the longer term, where do things stand on the packaging side? I believe it was a couple of quarters back that you pointed to R&D efforts aimed at broadening your addressable market to take in certain packaging work. Can you give us any update on that?
We've got a program underway with a top-tier packaging partner, and it centers on developing a CPO — a complete CPO. Beyond that, we definitely have additional efforts tied to NPO, substantial ones involving through silicon via, which is needed for NPO and could also be needed for CPO. We're moving forward on those. It's not as though a great deal of CPO is in deployment at present. It isn't. We're advancing well; we also have further initiatives that are generating extra, genuinely robust capabilities; we're discussing several generations ahead, yet there's a very large program centered on 51.2T to deliver extremely advanced modulation that might be needed inside a PIC structure for whichever CPO gets deployed.
All right. If these opportunities do turn into revenue, it's mostly going to happen later in this decade. Commercialization would need a couple of years, correct?
My view is that it will be a couple of years before CPO is heavily commercialized, full stop, regardless of Tower.
All right, got it. I appreciate all the specifics, and congratulations.
Thanks so much.
Yes. Hello. I wanted to ask whether you might spend a moment discussing 3.2T. It appears that perhaps this technology cannot make the leap from 1.6 to 3.2. Could you share, more or less, exactly where they stand when it comes to attempting to fix that issue? Does a danger remain that a completely brand-new technology could be required?
My view is that no completely brand-new technology will be required. Even regarding the DSP, particular problems exist that are being addressed. No, a new technology is not something I anticipate. As we see it, what is genuinely needed is the modulation referenced at 400G. Just as the script states, we are pursuing three separate pathways, based on what customers need and want, for the 400G modulator. Purely from a modulation perspective, I believe we are tackling it quite forcefully, with quite solid progress. When it comes to the whole transceiver, the build-out or CPO, I mean, those represent separate matters, yet I don't believe they cannot be attacked or solved. I don't believe 3.2T will be delayed, no.
All right, I just want to make sure I understand this correctly. When it comes to SiPho, are those revenues entirely constrained by how much capacity is available? In other words, assuming you could produce it, does the demand exist?
That's right.
All right, just one quick thing. When will you begin disclosing Agrate utilization? Is that a long way away?
It doesn't have to be. Reporting it could begin at any point. We simply haven't done so; the initial ramp is currently underway at that site; we could begin disclosing it. Nothing would prevent us from doing that.
Alright, that covers everything I wanted to ask. Thanks a lot.
Thanks, Lisa.
Great. Thanks for allowing me to pose another question. Russell, stepping back a bit, when you consider your whole business and clearly include Silicon Photonics in that view, what's your thinking regarding 300mm? You clearly possess some via a pair of joint ventures — or whichever term fits best — with Intel and ST, plus you have some capacity in Japan. How do you view obtaining additional 300mm? Having it seems quite crucial for you, and clearly it's hugely important for Silicon Photonics; that's plainly your major growth engine. Can you locate additional partnerships resembling those you've already announced with the two partners previously, or is Greenfield the path, or in what manner do you plan to tackle this? Thank you.
Richard, that's a superb question, and it's one we're concentrated on. No, I don't believe we'll tackle it via a partnership like the one with ST or the one with Intel; if we move ahead there, we'd tackle it in a somewhat organic way. I think it's a superb setup for being eager about our Q4 release, and at that point we'll discuss it further. All right. Then I'm looking forward to that call.
That wraps things up on my end. Thanks, everyone.
Thanks.
Honestly, I want to express gratitude for the ongoing attention toward Tower. I appreciate the backing for Tower. And I definitely appreciate the questions that came up during this call too. We're genuinely thrilled to bring you updates across the next several quarters regarding that capacity expansion beyond 3X that we've discussed for SiPho/SiGe growth as it becomes operational, plus updates on other advances and initiatives. Once more, that's a great question. The question concerns how this CapEx spending shapes the financial model. We've already noted that it pulls forward the timeline for hitting particular revenue levels, and it speeds up that timeline thanks to SiPho — as was said, margins that are accretive, executed under a somewhat different financial model; we're really eager to share updates on every one of these items. As we prepare and get set and deliver our end-of-year remarks, which always consist of a recap of the year plus an outlook on what lies ahead in the years to come, I believe there are numerous thrilling subjects we'll discuss with you regarding where the company is headed and how each of these accretive moves truly proves to deliver a very robust advantage and a powerful ROI. When examining our SiPho efforts regarding CapEx, what's genuinely excellent about SiPho and the CapEx is that starting right when you begin shipping wafers, you're looking at a half-year ROI on the CapEx invested in the tools. That isn't measured from when the CapEx is ordered, but rather from when you're actually qualified and shipping wafers and executing that ramp — ROIs come very, very fast. As Oren said, given that we're pursuing a very large expansion, given that this expansion is validated by customers asking for the capacity, and given the fast turnaround this market delivers on ROI, you can then understand how the revenue timing is truly accelerated by a great deal. We'll provide plenty of detail on that when we deliver the end-of-year and forward-looking statements alongside our Q4 release. Once more, I'm very much anticipating walking through that with you. This is an extraordinarily thrilling period for the company, extraordinarily thrilling, I believe, for our customers and for us amid customer partnerships with those who trust us and who seek us out for their solution and for expanding their own operations. I'm very excited to host our 2025 Technical Global Symposium in Santa Clara next week, taking place November 18th, open to all customers. I'm very pleased that you'd attend and listen to us. As mentioned, Dr. Charlie Kawwas from Broadcom will deliver the customer-invited talk there. It should be extraordinarily interesting. I've heard him speak on occasion previously, and he's a fantastic speaker possessing an immense amount of knowledge and capability. Dr. Racanelli will provide an in-depth overview. It may sound contradictory, yet an in-depth overview covering all our technologies. To start, I'm going to deliver the opening remarks, which will focus on Tower's culture, our direction, and the foundation for what I consider to be unmatched customer partnerships. Beyond this technical symposium, we will be taking part in San Francisco at the 23rd Barclays Annual Global Technology Conference on December 10th. Then, on January 13th and 14th, we'll be at the 28th Annual Needham Growth Conference in New York. Naturally, if any of you who are investors or analysts would like to schedule further calls with the company, please reach out to Noit, and we're glad to arrange those based on whatever you need and prefer; that brings my remarks to a close. Thanks once more. It's a very exciting time.
That brings us to the end of today's conference call. Thanks for joining in. Everyone is free to disconnect at this point. Have a great day.
Read the full transcript via the API
You're viewing the first half of this call. Get the complete Tower Semiconductor Ltd. transcript - plus 132,000+ transcripts from 7,000+ companies, speaker segments and full-text search - through the EarningsAPI REST API or hosted MCP server.
Get an API key View API docs →For developers and AI pipelines
Programmatic access to Tower Semiconductor Ltd. earnings transcripts and 132,000+ others is available through the
EarningsAPI REST API and the hosted MCP server.
Quarterly plans from $162 - full transcripts, speaker segments, full-text search,
and the /api/v1/transcripts/recent polling endpoint for ETL pipelines.