Navitas Semiconductor Corporation (NVTS) Earnings Call Transcript
November 3, 2026
Earnings Call Speaker Segments
Thanks, and good afternoon, everyone. Welcome to the conference call covering Navitas Semiconductor's financial results for the third quarter of 2026. With us today are Chris Allexandre, Navitas' President and CEO, and Tanya Stevens, our CFO. I want to remind listeners that the prepared remarks from management include forward-looking statements, and those carry risks and uncertainties; management might also make further forward-looking statements when answering your questions. As a result, the company is claiming the safe harbor protection for forward-looking statements provided under the Private Securities Litigation Reform Act of 1995. Actual results could come in different from what we discuss today, so we point you to a fuller discussion of the risks and uncertainties in the company's filings with the Securities and Exchange Commission, including its filings on Forms 10-K and 10-Q. Also, any projections regarding the company's future performance reflect management's estimates as of today, November 3, 2026. Navitas undertakes no obligation to update these projections later as market conditions may or may not change, except to the extent applicable law requires. In addition, the company's press release and management's statements on this conference call will include discussion of certain measures and financial information presented in GAAP and non-GAAP terms. The company's press release includes definitions and reconciliations of GAAP to non-GAAP items, which offer additional details; for anyone who can't listen to the full call right now, a recording will be accessible via webcast for 90 days in the Investor Relations section of Navitas' website at www.navitasseemi.com. Now I'm pleased to hand the call over to Navitas' President and CEO. Chris, please go ahead.
Good afternoon, and thanks for being with us on today's call covering third quarter 2026 earnings. We're grateful for your ongoing interest and support as we get close to finishing our strategic transformation and start operating as Navitas 2.0. Results for the third quarter show yet another stretch of strong financial performance along with quickening momentum. Total revenue landed at the top end of guidance, at $14 million, which is up 33% sequentially and 38% year-over-year — the company's first quarter of year-over-year growth since 2024. High-power markets drove that revenue growth, rising more than 50% sequentially and growing by triple digits year-over-year. As we anticipated, the mobile and low-end consumer segments continued to decline as a portion of revenue and are still expected to be negligible by the fourth quarter. When the year closes, high-power will basically represent the entire company, and we will have considerably exceeded the mobile and low-end consumer markets; indeed, after delivering several consecutive quarters of double-digit sequential revenue growth along with gross margin improvement—driven by the high-power market and a pickup in AI infrastructure, which is our core focus—gross margin rose sequentially and landed above the midpoint of guidance, reflecting the continued advantages of a favorable revenue mix and growing scale. One notable point: our non-GAAP gross margin for the third quarter came in at 40%, which is 130 basis points above the third quarter of 2025 — the period when the Navitas 2.0 journey got underway. Book-to-bill has stayed comfortably above 1 across the past several months, and right now we hold a healthy backlog that stretches into mid-2027. Looking ahead to the fourth quarter, we anticipate further revenue growth as well as margin expansion, and that quarter would mark the fourth in a row of double-digit sequential revenue growth together with margin improvement, with the company surpassing 2026 as a high per company. To give a bit of context, about a year back we made a conscious choice to move away from mobile and low-end consumer products and direct our resources toward the high-power market, where GaN, high-voltage SiC and ultrahigh-voltage SiC offer more differentiation and better long-term growth; we identified AI data center, energy and grid infrastructure, high-performance computing and industrial electrification as the high-power market's clear focus, with AI infrastructure as our North Star. That pivot's success now shows plainly in our revenue mix. When 2026 began, Navitas 2.0 was merely a strategy, and today it is ready. As we said before, by Q4 mobile revenue should be basically gone, and essentially all of the company's revenue will come from high power. Our departure from lower-margin mobile and low-end consumer has been more than offset by enormous growth in AI infrastructure, which is our North Star — AI data centers along with the grid and energy infrastructure needed to power them. We expect AI infrastructure to account for over 40% of total revenue as we exit this year, and we think this business is set to triple in calendar 2027. In recent months, it has become clear and widely accepted that power is now the number one constraint on AI compute. What matters now isn't just how many GPUs or XPUs we can build and get out into the field — it's how many of them we can actually power. Put another way, efficiency in power is turning into capacity. When power access is tight, each watt you save in conversion and delivery is a watt you can put toward compute instead. It expands how much compute fits within a set power envelope — and it isn't just about the energy bill. Alongside our high-voltage and ultra-high-voltage SiC, GaN, and the Claros IVR technology stack we recently acquired — which serves both vertical and embedded power delivery architectures — this puts us in a stronger spot to break through the AI infrastructure power wall. We do that by tackling the efficiency, power density, and performance bottlenecks throughout the entire power chain, from the grid all the way to the XPU. With the Claros acquisition completed in early October, Navitas now has a full platform able to handle the last stage of getting efficient power to the XPU itself. I want to emphasize that the count of processor sites continues to rise — CPU, GPU, TPU, XPU, plus various custom processors — and the firms constructing them are multiplying as well, with every one requiring a tailored power implementation for that last conversion stage down to core voltage. A modular strategy is needed for this, and it stands as one of the chief differentiators of the Claros IVR technology stack, because it enables numerous IVRs to be merged with considerable flexibility inside a power array to reach the required power level and density. Navitas already stood alone in providing both GaN and SiC. Now, with Claros, that's even more true: we enable efficient power delivery starting at the energy source — beginning in a modernized grid with our ultra-high-voltage SiC, then high-density, high-efficiency centralized AC-DC and DC-DC conversion using our high-voltage SiC and GaN, and finally power delivery to the XPU through the Claros IVR technology stack. Going a level deeper, IVR technology addresses a genuine scaling challenge in the real world. Processor power demands are climbing faster than lateral delivery can keep up with, as package power goes from roughly the mid-1 kilowatt range toward several kilowatts at sub-1 volt core and several thousand ramp. Placing conversion closer to the XPU — first with vertical power delivery directly under the processor, and eventually with power delivery embedded in the PCB package or substrate — shortens the route from centimeters to millimeters or less. That reduces resistance, inductance, voltage drop, and heat. Instead of being just another power FET, we offer an integrated IVR platform that combines digital control, power transistors, high-density capacitors, embedded inductors, and packaging in one architecture. The platform is designed so it can grow alongside processors as they are today, thanks to a repeatable building block of roughly 40 amps that extends through bigger chips and a modular power array reaching multi-kiloamp systems, all packaged and embedded. Put simply, the Cloud IVR technology stack—capable of supporting both upcoming VPD and EPD architectures—delivers a complete power solution platform for us. By layering GFS onto what we already offer, our 2030 SAM more than doubles to over $8 billion, and we gain a multiyear content growth opportunity that builds on the inflection points we're already pursuing with high-voltage SiC and GaN. The genuine benefit is that Navitas enables customers to convert a greater share of every megawatt available into productive AI compute. We have largely completed our evolution into a high-power business, but the path ahead continues to depend on the same core transformation pillars we have discussed over the past year: market focus, technology leadership, operational efficiency and financial discipline. On the market focus and technology leadership fronts, we continue to advance a range of customer and technology proof points. In the time since our previous conference call, we've made several announcements. First, we're working together with NVIDIA on a power delivery board that converts 800 volts to 12 volts, aimed at next-generation AI infrastructure. That effort extends the 800-volt 6 and 800-volt 250-volt DC-DC conversion board we announced earlier, and it further demonstrates just how broad our system knowledge and capability really are. We also licensed GeneSiC technology to Microchip — a move that highlights the strength of our trench assisted planner process foundation and supports wider SiC adoption, particularly in Asia, while also creating long-term potential for an additional foundry source. We recently partnered with Microchip as well to deliver an 800-volt reference design for next-generation AI data centers. Then, in late September, Navitas was selected by the U.S.
The government is developing a next-generation 10 kV SiC power semiconductor, and that effort bolsters both our technology lead and the strength of the domestic supply chain. We also reported initial deliveries of our fifth-generation GaNFast technology produced in the U.S., which stems from our long-term strategic partnership with GlobalFoundries. Having shipped first wafers in September for the first product family, 650-volt GaNFast included, we anticipate strategic customers will be able to sample very soon. Relatedly, we earlier talked about pursuing a Foundry plus model. We have since renamed that strategic operating model, calling it Integrated Fabless Manufacturing, or ISM—an intentional alternative to the conventional IDM approach. What really sets us apart, and the thinking behind it, is that we choose which things to own, where our money goes, and who we work with — so we hold on to the abilities that make our product distinctive while using partners to build at scale. The key point is that we retain ownership of essential technology IP and certain pieces of manufacturing equipment, along with focused co-investment, instead of relying only on supplier collaboration. On today's call, I'd like to take a moment to spell out and establish the basis of our IFM model, since I anticipate it will be strategically important in how we move forward. Focusing on the nearer term, I'd add that we're deliberately using our stronger balance sheet to build strategic inventory buffers—raw materials among them—and to lock in capacity for the future, all to support growth in the mid- to long-term that has been unexpectedly robust and steady. Now, an update on how we're progressing against the four inflection points I laid out on our previous conference call, which I plan to keep using as the framework for mapping our growth to architectural change in AI infrastructure. As of today, the shift to 800-volt architecture for next-generation AI data centers is a certainty. The question is no longer whether 800 volt will happen, but rather how it will play out. As we've said before, this shift will involve multiple turning points that create rising momentum and greater demand for higher-power solutions. Navitas's distinctive mix of high-power GaN, high-voltage SiC and our IVR capabilities will let us take advantage of every one of those turning points. I'd like to stress that we don't rely on any single architectural change and that we aren't sitting around waiting for the final data center architecture to take shape. These turning points are a series of opportunities, and each one can raise content per system as the broader market keeps growing and expanding. The first turning point, pre-800 volt, is already well along. We're seeing faster replacement of silicon for any ACDC above a few kilowatt. Our SiC-based products that target this opportunity are now in production, and they're fueling faster growth during Q4 and the first half of 2027 among several hyperscalers as well as OEM ODMs. Demand is picking up speed, especially for 1.2 kV SiC, where Gen 5 — which we've mentioned before — is building real momentum. The second inflection point, the SiC AR rack along with the initial rollout of plus or minus 400-volt and/or 800-volt, is speeding up too. We covered this during the last earnings call, and at October's OCP Summit it was clear that this is coming and will certainly land in 2027. We're observing broader design wins and programs with more customers and applications, spanning AC-DC, DC-DC and battery backup units, or BBU, each with higher dollar content. We've already booked substantial orders that are scheduled to ramp in the first half of next year. That will drive growth in 2027, building on the first inflection point that continues to expand across AI data centers that haven't yet transitioned to high-voltage DC. The rise in power level and density of AC-DC PSUs is genuinely remarkable, and that's what's behind this jump in content. The same is true on the DC-DC side. At present, we are in the final stage of design and qualification work with many customers, and we already hold orders that will add programs fueling further growth over the coming year. The second inflection point is especially compelling because of how many applications it spans—DC-DC and AC-DC, BBUs and more—along with the blend of technologies spanning SiC and GaN, all of which bolsters Navitas's strategic position. Our third inflection point, Nive 800 volt, continues to move forward. Everyone knows by now—and OCP reaffirmed it—that this represents the only path to a genuinely high-density, high-power rack. So when I say 500 kilowatts and up, all the way to megawatt-level racks — that's the range I mean. For this high-density rack, we're accelerating design and prep work with several vendors, and that depends on the XP. It's getting a lot of attention. The Native 800-volt solution is still on track to hit production before 2027 wraps up, and we expect it to matter materially in 2028. Right now we're sampling and making headway with customers across more than a dozen programs, and with certain leading OEM ODMs we've reached the final stage on the 800-volt power DC-DC conversion board. One existing customer going after hyperscalers has already put in a sizable order for significant system testing, which highlights Navitas's strong position in the market. The fourth turning point—full modernization of energy and grid infrastructure—remains farther off, yet it is advancing too, driven by how quickly AI data centers need power and how much they consume. We continue pursuing initial design wins for our ultra-high-voltage modules in solid-state transformers, which is genuinely a growth driver for 2028 and beyond. Meanwhile, we are seeing quicker engagement around nearer-term revenue opportunities involving battery energy storage systems, or BSS, along with other power grid-related applications, and those are creating strong demand for ultra-high voltage SC. Several of these opportunities ought to keep expanding in calendar 2027, adding another layer on top of the first and second inflection points tied to AI data centers. In closing, I am very pleased with the progress. Navitas 2.0 isn't a plan anymore — it's a reality. You can see that in the results through the revenue mix, the growth, and the margin expansion. Once the year closes, we'll be a high-power business, having essentially exited mobile and low-end consumer. The market's logic has become clearer. Right now, power is the binding constraint on adding compute. So efficiency in power is how capacity gets expanded. AI infrastructure's architecture is only starting to be rebuilt, all the way from the grid to the XP. With Cloud acquired, Navitas' platform is now complete and so is our grid-to-XPU strategy; thanks to SiC Gallium IVR, we aren't just positioned to take part — we can enable that multiyear transition with multiple points of monetization. Our full transformation lands at an ideal moment, since we sit where 2 secular trends intersect: AI compute growth and the rearchitecture that growth demands. From grid through XPU, the end goal for us is helping customers convert more of each megawatt available into useful AI compute. As for the fourth quarter specifically, the midpoint of guidance means yet another quarter in a row of double-digit sequential gains and over 100% year-over-year growth, powered by high power and AI infrastructure and adding to further gross margin expansion.
For Q4, what we're expecting also means the full year gets back to growth — up roughly 7% in '26, with high-power growth more than making up for leaving the mobile business. We see this as just the start, since we're staying committed to the path of continued growth, incremental gross margin expansion and, in the end, profitability. With that, I'll hand things over to Tonya to go through our Q3 financial results and the Q4 outlook.
Thanks, Chris. Just a quick note before I start: unless I say otherwise, my comments will center on non-GAAP results. You can find a full reconciliation of all non-GAAP to GAAP financial measures in the press release we issued earlier today. Third-quarter 2026 revenue came in at the top end of our guidance, up 33% sequentially to $14.0 million — an acceleration over the 22% sequential growth we posted last quarter. That works out to roughly $3.5 million more than the $10.5 million we recorded in the second quarter. As Chris mentioned, this double-digit growth came from stronger traction in high-power markets, which rose over 50% quarter over quarter and by triple digits year over year, reflecting how nearly complete our revenue mix shift is away from our legacy mobile and low-end consumer business. We still expect mobile's revenue contribution to become insignificant — in the low single digits — by year-end. Thanks to the improved product mix and higher quarterly revenue, gross margin rose another 50 basis points sequentially to 40.0%. That marks our third quarter in a row of sequential gross margin expansion. Our accelerating shift in overall revenue mix toward higher-value, high-power markets and away from mobile and low-end consumer remains central to our gross margin expansion strategy. We anticipate further gradual gross margin expansion over the coming quarters as we keep driving top line growth in high-power markets. Third-quarter operating expenses were $16.5 million, versus $15.5 million in the prior quarter and $15.4 million in the same quarter a year ago. In line with the guidance we gave last quarter, OpEx rose by about $1.0 million as we began making targeted incremental investments in the business, particularly in new R&D programs to support accelerated growth. Having diligently kept OpEx effectively flat through the first half of the year, we are increasingly focused on deploying resources and investments in a deterministic way to support the longer-term success and sustained growth of the transformed company. OpEx for the quarter grew roughly 6% sequentially, meaningfully below our 33% top line growth rate. This and future incremental OpEx, including expenses related to the Claros IVR technology stack and product development, will be allocated to scaling the business, including investments to accelerate new product development, strengthen our engineering and application support for key committed programs, and reinforce operational readiness ahead of expected growth in ramping shipments. Our third-quarter loss from operations was $10.9 million, compared with a loss of $11.4 million in the prior quarter and $11.5 million in the third quarter of 2025. In Q3, weighted average basic and diluted shares outstanding were approximately $261.5 million, producing a Q3 loss per share of $0.03, compared with a $0.04 per share loss in the prior quarter and a $0.05 per share loss in the year-ago third quarter. Turning to the balance sheet. At the close of Q3 2026, our cash and cash equivalents stood at $531.5 million, down from $557.4 million when Q2 ended. That decline mostly mirrors the TSMC buffer inventory we've been building steadily. One thing to keep in mind: the quarter-end cash figure came before we paid the cash component of the Claros deal, which closed on October 5. Had we adjusted for the roughly $126 million Claros payment made after the quarter closed, our effective pro forma cash balance would have been about $405 million. As a reminder, the company still carries no debt. I'd also like to note briefly that we drew nothing from our ATM facility in Q3 and feel good about where our cash stands now that Claros is behind us. Following the Claros transaction, we still have plenty of capital resources to speed up and scale further as a high-power company. That covers anticipated strategic investments to push forward our integrated fabless manufacturing, or IFM, model initiatives, possible capacity expansion and supply reservation agreements with our U.S.-based Foundry partners, and the possible pursuit of selective strategic opportunities. On inventory: we finished Q3 with $32.4 million, of which $11.2 million sits classified as long term on the balance sheet; the prior quarter stood at $19.5 million, and the change reflects the buffers of TSMC wafers we'd already planned to build so customers see a smooth transition, along with other supply chain-related investments. Just like last quarter, the $12.9 million sequential rise in Q3 inventory mainly reflects our measured investment supporting AI data center customers' anticipated long-term growth. More broadly, inventory at the channel and distributor level stays healthy. Turning to our Q4 2026 guidance: we expect sequential growth to continue, with revenue up 14% to $16.0 million, plus or minus $0.5 million. At the midpoint, that's 119% year-over-year growth and marks our fourth straight quarter of sequential growth, powered by high-power markets and AI infrastructure in particular. On that same guidance, we anticipate 7% revenue growth for the full year 2026, with our core high-power business still delivering triple-digit growth even as our mobile business falls to effectively zero after accounting for a large majority of total revenue last year. As we exit the year, we expect high-power markets to make up almost all of our business, with over 40% of total revenue coming from AI infrastructure, which spans both our AI data center business and the grid and energy infrastructure business that AI data centers require. Finally, I'd add that we ended Q3 with a very strong book-to-bill ratio, well above one, and we currently hold a solid backlog stretching into mid-2027. For non-GAAP gross margin, we anticipate expansion to 40.5%, give or take 100 basis points. At the midpoint, that works out to a 50 basis point improvement, and it marks our fourth quarter in a row of margin expansion. It also reflects the ongoing favorable mix shift toward high-power markets, along with a modest added benefit from better scale. As we mentioned earlier, OpEx will step up in Q4 as we absorb the overhead operating expenses tied to Claros, and we'll also keep making incremental organic investments, mainly in R&D and operational readiness ahead of the growth we expect over the next year. Non-GAAP operating expenses for the fourth quarter, which again will include Claros, should come in at $25.0 million, plus or minus $1 million. That wraps up our prepared comments. Operator, please open the line for questions.
Shadi Mitwalli here, standing in for Quinn. Congratulations on those results. I believe you said the backlog now stretches into mid-2027. And with these high-power markets beginning to inflect, I'd love to get a sense of how we should think about growth as we head into next year.
Thanks, Shadi, for that question, and I appreciate the chance you've given us to discuss next year's outlook, right? As we discussed over the past two earnings calls, we've witnessed a steady rise in book-to-bill. In fact, book-to-bill for last quarter, Q3, came in at two, and it has stayed well above one for two quarters running, right? So the way we look at '27, with the inflection point unfolding on top of each other, is that it's the path for us to deliver double-digit growth quarter-over-quarter, fueled by the acceleration of AI infrastructure, which, as we noted in the earlier remarks, will triple from the baseline in '26. And that double-digit growth will keep accelerating as each inflection point comes on top of each other. So essentially, the message today is that we've completed the first part of the company's transformation, which is the financial pivot we discussed, consumer to high power, as you mentioned. And now we're entering the next phase of the company, where the growth is just beginning, driven by the AI infrastructure, which will give us multiple years of double-digit quarter-over-quarter growth.
Great, that helps. Now, turning to the Claros acquisition — we've seen several IVR deals announced in recent months. I'm curious: could you discuss what you saw in Claros, and perhaps what advantages their technology offers over rival IVR technology?
Thank you, that's a very good question. Before answering your specific question, the first thing I'd point out is that the fact other MOSFET and silicon companies are acquiring IVR confirms what we've said — namely, the architecture has to change, right? The current approach of lateral power delivery shifting to vertical power delivery, using the same technology sort of assembled into a vertical package, won't sustain the density and current levels those processors require, right? What we saw in Claros — setting aside the culture, which is very similar to ours, and the engineering team's quality — is a genuine IVR technology stack built for XPUs. Rather than IVR designed for computing that was converted, okay, to AI data centers, as we've seen at some of the companies just acquired, right? A true 3.3-volt starting point, enabling a 3:1, with everything integrated in the stack — the capacitors, the inductors, the controllers, the FinFET — basically letting you achieve real high density because no external component is needed, right? And that's what we saw; and the modularity, which I mentioned in the webcast, also matters. What we observe is that, of course, everyone talks about the leading GPUs, but there are as many processors now in the use cases we have; and all those processors — call it CPU, XPU, NPU, LPU, PU customs, right? They all require super high density, high current IVR and power delivery. And I think the flexibility and modularity we saw at Claros is unique, and that's why we made that move.
Chris, congratulations on the strong results and all the headway. Following on from Shadi's question, could you give us a feel for how much of the 2027 growth is backed by awarded programs, and what already sits in your backlog today versus design conversions you expect?
I'll begin, and Tonya may want to add a few comments, okay? So, our backlog already extends well into '27. So we can't claim the entire year is locked in, right? But here's what I'd say: we've been very surprised — in a positive way — by how Inflection 1 has accelerated, which we really saw in Q3 and Q4 this year, but also Inflection 2, right, the sidecar, where we're seeing a faster conversion in the number of programs across AC-DC, DC-DC and BBUs. So more programs are converting faster into higher power, moving from design into design win into backlog. And last but not least, since we discussed inflection in Inflection 3 on the last call, we're now beginning to see — and we mentioned this in the remarks, right, we've actually landed a pretty large order, 1 million units of GaN devices for an 800-volt DC-DC high-density brick for Inflection 3, right? So things are really taking shape. I'd tell you, Kevin, slightly faster than I even expected six months to three months ago.
Absolutely, and I’ll add to what Chris said. What he described matches precisely what I observe and how I reason about it. Inflection 1, as Chris mentioned, shows up in our revenue. We saw it there last quarter. This quarter we’re seeing it in revenue again, and it’s accelerating. Inflection 2 is in our backlog, and that is helping us fill out 2027. And with Inflection 3 as well, we’re beginning to see it go beyond only high-volume orders and sampling. Chris brought up the one million units. For Inflection 3, that represents a meaningful amount of volume we’re looking at. So that is exactly what we’re seeing.
Yes. Understood. Okay, that works well. And moving to gross margins—congratulations on the ongoing expansion. Could you help us understand how we should view the gross margin profile across silicon carbide, GaN and IVR as each of those businesses scales? So, we don't really provide detail on how gross margin splits among the various technologies. What I would say is, if you look at the gross margin expansion in '26, right?
So, let's step back. A year ago, to this day, we discussed transforming this company from consumer to high power. Essentially, gross margin has risen, based on the Q4 guidance, by two points. That came largely from shifting consumer to high power. The next phase of gross margin expansion is coming from scale as we keep growing double digit, but also with the AI infrastructure, which is now 40% of the company in Q4 and poised to triple next year. And that portion of the business is actually also at a higher margin, right? So what you're going to see is—I just had the question before, how do we see growth, right? I mentioned double-digit quarter-over-quarter with that double-digit kind of accelerating over time. I think you're going to see the same on gross margin, consistent gross margin expansion quarter after quarter as we scale. And of course, as we keep making fundamental transformation moving to 18-inch on the GaN, continue to work with our foundries on cost reduction and so forth. But I mean, look at this as continuous gross margin expansion organically with the migration to high power and now to AI infrastructure.
Yes. And I'd simply confirm that our framework isn't tied to the technology itself — it's tied to complexity. And we've shared this insight with you previously: the greater the power and the greater the complexity, the higher the ASP and margin tend to be. And that covers tougher technology problems that need solving, including the IVR technology stack; those tend to carry higher margins. versus splitting that between GaN and SiC.
Yes, Kevin, there’s one more point I’d like to make — and Tonya is exactly right here. Margins follow value, and value is generated by the complexity of the customer problems you solve. What we’re observing, if you look at Inflection 1 and Inflection 2, or even ahead to Inflection 3, is that we’re discussing several times the power density on the AC/DC side, shifting from silicon to silicon carbide, and then to GaN once the add-on becomes native. When we address those difficult problems quickly, customers essentially reward us with solid margins. And that, I think, brings us back to the heart of what we’re trying to accomplish as a company.
I'd also add my congratulations on the consistent progress. When it comes to the IVR market, that's a brand-new market. How early in the XPU or GPU design process do you need to be involved in order to define the IVR? Is it necessary to collaborate with them while they're designing the XPU?
That's a really good question, Kevin. I appreciate you asking it; first of all, what you need is something we refer to as an IVR technology stack; that's what we have with Cloud, correct? So, integrating the FinFETs, the controller, the inductors, and the capacitors all into one technology. Then you apply that either to vertical power delivery, which is essentially a chip placed on the backside of the PCB or within the PCB, right, in an array form, as we discussed, to then deliver power to the customer, or even taking it a step further where it's no longer a chip, you embed the entire system or a portion of the system into the substrate and the XPU. Either way, you must collaborate with the XPU vendors. Either way, you must collaborate with the OEM or power OEM that handles the final step of our delivery to the XPUs so that the right architecture fits. And that's what I got asked earlier by Sean, I believe it was Kevin. What appealed to us about Claros, right? Well, beyond what I consider the best density, what I liked is this modularity, since there are no caps outside; so each customer requires a different shape of the Power VRM and a different level of voltage and current, right? And that's sort of where you need to work with them. Another thing I'd mention is to keep in mind that a lot of customers are becoming vertically integrated. So we used to have GPU companies. Now we have hyperscalers making their own processors. We also have processor companies that become rack companies, right? So the answer to your question is the technology stack stays the same, but you must collaborate with the processor company, which are becoming rack companies or even hyperscaler companies, to essentially deliver the power delivery.
That's excellent. Yeah, the design cycle is becoming quite involved; and that leads me to a strategy you mentioned that's interesting — integrated service manufacturing, where you plan to purchase the equipment. And I suppose, will that be on the back end? For instance, as you describe the IVR, it could come in various shapes and packages for different designs. Is that where you're bringing in the equipment? Or will it also be on the front end?
So, allow me to set you straight for a moment, okay? We refer to integrated fabless manufacturing as IFM, correct? Mostly on the foundry side; so the best way to understand this is that if you consider both SiC and GaN — we actually own the technology stack, likewise on the IVR, right? So, regarding SiC, we have GenSiC, which is a full stack high-voltage, high-voltage technology. For GaN, we overlay our own IP on top of TSMC previously, now GF, to differentiate in reliability and other figure of merits, and the same applies with the IVR when we own the full stack, right? So, what we aim to do is own the stack but partner with foundries where we put in capital to essentially secure capacity and supply so we can enable cost predictability and competitiveness. So, the best way to understand this is somewhat similar to what we announced with GF on the GaN side and other foundries we work with — can we combine the best of both. We don't want to be an IDM. We don't want to be a fabless, we're somewhere in the middle, and we believe this is the most capital-effective way to deliver the high growth while controlling our technology.
Congratulations on a solid quarter and a good outlook. I was wondering — I believe I heard pretty clearly that you expect data center infrastructure revenue to grow by 3x as we go into next year. What do you think the non-data center portion contributes? Could you tell us what's involved there and what that contribution might be?
Yes. Jon, thanks for that question. As I mentioned, our backlog extends well into '27. In fact, the visibility we have in AI infrastructure stretches much longer and much further out than in other markets, right? So what we've modeled for next year is that anything outside AI infrastructure is more or less flat. We see some growth in industrial; we see things roughly flattish in computing. Remember, we address the super high-end computing market, not the computing market as a whole, right? So we've assumed flat there because we lack a full year or even two quarters of that kind of visibility. But on the AI infrastructure side, as you pointed out, thanks to the visibility we have, thanks to the accelerated momentum in Inflection 1 and Infelction 2, and thanks to a number of design wins converting, we clearly see potential for that business to more than triple next year, right? So the best way to think about it is that '26 was a year of high power basically offsetting the exit of mobile. That's done. It even allowed us to end up higher than where we started this journey from a revenue standpoint, and to get 2026 with $16 million. Next year, really, the growth is coming from AI infrastructure, and we assume the rest stays flat.
All right, great. And then what's more interesting is that, heading into '28, I believe you've got several accelerations working at once, correct? You've got NVIDIA's 800-volt products ramping up. The SST products are ramping, and Claros is ramping too. Do you expect even more acceleration in '28? Or is it tough to comp against the growth that's coming in '27, or perhaps the base is smaller and you can still speed up from that point?
Naturally, the further out you look, the harder it becomes to forecast, correct? What I'd say is this: from what we see, given the products we have, the technology we have, the engagement we have, and those inflection points, it's double digit, and that double-digit quarter-over-quarter figure will accelerate as those inflections stack on top of one another. So it isn't inconceivable that the quarter-over-quarter compound growth begins to accelerate in late '27, moving into '28 and then '29. That happens as each of those technologies — SiC, GaN, and even IVR — layers atop the others, and each of those inflections arrives on top of the rest.
Great. Could I squeeze in one more question — Tonya, with all this revenue growth, how do you expect OpEx to scale? Do you have a figure, perhaps as a percentage of revenue growth? Or is there some other way we should think about it over the next year or 2?
Yes, good question. That step-up you saw in Q4 — as we discussed, it reflected the normal run rate in our guidance plus Claros taking on Claros. Our thinking there remains what we've said previously: OpEx rises as a percent of revenue. So following that initial step-up, we expect that to play out in '27, as a fraction of our revenue growth. We aren't really providing the full year, but that's the way we're looking at it.
John, beyond what Tonya mentioned, I'd like to add one point: from the start of this transformation, we've been explicit that our sights are set on reaching profitability. Scale is where profitability comes from. There are two vectors for us. First, scale — growing revenue at a consistent double-digit rate, as we discussed. Second, gross margin expansion, and that expansion will keep going as we just discussed; and once IVR comes into play, we also see a step function in the company's combined gross margin; but we stay focused on ensuring we achieve that by growing OpEx at a fraction of revenue growth, so that we can reach profitability at the horizon, right, which continues to be a very important focus for us.
I want to pick up where that last question and answer left off, particularly since there's an obvious angle here — thinking through what a breakeven model looks like. I'm not asking when you might reach it, but how should we think about that model? And I assume it would — I suppose I'd rather have an answer that factors in IVR along with the gross margin dynamic you described, but sort of at scale, and what the margin structure is, or maybe I'll just let you explain how we should view the breakeven model. And I have a follow-up.
Thanks, Richard, good to hear from you. The best way to frame this, I think, is that we'll keep growing the consistent gross margin organically. We see the company picking up a couple of points over time; and then it's not inconceivable, as I mentioned, that once IVR is in place, the blended gross margin is north of 50%. Once you're there, if you take our Q4 OpEx this year, sort of multiply it by four to annualize and then increase — apply a yearly increase that's less than revenue, you're basically landing at OpEx in the low 30s, right, which at 50% gross margin would put the breakeven point at roughly the low 60s. So that's sort of — if you're asking me about the model, that's the kind of model you need to keep in mind.
All right. Great. That's quite useful. Chris, perhaps I'll just pose a very broad question here, since we've observed some movement both at Navitas and throughout the industry, along with some patent disputes, I suppose that's what we'd call them, and you're — you sit on both sides of this. Maybe you could help us grasp where you stand on it and how you anticipate the industry will develop over the coming couple of years given this dynamic.
Richard, I believe I sort of addressed the question about the litigation on the last earning call. I think I was pretty clear, okay, on that last earning call regarding my perspective on this. If you don't mind, I'd rather stick with what I stated previously. We'll keep defending our IP and we would expect IP, but we'll keep defending our IP. And I'd rather stick with what I stated on the last earnings, if you don't mind.
No issue on that. Perhaps I could pose a different question in place of that one, Chris. Could you discuss the inflections, since clearly you're picking up genuine momentum in backlog when it comes to inflections one and two? It would help if you could describe your share qualitatively in whatever way you find interesting — for instance, SiC against GaN, or particular elements inside those inflections.
Giving you a blended market share for Inflection 1 is tough, correct, but here are a couple of data points, right? First, Inflection 1 is basically a substitution as you shift to 3, 5, 10 or 8, 10-kilowatt density of PSU, right? Silicon gets replaced by silicon carbide. What's great about it is that the replacement happens with 1.2 kV. So the content is quite attractive. And typically there are two to three vendors, right? So on certain programs we're first source, on others we're second source. And with 3 vendors, right, one takes 40, another takes 30, and the last takes 30. So over time, I'd say it gravitates toward a solid 30% to 40% share of the market — of the PSU we take part in. What's interesting about inflection 2 is, first, it involves far more applications. There's AC/DC, which now essentially moves it to 15, 18, 25, 27, 30 kilowatt. There's also DC-DC with 15, 30-kilowatt DC-DCs and BBU, which are beginning to use both SiC and GaN. So the great thing about Inflection 2 is that it isn't only a SiC play. It's a GaN play as well, right? And here we've seen that in some cases we're sole source; in other cases we are gain sharing, right? So the way I see it is that what you want is to be first on the print, okay, and give customers a secured path from an availability standpoint and an execution standpoint, so they can ramp production, because as I mentioned earlier, we see an acceleration. Had you asked me six months ago whether Inflection 2 and sidecar would arrive that quickly, I'd have said it's coming, but it's certainly arriving faster than we expected. Same with Inflection 1, okay? That's why you saw us come in above the estimates, okay, that we had set for ourselves that you had set for ourselves, right? So I think the speed is helping, because if you're first — and speed is core to our values — if you're first, you can command more share, right? But it's hard to tell you how much market share of that market we own. What I'd tell you is that if you take out mobile, we grew in '26 based on the midpoint of the guidance for Q4, right? We grew 125%. I believe that's faster than the market. When you consider GaN and SiC addressing high-power and AI infrastructure, growth outpaces the broader market—so we have momentum to capture share, all right? And as I said, we're not here just to take part. We're here to capture share. So I believe we're positioned well, given our speed, our technology, and the focus we bring. We aren't a business that tackles EV on Monday, industrial on Tuesday, then AI data center on Wednesday and Thursday, correct? That is the sole thing we do. And that concentration lets us secure more share than rivals do.
Can you give us a little more detail on Claros? I think on the update call a week or two ago, you mentioned there's an anchor customer for the IVR product. First off, how big could that one customer potentially be? Second, what does the pipeline of other customers look like beyond that one? And how quickly might those start to matter in the '28 to '29 window?
Thanks, John, for circling back on that. That anchor customer is vertically integrated — they view things all the way from the processor up through the system level. Claros has been working with them since the beginning. As Kevin mentioned before, this has played a role in shaping the technology, since it's not just about the tech stack — it's about how you integrate that into a chip, into an array, into a system that delivers density, correct? So that customer, as we noted at the time of the acquisition, is going to ramp meaningfully in late '28 and '29. Prior to — and before the acquisition, Claros was basically focused on that customer to prove their technology and achieve scale, right? Since we merged the two companies, we've met pretty much everyone out there running an IVR program. And we're beginning to see good feedback, but also solid momentum in the technology that we had. And once more, the message we keep hearing is that density is what this race is about, all right? After that, the debate turns to Claros versus Ferric versus Empower versus Parlattice and others — over which has the strongest density and which approach is best, correct? I don't want to weary you with specifics, but density lacking caps outside differs from density having caps outside. Still, that isn't what customers report to me. What customers adore is Claros's modularity. You could say this was designed with the system in mind, not merely the XPU but the system in mind. And that is drawing considerable traction. With Navitas, I believe we can extend that to additional customers. It's a little soon to call it a win—there remains plenty of work ahead.
Thank you all for tuning in. I want to end with a point I've made before, but I believe it's really important for you to lock into your minds, alright? Exactly one year ago today, we committed to reshaping this business from one centered on consumers into a high-power company. We accomplished that with roughly the same OpEx. We picked up two points of gross margin, and we've managed to post consistent double-digit growth across the last few quarters even with the headwind created by deliberately stepping back from mobile. Now we're moving into this company's second phase. We've zeroed in on AI infrastructure. We hold the technology stack required, spanning ultra high-voltage SiC, high-voltage SiC, GaN and IVR. AI infrastructure — meaning data centers plus the grid, the power infrastructure — accounted for 40% of our sales as Q4 closed, and it will fuel future growth. And I believe we're only getting started. I also recognize there's plenty of work ahead. This market demands a great deal when it comes to reliability and speed, yet I believe we've got the right team and the right technology. And I think you should now view us as a company that grows at a consistent double-digit rate, capitalizing on the inflection points that stack on top of one another. That's my parting message; and I'd also like to use this moment to thank our employees — transforming the company takes a great deal of work, and they've shown real resilience and determination in getting it done. And now we can take Navitas to the next level. So my closing words go to them, to all they've delivered over the last 12 months and to what they'll accomplish going forward. Thank you.
That wraps up our session for today, everyone. We appreciate each of you being here. You're free to leave the line now.
Read the full transcript via the API
You're viewing the first half of this call. Get the complete Navitas Semiconductor Corporation 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 Navitas Semiconductor Corporation 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.