Thank you for being here, and welcome to the Navitas Semiconductor Q2 '25 Earnings Call. I would now like to hand it over to Lori Barker from Investor Relations. Lori, you're up.

Good afternoon, everyone. I'm Lori Barker, Investor Relations for Navitas. Thank you for joining Navitas Semiconductor's Second Quarter 2025 Results Conference Call. I'm joined today by Gene Sheridan, our President, CEO and Co-Founder; and Todd Glickman, CFO. A replay of this webcast will be available on our website approximately 1 hour following this conference call, and available for 30 days. Additional information related to our business is also posted in the Investor Relations section of our website. Our earnings release includes non-GAAP financial measures. Reconciliations of these non-GAAP financial measures with the most directly comparable GAAP measures are included in our second quarter earnings release, and also posted on our website in the Investor Relations section. Non-GAAP expenses and operating margin excludes stock-based compensation, amortization of intangible assets, and other non-recurring items. In this conference call, we will make forward-looking statements about future events or about the future financial performance of Navitas. You can identify these statements by words like 'we expect' or 'we believe' or similar terms. We wish to caution you that such forward-looking statements are subject to risks and uncertainties that could cause actual events or results to differ materially from expectations expressed in our forward-looking statements. Important factors that can affect Navitas' business, including factors that could cause actual results to differ from our forward-looking statements, are described in our earnings release. Please also refer to the risk factors sections in our most recent 10-K and 10-Q. Our estimates or other forward-looking statements may change, and Navitas assumes no obligation to update forward-looking statements to reflect actual results, changed assumptions or other events that may occur except as required by law. And now over to Gene Sheridan, CEO.

Thanks, Lori. And thanks to all of you for joining our Q2 2025 earnings call. I am pleased to announce Q2 revenues of $14.5 million which are in line with our guidance, despite a number of headwinds that continue to challenge near-term results. Our industry is in a classic semiconductor downturn with a slowdown in projections for solar, industrial, EV sectors, the continued impact of tariff conflicts, and now, the removal of tax credits for the solar and EV industry. Despite these short-term industry headwinds, Q2 was a very strategic quarter for our company. We have decided to more aggressively transition and invest in a leadership position for AI data centers. AI data centers represent an extraordinary opportunity and one that Navitas is uniquely positioned to capitalize on. In Q2, we announced that NVIDIA has selected Navitas to support their vision for next-generation 800-volt data centers. We also raised nearly $100 million of new capital in the quarter, providing additional cash flow to support our plans for growth, including investments needed to execute on development milestones over the coming quarters to support a significant ramp expected in late '26. Furthermore, in Q2, we announced our new GaN foundry partner, Powerchip, which is enabling our next-generation 8-inch, low-cost manufacturing platform to support our cost and capacity goals for this expanded AI data center opportunity. Our existing high-voltage GaN supplier is TSMC, which is utilizing 6-inch wafers. Over the next 2 and a half years, we expect our high-voltage customers to transition to Powerchip's 8-inch factory, which can produce nearly 80% more chips per wafer compared to 6-inch for little incremental cost. We expect this will yield more attractive price points for our customers and higher gross margins for Navitas. We are already planning to ramp our medium voltage 80- to 200-volt GaN from Powerchip's 8-inch line. Last, we made the decision to sharpen our focus on the high-end performance applications of the mobile, consumer and appliance sectors. Going forward, we will reduce our focus, investment and revenue expectations around the more mainstream, price-sensitive applications in these sectors. We expect this transition will allow Navitas to maintain its spending discipline, with operating expenses at or below current levels, while shifting significant customer acquisition and R&D investments to next-generation AI data centers and the related energy infrastructure markets. With that in mind, I'd like to share much more detail around the opportunities that we are targeting with this increased focus. AI is about to transform everything in our lives and on our planet. The impact will be felt in all sectors, but it starts with the cloud. The world's most powerful AI processors are rapidly being developed and deployed in the cloud, and with this massive processing capability, we are also seeing an explosion in power requirements that has never been seen before. While we now have AI processors that are rapidly approaching the intelligence of a human brain, we should also consider that the energy consumption of the human brain is believed to be less than 20 watts, while AI processors may require over 1 billion watts to generate a similar level of intelligence. So this presents a very large-scale power problem for our industry. In particular, total AI processor power consumption is projected to go from 7 gigawatts in 2023 to over 70 gigawatts by 2030. That 10x increase is creating a number of power challenges for our industry, but will also generate significant commercial opportunities for the suppliers that deliver leading solutions to these major power problems. In power electronics, when any system requires a significant increase in power capability, one of the most effective strategies is to increase the operating or system bus voltage. Since power distribution losses are inversely proportional to the square of the operating voltage, a 4x increase in voltage translates into a 16x reduction in power loss and a very dramatic improvement in energy efficiency. Before AI, traditional data centers operated with a 12 volt bus and achieved overall energy efficiencies in the 70% range. In the last 2 years, with the deployment of the first AI chips, the data center industry has quickly transitioned to a 48-volt bus architecture, reducing power distribution losses by up to 16x and targeting overall efficiencies in the 80% range. These 48-volt data centers are also supporting a major upgrade from 10 to 20 kilowatts in a traditional server rack, to a 100-200 kilowatt rack in a 48-volt data center. While such a change is dramatic, it is far from adequate to handle the exponential growth in AI power requirements. Thankfully, NVIDIA has announced their intentions to enable 800-volt data centers in the future. With a 15x increase in operating voltage, overall data center efficiencies should improve significantly while targeting rack power that can achieve 1 megawatt or more. This is an important step and vision for the data center industry, but also poses significant challenges and opportunities for the power electronics industry and semiconductor suppliers such as Navitas. Traditional 12 volt data centers had moderate power semiconductor content and little or no demand for the ultra-efficient gallium nitride or silicon carbide technology that Navitas is producing. The near-term move to 48-volts has increased that power semiconductor content to around $10 million to $20 million per gigawatt of power demand, which may represent over $1 billion per year power semiconductor market opportunity globally over the next 5 years, but GaN and silicon carbide will be required only selectively in these lower voltage and lower power systems. However, as we look at 800-volt systems, these super-charged data centers will require 3 different stages of power conversion, all of which, we believe, will need a combination of gallium nitride and silicon carbide technologies to meet their power, efficiency and density requirements. As a result, we expect power semiconductor content of AI data centers to expand to $30 million to $50 million per gigawatt of power delivered, with the majority demanding gallium nitride or silicon carbide. When combining this growing content with the 10x increase in AI power projected in the future, we believe this could translate into a $2.6 billion per year opportunity by 2030 for gallium nitride and silicon carbide. Let me break down this opportunity further and explain Navitas' capability and plans. There are 3 power conversion applications or stages for the 800-volt data center opportunity. The first stage starts with the power grid itself, which is not well prepared to handle this significant surge in energy demand. The existing grid was designed over 100 years ago and is antiquated, inefficient and lacks the stability needed to electrify our planet. The existing grid is built on low-frequency transformers or LFTs, which have no semiconductor content and don't offer good flexibility to deal with the growing instability of the grid itself and the fast charging power requirements of data centers and our planet. The answer is in solid-state transformers, or SSTs. SSTs enable up to 75% reduction in size and weight and offer an inherent robust ability to efficiently deliver more power under a wide range of operating conditions. SSTs require the most efficient power semiconductors, like silicon carbide, and require silicon carbide that withstands ultra-high voltages to interface directly to the grid which operates at 10,000 volts or higher. Our GeneSiC technology is a leader in ultra-high voltage or UHV silicon carbide technologies, with voltage ratings up to 6.5 kilowatt. This positions Navitas to support this mega-trend to upgrade the grid with robust, efficient solid-state transformers. In the process, we expect to participate in creating a brand-new $1 billion per year market opportunity for Navitas. We have completed initial sample evaluations with positive customer results. We are now developing engineering samples, initially at 2.3 and 3.3 kilowatt in our new SiCPAK package optimized for these ultra-high voltages and designed for very high reliability. We expect to deliver these final samples to customers in Q4 this year. Our target customers include major power system integrators such as Schneider, Eaton, Vertiv, among others. We expect first customers will create system designs during 2026, with many targeting 2027 for mass production ramp-up. We currently see limited competition in this exciting new market, and we are aggressively investing in expanding our technology lead and forging strategic relationships with the early leaders that are creating these SSTs. These SSTs convert the grid power from over 10,000 volts down to 800-volts needed for next-generation data centers. We estimate this stage requires about $8 million of power semis per gigawatt of power delivered, which translates to about $0.5 billion per year opportunity for our silicon carbide technology in the next 5 years. But this is only the estimate for grid-powered data centers. We see the same trends and opportunities to upgrade the grid to connect with large-scale renewable energy and storage systems and also to power the rest of our increasingly electrified planet. In total, we estimate SSTs and grid power reflects well over a $1 billion per year total market for Navitas silicon carbide technology by 2030. This takes us to the second stage, which converts that incoming 800-volts from the grid down to 48-volts. This second stage requires 800-volt to 48-volt DC-to-DC converters that are integrated directly onto the server motherboard, where real estate is at a premium and thermal management is critical. As a result, this second stage will require the highest frequency, efficiency and density technologies which GaN and silicon carbide can offer. While the 800-volt input of this stage can utilize high-voltage GaN or silicon carbide, the 48-volt output of this stage is a perfect fit for our new mid-voltage 80- to 200-volt GaN, which we expect to introduce beginning later this year. Our target customers for this stage include Flex, LiteON, and Delta, among others. Similar to the first stage, initial engineering samples have been evaluated by our early customers with positive results, and we are now developing final engineering samples for lead customers in Q4 and supporting them to finalize their system design and supplier selections in advance of early production ramps expected to start in late 2026. We are one of only two suppliers that can offer the full range of high-voltage GaN, high-voltage silicon carbide, and the mid-voltage GaN for this application and we believe we offer the best performance versions of each. We estimate this application represents a $1 billion per year TAM given $15 million of power semis per gigawatt delivered. And finally, we come to the third stage, which takes the 48-volts from the second stage and converts it down to 12 volts or less to eventually power the AI processor itself. This 48-volt DC-to-DC converter has the most critical demands for high efficiency, density, and reliability given it is the closest to the AI processor on the motherboard. For this stage, Navitas will use our new 80- to 200-volt GaN technology. This technology offers industry-leading frequency, efficiency and density, and does it in an innovative new package which efficiently removes the heat from the top and bottom of the package, all in a very tiny 5 by 6 millimeter footprint. Initial customer evaluations are complete with positive results. As with the other stages, we plan to deliver final engineering samples in Q4 and expect customers to finalize their system design and supplier selections in advance of early production ramps in late 2026. We estimate this stage represents the largest of the 3 stages with a $1.2 billion per year TAM, given $20 million of power semis per gigawatt of power delivered. In aggregate, we believe the AI data center and related energy infrastructure opportunities represent a very sizable $2.6 billion per year market opportunity for our industry and for our company. We are pleased that our years of collaboration with industry leaders like NVIDIA have led Navitas to be recognized among this sector's key ecosystem partners. We are excited by the positive initial customer evaluations and look forward to developing the final technology for each of the 3 stages I described. We believe we are at the right time, in the right place, with the right leading-edge technology to establish a leadership position in this fast-emerging AI data center and energy infrastructure market. While some of the market headwinds and our decision to sharpen our focus within the mobile and consumer segments will constrain our near-term financial performance, we believe we are making important decisions and investments to capitalize on this exciting opportunity. We recognize that a lot of important information and updates have been shared today. To assist investors to better understand Navitas plans and opportunities for AI data centers and related energy infrastructure, we are posting a short PowerPoint on our investor website today that further explains many of the opportunities I just covered. In addition, we will be holding a live presentation and Q&A event open to the public later in the week. Now, let me turn it over to Todd to expand further on the financial implications of the strategies and initiatives that I have described.

Thank you, Gene. In my comments today, I will take you through our second quarter 2025 financial results, and then I'll walk you through our outlook for the third quarter and explain further on the financial implications of some of the important Q2 achievements, market dynamics and announcements that Gene has described. Revenue in the second quarter of 2025 was at the midpoint of guidance at $14.5 million. As expected, revenue and the industry environment remained relatively static compared to the first quarter of 2025. The decline compared to a year-ago quarter was primarily the result of lower revenues in the China EV and industrial markets as semiconductor customers wait for improved economic indicators. Before addressing gross profit and expenses, I'd like to refer you to the GAAP to non-GAAP reconciliations in our press release. In the rest of my commentary, I will refer to non-GAAP measures. Gross margin in the second quarter was 38.5%, which was up sequentially compared to 38.1% in the first quarter, primarily due to a slight favorable change in product mix. In the second quarter, we executed on further synergies and operational efficiencies associated with prior acquisitions as we reduced operating expenses sequentially from $17.2 million to $16.1 million. Operating expenses were comprised of SG&A expenses of $6.9 million and R&D expenses of $9.2 million. Expenses were slightly higher than projected as we incurred additional second quarter only R&D expenses related to our high power GaN development. Consolidating certain support groups and sites, and further streamlining day-to-day functions has allowed us to significantly reduce SG&A by 17% or $1.4 million from the first quarter, while we continue investing in next-generation GaN and SiC technology platforms to serve the increasing power consumption across AI data center and energy infrastructure markets. Adding all this together, the second-quarter 2025 loss from operations improved sequentially to $10.6 million from $11.8 million in the first quarter of 2025 by leveraging SG&A cost reductions. Our weighted average share count for the second quarter was 199 million shares. Approximately 20 million shares were issued, yielding net cash proceeds of $97 million through our at-the-market offerings that we concluded during the quarter. Turning to the balance sheet. Accounts receivable was relatively flat since last quarter at $12.5 million. Inventory is $15.1 million down from $16.1 million in the first quarter. We took an approximately $3 million China SiC inventory reserve on U.S.-produced SiC products given the impact of the unstable tariff environment on our sales into China. Ultimately, we believe our U.S. manufacturing location for SiC wafers will provide Navitas with a significant strategic advantage with our U.S. customers for AI data center and energy infrastructure over time. Our balance sheet remains very strong as we exit Q2 2025 with high levels of liquidity and an improved working capital position. Cash and cash equivalents at quarter end were $161 million and we continue to carry no debt. Moving on to guidance for the third quarter, we currently expect revenue of $10 million plus or minus $500,000. This expected revenue reduction reflects both adverse impacts from China tariff risks for our silicon carbide business and our strategic decision to de-prioritize lower margin China mobile business while we accelerate our investment and leadership in AI data centers and associated new energy infrastructure. These applications have ever-increasing power demands ideally served with Navitas' differentiated GaN and SiC technologies that we expect to yield strong growth potential and higher margins over time. We believe this transition toward AI data center and energy infrastructure markets will take multiple quarters, and we have set up our balance sheet accordingly. In addition to the completion of our $97 million net capital raise, we have reduced operating expenses 25% from second quarter 2024. We believe we are well-positioned with the resources and runway to execute on opportunities for our next wave of growth driven by increased scale and profitability. Gross margin for the third quarter is expected to be flat compared to the second quarter, with our guidance at 38.5% plus or minus 50 basis points, as our expected mix remains consistent in the near term. Turning to operating expenses, we anticipate operating expenses of $15.5 million in the third quarter, down from $16.1 million in Q2 2025 as we continue to execute on our plan of focusing on fewer markets to drive capital efficiency across the business through the transition. For the third quarter of 2025, we expect our weighted average share count to be approximately 214 million shares. In closing, we are pleased with our Q2 results, particularly around the recognition of our technology attracting fresh investment, the announcement of strategic ecosystem partnerships and the acceleration of our focus and investment in data centers and energy infrastructure. Looking forward, while some industry headwinds continue and we will expect to further lessen our reliance on mobile revenues, we are confident these strategic moves position the company for its next wave of significant growth.

And your first question comes from the line of Ross Seymore with Deutsche Bank.

Gene, in this transition, I guess let's get the bad news out of the way first. How do you expect the revenues to behave between now and the time in, it sounds like, the second half of next year where the bigger ramp will occur in the data center side? So as we drop down first before we go back up on the other side, how should we think that shape looks?

Yes. Thanks, Ross. Good question, the obvious question, right? So as we make the transition, we're going to both reduce dependency on mobile, as Todd and I described, which does involve reducing some revenues being more selective in that market. At the same time, we're layering in all new design wins from other sectors and even the 48-volt data centers. So I think those net out to some softer quarters over the next 1 or 2 quarters, but sets us up well as we shift all of our investment or heavy investment into the AI data center, first 48-volt, but much bigger as we described with the 800-volt. So it will be a transitionary period, but some softer quarters in the near term, but setting us up well for big growth as we get into '26.

And then I guess that's the perfect transition to the good side of the equation. When you get into '26 and you say most of these ramps are late '26, are you talking second half or fourth quarter? And then looking even further beyond that, what's the sort of margin structure you think this business could offer? Especially on your gross margin line, it seems like it would have significantly less pressure from competitors perhaps, but inherently a more performance-centric market. So what sort of gross margin do you think that could deliver?

Yes. Let me, so covering the first part, recognize the 800-volt data center is heavily a 2027 play. That's a pretty extraordinary one with really high content, we believe, GaN and silicon carbide across those 3 stages we talked about. The 48-volts are still ramping, right? We previously announced 70 customer projects, over 40 wins. They're not as big. They're not as significant, but those will be contributing and layering into growth and help offset some of that reduced mobile dependency we talked about throughout '26. So I think that's the dynamic for '26 on the top line. I think on margin, we still have our same long-term margin model north of 50 points. As you point out, the AI data center and the energy infrastructure markets should be very high-value markets. They're really driven around performance, efficiency, density, Cost is secondary, even supply chain output is a higher issue than cost. And that's also, obviously, part of what we're doing with mobile and reducing that dependency being more selective. So I think all that sets us up well. I'd add in another factor, which is a big announcement in the quarter, Powerchip, low-cost 8-inch, that's also going to be layering in over the next 12 to 24 months. That's going to drive great cost reduction for us, better price points in the market, better upside on growth, but also add to the incremental gross margin as we go from high 30s, low 40s and head towards that longer-term model of 50% and beyond.

And your next question comes from the line of Blayne Curtis with Jefferies.

I would like to follow up on Ross's point. Can you clarify what portion of your mobile business you consider to be the lower gross margin segment that you are moving away from? You mentioned it as a continued headwind. I'm trying to understand if the absolute amount will decrease from here on out, or if you plan to maintain the remaining segments a bit longer.

Yes. Yes. Good questions, Blayne. And so we've often highlighted how the value of GaN in chargers and in other applications goes up as you go up in power. Ultrafast chargers we've talked about in the past are really north of 100 watts. We participated in things below that 45 watts, 65 watts. Those are popular power levels even with silicon chargers. There's a lot of volume there. A lot of that volume is in China, but we don't love the price points, we don't love the margin profile, and we don't love the price kind of profile going forward. So we're going to really refocus on those ultrafast chargers, 100-watt and up. Case in point, we just announced a 90-watt Xiaomi aftermarket charger. That's a great one in China that's with a great brand and a great partner and an incredible power density. It's the power that is on the upper end of most notebooks, but in the size of a typical 12-watt silicon charger. So that's the kind of thing we'll keep going with. We're going to like that margin profile better. But admittedly, it's going to be less revenue base. The mainstream where we're reducing is more China, more 65-watt and 45-watt, and we see that being far more than offset as we start to ramp AI data centers later in '26.

I was curious about the transition to Powerchip. Can you explain your ability to achieve any volume before Powerchip ramps up? Is there any effect from this transition on your revenue outlook?

Yes. And you can kind of break it into 2 pieces. We've actually got the, what we call, the mid-voltage GaN that's brand new, 80- to 200 volts. That's very important for 48-volt data centers today and 800-volt data centers tomorrow in this, what I call, the Stage 3. That's starting out straight away from Powerchip. We're already sampling that next quarter from Powerchip. That will start ramping production in early '26. The high-voltage GaN, we're already shipping today from TSMC, we'll start sampling that to customers, if not late this year, early next year. That will start ramping in late '26, and we expect a lot of our customers to migrate quite quickly from TSMC to Powerchip given the big advantages in technology cost and capacity.

In summary, there is no supply issue. The revenue decline is solely due to changes in pricing and margins in the mobile sector.

Yes. Yes, exactly. Yes. Thanks for clarifying. No supply at all. In fact, by bringing up 8-inch, that gets you 80% more die per wafer. We'll have a lot of capacity. No shortages or supply chain issues on our mobile decisions in the short term.

And your next question comes from the line of Kevin Cassidy with Rosenblatt Securities.

Since NVIDIA's announcement, have you noticed any increase in the adoption of gallium nitride in 48-volt data centers? They seemed somewhat hesitant before, but does this help overcome that hesitation?

Yes, that's a good point, Kevin. It may be a bit early to make a definitive call, but your question highlights an important aspect. We already see GaN designs in progress for the 48-volt system of an 800-volt data center. The 48-volt converter using GaN, which I refer to as Stage 3, can also be implemented in 48-volt data centers. We are optimistic that as we validate it for the 800-volt system, we might see some positive outcomes next year, even before the 800-volt rollout, with the use of it in 48-volt data centers. However, it's still too early to predict, and we will monitor how this develops and support it as needed.

Okay. I see. And then also on the transition from TSMC to Powerchip, should we expect to see inventory build? Are you going to get a safety supply of TSMC wafers before making the transition?

Yes. Yes. So TSMC is committed to at least a 2-year supply through mid of '27. That might get extended. But even if it doesn't, we can do additional last time buys as you're implying. So our message to our customers, if it takes them a little bit longer to transition to Powerchip, we can supply them through all of '27, probably even into '28 if it's needed. I don't think it will take anybody that close. And there are so many big advantages on cost capacity and tech to make that move, but that gives us a nice cushion and a high confidence on supply chain.

And your next question comes from the line of Jack Egan with Charter Equity Research.

So just on the near term, I was hoping you could kind of go through the drivers for the big sequential decline in September. So I mean, you have a weaker demand environment, some tariff headwinds and then the narrowing of your products portfolio. So how much did each of those kind of contribute to the guidance for September?

Yes, you summarized it really well, Jack. They carry almost equal importance. We anticipated that silicon carbide would pose a risk for us in China due to tariffs. We are among the few companies engaged in U.S. manufacturing, and that risk is becoming a reality. However, this same risk associated with U.S. manufacturing is evolving into a significant advantage as we consider AI data centers and energy infrastructure in the long run, since our predominantly U.S.-based customers appreciate having a U.S. supply chain. In the short term, this has affected us, primarily impacting Q3, with possibly some effects in Q4. Additionally, we are aiming to be more selective in mobile to lessen our dependency, which will require several quarters to achieve, as Todd mentioned. We haven't observed the expected ramp-ups from new design wins due to ongoing industry slowdowns that may persist for a few more quarters, and it’s challenging to forecast even for our larger competitors. Thus, I would say both factors are equally responsible for the sequential decline we saw in Q3.

Okay. That's helpful. And then on the data center side, has the 800-volt announcements with NVIDIA led to more engagement with other data center customers? And I mean, would Navitas have the design and support resources to handle those additional products? Or I mean, would NVIDIA really just take the bulk of your focus for the foreseeable future?

Yes, we see many new opportunities ahead. We believe we are already well positioned in the market. We are familiar with many of these customers globally, particularly in various power supply applications such as notebooks, desktops, and some early server projects. Our knowledge of these customers extends to the newest technology, solid-state transformers, where we have received significant interest. This has led to numerous new prospects, which is very encouraging. Regarding the topic of new opportunities, NVIDIA plays a major role, and we are very enthusiastic about their vision. However, they are not the only company aiming to advance the shift towards higher voltage data centers. There are other emerging opportunities, although we don’t have any specific announcements to make at this time. We anticipate that more players will also contribute to market growth.

And your next question comes from the line of Joe Moore with Morgan Stanley.

Could you discuss the competition for these NVIDIA products? You've had ten different companies announce their participation in this partnership. Clearly, you have a broader range of wide band-gap products to address this. Can you explain your position within that serviceable available market?

Yes, definitely, Joe. And you touched on the first point, which is, this is a pretty extraordinary challenge from grid power at tens of thousands of volts to step it all the way down to GPU power at sub 1 volt. And each of these 3 stages has a big demand on high efficiency and high density. So you're going to need the best high-performance, high reliability and high-efficiency technology in each stage. We feel like we're starting from a great place, having it, not just having the range, but on the silicon carbide, what we call ultra-high voltage, 2.3 kilovolt, all the way up to 6.5 kilovolt, very few suppliers in that space. We have the best performance, the best reliability, in our opinion, gives us a great starting point. And we're investing aggressively to expand that lead generationally and in packages and portfolio. You go to that second stage of high-voltage GaN, which companies have high-voltage GaN, high-voltage silicon carbide, and it may take a combination of those 2 in the second stage and the output of that, the mid-voltage GaN, 80- to 200-volt. Very, very few, even of the named suppliers that are participating have that combination. And then that third stage is all mid-voltage GaN, 80- to 200-volt GaN with a 40-volt converter, demanding the highest efficiency, highest density. And part of our strategic decision to defocus and reduce some of our dependency on mobile is shifting aggressively to accelerate that R&D, push that technology advantage, push the focus, and increase the customer intimacy. So I think those are all positives. Another big positive for us is our size. While we're up against some big guys, I think our small size is our advantage, speed, flexibility, innovation, risk-taking and focus. It all comes back to focus. And so a lot of what you heard today is us really doubling down and increasing the focus in this critical fast-moving market.

And your next question comes from the line of John Tanwanteng with CJS Securities.

I was wondering, since there's obviously no design wins as part of the announcement, there's a long time until you get there. You're announcing this transition away from your core bread and butter markets. Are you getting any design and engineering revenue along the way? Or is it really just risking it all with the knowledge that you do think you have a performing product before we get there? And how should we think about the cash flows along the way as well? Obviously, nice to see that you do raise the capital, but what should we expect before the cash flows start ramping again?

Yes, it is an important point. While the 800-volt technology is exciting and we are on the right track, primarily it will impact 2027. In 2026, we are still focusing on ramping up 48-volt data centers. We have announced 40 design wins, totaling over 70 customer projects that will be scaling up. Although these gains are somewhat countered by reduced dependency on mobile, we anticipate seeing significant growth in 2026. We expect to share impressive announcements and design wins during this period. We didn't emphasize our pipeline this quarter as we were primarily explaining the AI data center opportunity, but you will see those design wins and growth indicators throughout 2026. Todd, I believe there was another question for you.

Yes, John, I think you mentioned something about cash flow. Yes, this last quarter, our operating cash flow was around $11 million. With the $106 million on our balance sheet today, we expect to maintain cash flow usage of around $10 million to $11 million going forward. So that's our profile going forward because we expect to also keep operating expenses pretty flat.

Okay. Great. And just to clarify, Gene, when you say things improving through '26, are you talking about just data center? Or are you expecting other markets to recover as well?

Yes. I think that depends somewhat on the markets. We did enter this year with a strong pipeline, strong design wins. Some of those forecasts have come down or delayed a bit with the continued softness in the market. But I would certainly expect to see recovery, as I think most do going into '26, and that's going to add to some tailwinds for us.

And your next question comes from the line of Richard Shannon with Craig-Hallum Capital Group.

A couple of questions here. Gene, curious to understand kind of the change in focus here and what's driving this. Have we had a material change in pricing trends in certain markets here? I know you've got fairly aggressive competitors, specifically in China with mobile, but wondering if there's been any change here in the last quarter or 2 that's dictated this? Or is it just a continuing trend that's been made it harder?

Yes. No, I think you nailed it on the head, Richard. No big announcements here. We've seen this trend for a while. We've talked about the importance of ultrafast charging and where we bring the real value. We participated in some of the more mainstream price sensitive, as I said, 45, 65-watt, decent volume there in China. But those trends continue. And when the market is soft, people kind of get more aggressive on price, it's sort of normal. But we don't like that pricing trend. We don't like the margin profile. And frankly, we're getting ready for a much more attractive pricing margin profile and where we really want to put our investment and focus as you're hearing throughout the call.

Okay. Fair enough. And if I missed this, I apologize. But I just want to get a sense of how you're thinking of the trajectory in gross margins over the near term. I think it was an earlier question about what to expect from a revenue perspective, which seems kind of maybe bumping along the bottom or whatever phrase you'd like to use. But how do we think about gross margins here, particularly as we deemphasize some of these lower-margin markets? Will we get up to bump above 40% here fairly soon? Or does that really take the revenue inflection to make that happen?

Yes, it's actually going to take the revenue inflection to make that happen. To your point, we've delivered 38.5%. We're guiding to 38.5% in Q3, and we expect it to remain that level as some of the other businesses are experiencing some tariff pressure on their margins, mainly in our silicon carbide business. So you're not going to see that gross margin profile increase until the other sectors kick in.

There's no further questions at this time. That concludes today's call. Thank you all for joining. You may now disconnect.